Have you ever watched a warning light blink for months and still told yourself the engine was fine? That is the uneasy feeling hanging over the latest review of Silicon Valley Bank. An outside look at how supervisors handled the 2023 collapse now says staff at the central bank knew, or should have known, that the lender was already brittle long before depositors stampeded for the exits.
I sat with that line for a while. Knew, or should have known is not a throwaway phrase. It is the kind of wording that turns a messy episode into a live argument about accountability, timing, and whether the people paid to notice risk actually noticed it. In my experience, banking stories rarely fail because one number was missing. They fail because the numbers were sitting there, and the system treated them as background noise.
What The New Review Actually Changes
The collapse itself is familiar by now. In March 2023 the bank disclosed that it had sold securities at a steep loss and needed fresh capital. Depositors did not wait around for a tidy recapitalization plan. They ran. Fast. The institution that had become a default home for venture-backed tech cash suddenly looked like a one-way door.
What is new is the tone of the latest assessment. An earlier internal look after the failure leaned toward a story of caution that came too slowly. This newer review, presented publicly by the current vice chair for supervision, goes further. It argues that supervisors had enough in front of them to treat the bank as vulnerable well before the public drama.
Staff knew, or should have known, that the bank was vulnerable prior to the crisis.
That sentence is doing a lot of work. It does not claim a secret memo predicted the exact hour of the run. It claims the risk profile was visible. Concentration. Duration. Uninsured money that could leave in an afternoon. Bond holdings that looked safe on paper until rates moved the other way.
A Bank Built For One Kind Of Weather
Silicon Valley Bank was not a mysterious black box. Its client base was famous. Startups. Funds. Founders parking large operating balances. That model can feel brilliant in a cheap-money boom. It can feel terrifying when funding tightens and every treasurer in the same industry reads the same headline at the same time.
The review highlighted a deposit base that was heavily uninsured and tightly clustered around venture-backed technology firms. Roughly nineteen out of twenty dollars were outside ordinary insurance limits, according to the summary presented this week. You do not need a doctorate in banking to see the problem. Insurance exists for a reason. When almost nobody is covered, confidence becomes the only backstop.
Then came the rate cycle. The same Treasuries that had been treated as the conservative choice lost market value as policy rates climbed. Hold-to-maturity accounting can hide pain until it cannot. The moment management sold paper and booked a large loss, the story flipped from “we are liquid” to “we need capital.” Markets do not always wait for the second paragraph of a press release.
- Deposits concentrated in one ecosystem rather than spread across households and industries
- A very high share of balances above insurance coverage
- A securities book sensitive to rising rates
- A capital raise announced after a realized loss, not before it
None of those points is exotic. That is what makes the “should have known” language sting. If the ingredients were ordinary, why did the supervisory response feel extraordinary only after the run had started?
The First Review And The Second One Are Not The Same Story
Shortly after the failure, the then top supervision official commissioned a look-back. That earlier account described staff as too cautious in pressing the bank. Delayed findings. Soft language. A culture that preferred process over speed. Fair enough. Plenty of people inside large institutions recognize that pattern.
The newer review does not simply repeat the same sermon. It shifts the center of gravity. Instead of “they were slow once trouble was obvious,” it says trouble should have been obvious earlier. That is a different charge. Slowness after a flare-up is a management problem. Blindness before the flare-up is a detection problem.
I have found that institutions hate this distinction because it is harder to hide behind. You can always argue that hindsight is cruel. You cannot as easily argue that duration risk plus uninsured concentration plus a single-industry deposit franchise was some kind of riddle.
Why The Political Weather Matters Here
Personnel changes make this report land differently than it would have in 2023. The official who led supervision during the crisis later left the role so a new administration could choose its own top regulator. The current vice chair for supervision unveiled the consulting firm’s findings in a London speech. She did not need to name her predecessor for listeners to hear the implied question.
Will this become a push to reopen old fights about who stays on the Board? Maybe. Markets love a personnel drama. I am less interested in the soap opera than in the supervisory lesson. If every failure becomes a score-settling document, the public still does not get a cleaner early-warning system. If the document forces a harder look at how staff escalate rate risk and deposit flight risk, then the argument is worth having.
Perhaps the most interesting aspect is how quickly a technical review becomes a proxy war about the central bank’s broader posture. One camp hears “the staff missed it.” Another hears “the last team was too timid with a favored industry.” A third hears “this is about control of the institution, not about duration gaps.” All three can be true at once. That does not make them equally useful.
