Why did SVB fail so quickly in March 2023?

The short answer

Between 2020 and 2022, SVB’s deposit base nearly tripled from approximately $62bn to almost $200bn as the venture-backed tech sector enjoyed pandemic-era cash infusions. The bank invested most of these deposits in long-duration Treasury and agency securities at very low yields, classifying the bulk as held-to-maturity to avoid mark-to-market volatility on regulatory capital.

When the Fed raised rates 525 basis points between March 2022 and July 2023, the market value of those securities fell sharply. Simultaneously, venture funding tightened and tech firms began drawing down their SVB deposits to fund operations. By March 8, 2023, SVB announced a $1.8bn after-tax loss from selling its available-for-sale portfolio and a $2.25bn capital raise.

The announcement triggered VC-coordinated withdrawals: $42bn left the bank on March 9, with another $100bn queued for March 10. The California Department of Financial Protection and Innovation took possession on March 10 and appointed the FDIC as receiver.

→ New to bank failure mechanics? Systemic fragilities pillar

What the data shows

The SVB collapse compressed in numbers (Federal Reserve, FDIC, SVB regulatory filings 2022-2023):

• Total assets at year-end 2022: $209bn (16th largest US bank). Total deposits: $175.5bn, of which 86.4% (~$151.6bn) were estimated uninsured
• HTM securities portfolio: $91.3bn at year-end 2022, with only $563m in interest-rate swaps protecting it
• Available-for-sale portfolio sold March 8, 2023 for an after-tax loss of $1.8bn — the trigger announcement
• March 9, 2023: $42bn in deposits withdrawn in a single day. SVB had a negative cash balance of approximately $958m by close of business
• March 10, 2023: $100bn in withdrawals queued, representing 81% of total deposits when combined with March 9 outflows
• March 12, 2023: Systemic Risk Exception invoked by Treasury, Fed, FDIC. Bank Term Funding Program (BTFP) launched simultaneously

The exception that nuances the timeline: the Federal Reserve’s Vice Chair for Supervision Michael Barr testified that supervisors had warned SVB management about interest-rate risk to its balance sheet starting in November 2021 — over 16 months before the failure. The fragilities were known; the speed of collapse was not anticipated.

→ Dataset: US bank reserves

Why it happens — the macro mechanism

SVB’s collapse was the convergence of five reinforcing fragilities, each manageable in isolation but catastrophic in combination.

The duration channel. The bank held $91.3bn in HTM long-duration securities purchased at low yields. The 525 bp rate-hike cycle drove unrealized losses that, if marked to market, would have eroded most of the bank’s equity capital. 2023 vs 2008 crisis details how this differed from a credit-cycle crisis.

The sectoral concentration channel. SVB served roughly half of all US tech startups and venture capital firms. When venture funding slowed in 2022-2023, tech firms drew on their SVB deposits to fund payroll and operations. The same shock that pressured asset values also pressured the funding side. Contrary to the assumption that diverse client bases protect against deposit outflows, SVB’s homogeneous tech clientele meant that all depositors faced the same liquidity pressure simultaneously.

Bridging point: the third channel concerns regulatory perimeter and supervisory intensity.

The supervisory channel. The 2018 Economic Growth, Regulatory Relief, and Consumer Protection Act (EGRRCPA) raised the threshold for enhanced prudential standards from $50bn to $250bn in assets. SVB sat just below this threshold and faced lighter stress-testing requirements. Bank stress tests details how this gap left SVB-class banks under-monitored. This is the angle distinctive — five fragilities together, not one in isolation, drove the failure.

The deposit-coordination channel. 86.4% of SVB’s deposits were uninsured, concentrated among VC-connected tech firms. When the March 8 capital-raise announcement triggered concern, VC firms used group communications — Signal, Slack, Twitter — to coordinate immediate withdrawals across portfolio companies.

Synthesis by regime: in the pre-2022 low-rate regime, SVB’s strategy worked — long-duration securities yielded enough above zero-rate deposits to generate net interest income, and the EGRRCPA-relaxed supervision was tolerable given low rate volatility. In the post-March 2022 regime, the same strategy became untenable: rising rates crystallized duration losses, the supervisory framework had not adapted, and the depositor base behaved as a single unit under stress. SVB’s failure marks the regime transition itself, not a breakdown within either regime.

SVB failed not from one fragility but from five — duration, sector, uninsured concentration, lighter supervision, and coordinated VC withdrawals. Each was individually manageable; the combination was lethal.

→ Framework: Liquidity transmission

What it means for different economic actors

Bank shareholders. SVB Financial Group equity went to zero — shareholders received nothing in the resolution. The lesson: capital ratios that look healthy under HTM accounting can be illusory when forced sales materialize, even at fundamentally creditworthy banks.

Bank creditors and uninsured depositors. The Systemic Risk Exception protected uninsured depositors at SVB and Signature, but this was a discretionary decision rather than a structural guarantee. AT1 holders at Credit Suisse the same week were wiped out entirely, illustrating that subordinated debt remains structurally subject to bail-in.

Macro analysts. The episode validated specific signals — HTM unrealized losses relative to tier-1 capital, uninsured deposit concentration, sectoral homogeneity — as bank-stress indicators that traditional metrics like CET1 ratio missed. These signals are now standard in regional bank screening.

A common error is to treat SVB as a one-off event attributable to mismanagement. The point is set out at length in the most common misreadings of bank runs and banking crises. The Federal Reserve’s post-mortem identified systemic regulatory gaps — lighter supervision under EGRRCPA, slow supervisory response to known risks, and stress-test scenarios that did not capture rapid digital-deposit outflows. The fragilities were structural.

Practical observation

Go deeper

Related questions

• How did the 2023 regional bank crisis differ from 2008?
• What is deposit concentration risk for banks?
• How does social media accelerate bank runs?
• How do stress tests assess bank resilience?

Frequently asked questions

Could SVB have been saved?

Once the run started on March 9, no. The $42bn outflow on a single day exhausted available cash and asset-sale capacity, and the queued $100bn for March 10 made it operationally impossible to meet redemption requests. Earlier intervention — supervisory action in 2022 demanding hedging of duration risk, or a supervised capital raise before March 8 — could plausibly have prevented the trigger event. The Fed’s post-mortem identifies multiple intervention points that were missed.

What is the angle distinctive about the five-fragility convergence?

Most analysis of SVB focuses on a single cause — duration risk, or uninsured deposits, or social media coordination. Each is partially right but individually insufficient to explain the speed and totality of collapse. The structural insight is that any one of these fragilities was manageable on its own; the combination was uniquely lethal. Banks elsewhere had similar duration risk (Bank of America held large HTM losses) without failing. Banks elsewhere had high uninsured ratios (custody banks routinely do) without running. The intersection is what made SVB unique.

What lessons did regulators draw post-SVB?

The Federal Reserve’s framework review accelerated several reforms: enhanced supervision for banks above $100bn assets (effectively rolling back parts of EGRRCPA), updated stress-test scenarios incorporating rapid deposit outflow assumptions, and proposed changes to capital treatment of HTM unrealized losses for larger regional banks. Most are still in implementation as of 2025. The structural questions about deposit insurance design and TBTF perimeter remain unresolved.

Last updated — 14 June 2026