May 28, 2026Updated June 7, 20265 min readterminal
Table of Contents ▾
  1. The Unrealized Loss Structure That Made Deposit Flight Rational
  2. How Deposit Outflows Compressed Capital Adequacy in Real Time
  3. Why February Call Reports Showed the Risk But Markets Repriced in March
  4. What the Terminal Stress Metrics Would Have Signaled

SVB 2023: What the Balance Sheet Showed

Silicon Valley Bank's collapse on March 10, 2023 was not a surprise to anyone reading the quarterly call reports. The bank held $212 billion in assets, $91 billion locked in long-duration mortgage-backed securities and Treasuries, and carried $15.1 billion in unrealized losses on its held-to-maturity (HTM) portfolio by Q4 2022. The balance sheet was transparent. The math was visible. What happened between January and March was not a discovery—it was a repricing of information that had been there for months.

SVB's HTM portfolio carried $15.1 billion in unrealized mark-to-market losses as of December 31, 2022. This was not a minor accounting artifact. The bank held $91.3 billion in securities overall, meaning 16.5% of the portfolio was underwater on day one of 2023.

The composition matters. Of the $91.3 billion in securities:

  • $79.5 billion in mortgage-backed securities (87% of the total securities portfolio)
  • $15.3 billion in U.S. Treasury securities
  • $1.0 billion in other debt

MBS duration risk was extreme. These securities, issued when 30-year mortgage rates were 2.5% to 3.5%, were pricing for a world that no longer existed. By December 2022, the 10-year Treasury yielded 3.88%. The Fed had raised rates 425 basis points in nine months. Every basis point higher in the yield curve meant thousands in mark-to-market losses on a $79.5 billion MBS book with average duration of 5.5+ years.

What made the situation unstoppable was the deposit composition. SVB reported that 93% of deposits were uninsured—above the $250,000 FDIC limit. The largest 20 depositors held roughly 47% of total deposits. When venture capital and tech sector clients began moving capital in February 2023, the bank faced a depositor base with every rational incentive to leave first. There was no sticky core. There was no retail base. There was only business clients with direct knowledge of SVB's interest rate exposure and access to real-time market pricing.

Try it yourself

The only calculator that shows CPI plus the USD reserve premium — side by side.

Open the calculator →

SVB's Tier 1 capital ratio stood at 11.8% in Q4 2022, above the 10.5% regulatory threshold. Capital adequacy looked compliant on paper. This ignored a critical dynamic: if the bank had to sell its MBS and Treasury portfolio to meet deposit withdrawals, the $15.1 billion in unrealized losses would immediately convert to realized losses, eroding tangible capital.

The bank experienced a $42 billion deposit outflow on March 9, 2023 alone—a single day. This forced immediate asset sales. When SVB liquidated portions of its securities portfolio to cover the run, the mark-to-market losses became real. The bank reported that in the process of managing the outflow, it incurred a $1.8 billion loss on the sale of available-for-sale securities. This single transaction burned 1.8% of the bank's total tangible capital in hours.

By Friday March 10, 2023, when regulators seized the bank, the uninsured deposit base was in full flight. The bank could not raise funds at any rate. It could not sell assets without triggering cascading losses. The capital ratio had collapsed below minimum thresholds.

The Q4 2022 10-Q filing was public on February 2, 2023. Every metric that led to failure was visible:

  • $15.1 billion unrealized losses on HTM securities
  • $91.3 billion concentration in long-duration fixed income
  • 93% uninsured deposit ratio
  • Net interest margin compression from 1.96% to 1.60% year-over-year
  • Deposit growth stalling despite massive inflows to the bank between 2020-2021

The balance sheet was screaming. Yet SVB stock traded at $106 on February 7. On March 9, it closed at $28. The market did not reprice the risk until depositors actually began moving money. This was not a failure of disclosure. It was a failure of real-time capital allocation. Professional investors and financial analysts had weeks to read the filing. The pricing signal came only when client phone calls started happening in venture capital offices.

This timing gap—between data availability and market repricing—is the core failure in real-time banking stress monitoring. Call reports are quarterly. Deposit flows are daily. By the time a 10-Q file is public, the regime may have already shifted.

A banking stress monitoring system focused on forward-looking metrics would have flagged SVB in early February 2022—not March 2023. The signals:

  • Deposit volatility index: SVB's deposit base was concentrated in a single sector (venture capital and technology). Sector-specific capital flows are measurable. When VC funding velocity shifted negative in late 2022, SVB's deposit retention rate should have triggered a yellow flag.
  • HTM loss acceleration: As the yield curve steepened beyond 4.0%, the theoretical mark-to-market loss on a $79.5 billion MBS portfolio with 5.5-year duration should have been modeled forward. At 4.5% yields, the loss would exceed $18 billion. This was calculable.
  • Uninsured deposit ratio trend: 93% uninsured is a structural liquidity risk, not a capital risk. It signals that any loss of depositor confidence converts to immediate funding pressure. This metric rarely improves organically.
  • Asset-liability duration mismatch: SVB's liabilities (deposits) had an effective duration near zero. Its assets (MBS and Treasuries) had duration of 5.5+. A parallel yield curve shift of 425 basis points over nine months guaranteed losses. This is mechanical, not probabilistic.

The balance sheet showed all of this. The failure was in real-time capital flow monitoring and sector-specific deposit concentration analysis. SVB died not because its finances were hidden. It died because the repricing of known risks happened faster than the funding mechanisms could support.

The Reserve Global Terminal monitors banking stress across 25 countries in real time. $49/month.

Frequently Asked Questions

Why did SVB's capital ratio look adequate in Q4 2022 if the bank collapsed three months later?

SVB's Tier 1 capital ratio of 11.8% was above regulatory minimums, but this ratio ignores the $15.1 billion in unrealized losses on the HTM portfolio. Once deposits fled and the bank was forced to sell securities, those unrealized losses became real, erasing capital immediately. The ratio was compliant but fragile.

What made the 93% uninsured deposit ratio so critical to the collapse?

Uninsured deposits sit above the $250,000 FDIC guarantee. When depositors lose confidence, every dollar is at risk. With 93% of deposits uninsured and concentrated in venture capital clients, the entire depositor base had rational incentive to withdraw simultaneously. SVB experienced a $42 billion outflow in a single day because there was no sticky core.

Could regulators have seen this coming from public filings?

Yes. The February 2, 2023 10-Q filing showed all the critical metrics: $15.1 billion unrealized losses, $91.3 billion in long-duration securities, and 93% uninsured deposits. The balance sheet was transparent. What markets failed to price until March was the speed at which depositor confidence could shift in a concentrated client base.

See the real numbers for your currency

The only calculator that shows CPI plus the USD reserve currency premium — side by side.

Open the calculator →

Related Articles

2008 GFC: The Banking Data Nobody Read
May 14, 2026
Argentina 2018: The IMF Call Nobody Expected
June 25, 2026
Egypt 2023: Three Devaluations and a Lesson
July 2, 2026