Uninsured Money Moves Like A Crowd, Not Like A Queue
Retail depositors with modest balances often stay put out of habit. Corporate treasurers do not. They have dashboards, group chats, and a duty to protect payroll. When a bank serving one industry announces a loss and a capital need, those treasurers do not form a polite line. They compare notes. Then they wire.
That is why the 94 percent uninsured figure is not a trivia item. It is the match next to the dry grass. Insurance is imperfect. It also buys time. Without it, the only thing standing between a rumor and a run is the belief that someone else will stay. Belief is a thin asset.
A concentrated, uninsured deposit base can turn a solvency story into a liquidity event before the solvency story is even finished.
I keep coming back to that. People still talk about Silicon Valley Bank as if the main sin was holding bonds. Bonds were the amplifier. The client mix was the fuel. Rising rates were the weather. Supervision was supposed to notice the combination, not each piece in isolation.
Rate Risk Was Never A Hidden Science
For years, cheap policy rates made long-duration government paper look like a sleepy place to park liquidity. Then the hiking cycle arrived. Mark-to-market losses piled up on securities that were still “safe” in credit terms. Credit risk and rate risk are not the same animal. A lot of comfort language in banking still pretends they are cousins.
When the bank sold holdings and booked about $1.8 billion in losses, it was not revealing a brand-new fact of nature. It was converting paper pain into recognized pain. Investors and depositors can live with unrecognized pain for a surprisingly long time. Recognized pain demands a plan. The plan, in this case, included raising capital after the market had already seen the bruise.
| Risk | What It Looked Like | Why It Mattered |
| Interest rate risk | Longer-duration Treasuries losing market value | Capital and confidence both take a hit when sales crystallize losses |
| Deposit concentration | Tech and venture cash in the same network | One shock travels through the whole client graph |
| Insurance gap | Vast majority of balances uninsured | Little reason for large clients to wait and see |
| Communication timing | Loss announcement paired with a capital need | The message can read as distress, not housekeeping |
Look at that table for ten seconds and ask yourself what a supervisor is for. If the job is only to confirm that loans are not exploding, this bank would have looked dull. If the job is to ask what happens when rates jump and the biggest clients all share the same group chat, the bank looks anything but dull.
Supervision Is A Craft, Not A Checklist Theater
There is a temptation after every failure to invent one more form. Another memo. Another traffic-light dashboard. I am skeptical. Paperwork did not fail Silicon Valley Bank by itself. Judgment did. Or at least the willingness to turn judgment into action while the bank was still standing.
Good supervisors sound a bit rude in private. They ask why so much of the balance sheet is one trade on rates. They ask why the deposit franchise has no ballast. They ask what happens on a Tuesday afternoon if a funding round freeze hits twenty clients at once. Those questions are uncomfortable in a boom. They are obvious in a bust. The whole point of the job is to ask them in the boom.
- Map who actually holds the deposits, not just the average balance.
- Stress the securities book against a real hiking path, not a polite one.
- Ask how fast uninsured money can leave, not whether it “should” leave.
- Force a capital conversation before a sale makes the loss public.
- Write findings in language that requires a deadline, not a workshop.
None of that is glamorous. It is also not mysterious. If the latest review is taken seriously, those steps stop being optional color commentary.
What Depositors And Founders Should Take From This
If you run a company, the lesson is blunt. A bank that “understands your industry” can also be a bank that shares your industry’s worst week. Diversifying cash across institutions is boring. So is monitoring uninsured exposure. Boring is underrated.
If you are an ordinary saver, the lesson is slightly different. Systemic backstops arrived after the run, and that fact still shapes how people think about safety. Guarantees after the fact can stop contagion. They also teach the next cycle that someone will clean up. That is a dangerous education if it becomes the only education.
I do not say that as a sermon against rescue. I say it because risk management dies when everyone assumes the rescue is pre-purchased. The 2023 weekend actions protected a lot of payrolls. They also left a political and regulatory hangover that is still being argued in 2026.
The Governance Question Nobody Wants To Make Small
Was this a staff failure, a leadership failure, a cultural failure, or all three? Reviews like this usually try to sound clinical. Readers hear personnel. That is inevitable when the person who now presents the findings occupies the chair once held by the official associated with the first review.
Keep the personalities in view, but do not let them eat the file. The file says a large, well-known regional specialist bank carried a deposit structure that could not survive a confidence nick, and a securities strategy that turned a rate cycle into a capital event. If that profile did not trigger urgency, the next profile that looks similar will not either, unless something in the watch-tower actually changes.
Some analysts already treat the document as kindling for a broader fight over remaining Board seats. That fight may happen with or without this text. The text still has to stand on its own: did supervisors have enough information to act earlier? The review’s answer is yes.
Lessons That Travel Beyond One Lender
Every cycle finds a new way to hide old risks. Last time it was housing credit dressed up as math. This time it was rate exposure dressed up as prudence. Next time it may be liquidity that looks plentiful until it is all the same kind of liquidity.
Simple watch list after SVB: 1. Who can leave in 24 hours? 2. What asset sale would announce a loss? 3. How many clients share one rumor mill? 4. Does the capital plan exist before the sale, or after?
Those four questions would have sounded almost too basic in 2021. They do not sound basic now. That is the quiet tragedy of late-cycle supervision. The basics only look basic after they have already been expensive.
I keep a private rule when I read these reports. If the findings could have been written with a yellow highlighter on the public financials, the failure was not a lack of data. It was a lack of nerve, priority, or both. This case looks a lot like that.
Where The Argument Goes From Here
Expect more hearings language, even if the formal venue changes. Expect supporters of the earlier team to say the new review is hindsight with a political accent. Expect critics to say the first review sanded down the edges. Both claims can be partly right. The public still needs a practical standard: when a bank’s deposits are almost entirely uninsured and one sector deep, supervisors do not get to call the situation “watchful waiting” forever.
Will rules get rewritten again? Probably at the margin. Will culture get rewritten? That is the harder part. Culture is who gets promoted for raising an awkward flag in year two of a boom. If that person still looks like a nuisance, the next speech will sound a lot like this one.
The expensive part of bank supervision is not the exam. It is the moment someone has to say the popular client franchise is also the risk.
That moment arrived too late in 2023. The new review is essentially an argument that the moment was available earlier. Agree or disagree with the politics around the messenger, the balance-sheet arithmetic has not become more complicated with time.
A Cleaner Way To Talk About Blame
Blame is a sloppy word. Management chose the business model. Depositors chose convenience and yield and industry fluency. Policymakers chose a rate path for inflation reasons that had little to do with one California-centered franchise. Supervisors chose how loudly to talk about the mix. You can hold more than one of those facts without turning the story into a cartoon.
Still, the unique job of supervision is to be the adult in the room when the business model is winning. Winning models attract fans. Fans do not write CAMELS letters. If staff truly had the ingredients in view, then “should have known” is not a smear. It is a job description that was not met at the speed the risk required.
I will say this as plainly as I can. A bank can be solvent on a hold-to-maturity fantasy and insolvent in the court of depositors the same afternoon. Any review that does not put that sentence near the top is doing public education a disservice.
Practical Takeaways If You Follow Markets For A Living
Watch regional specialists with fashionable client lists more closely than generic “community bank” labels suggest. Watch securities losses that management insists will never be sold. Watch insurance ratios, not just loan quality. Watch how quickly a capital raise is framed as optional strategy rather than emergency plumbing.
- Do not treat “government bonds” as a synonym for “no market risk.”
- Do not treat “sticky deposits” as a fact when the clients are institutions.
- Do not wait for a run to decide whether concentration was a feature or a flaw.
Those habits would have helped in March 2023. They will help the next time a beloved niche lender discovers that niche is another word for correlated.
The Human Tempo Of A Modern Run
Older bank-run pictures involve lines on sidewalks. This one involved group messages and same-day wires. Speed changes the moral of the story. Supervisors who still think in quarterly exam cycles are bringing a calendar to a stopwatch fight.
That does not mean every bank needs a daily public stress theater. It means internal escalation has to match the speed of the clients. Tech treasurers did not need a 200-page report to decide. They needed a headline and a loss number. The supervisory apparatus cannot be the slowest reader in that room.
Is that fair to examiners who were already writing memos? Fair is the wrong test. Effective is the test. If memos existed and urgency did not, the paper trail is not a defense. It is evidence of a translation failure between observation and action.
Closing The Loop Without Pretending The File Is Closed
So where does that leave a reader who is not paid to sit on a Board? It leaves you with a sharper eye for business models that rhyme with 2023. One industry. Big uninsured balances. A bond book that works only if rates stay polite. A capital story that begins after a sale rather than before it.
The latest review will be quoted selectively. That is how these documents live. Quote the part that says staff should have seen the fragility. Quote the part that implies earlier leadership did not force the issue. Quote the political subplot if that is your beat. Just do not lose the boring center: a visible risk mix was allowed to sit in plain sight until depositors did the supervising themselves.
I started with a warning light. I will end with one. The next bank that looks “unique” in a flattering way will also look unique in an ugly way when the cycle turns. Unique is not a synonym for safe. If this report does anything useful, it is to make that sentence harder to laugh off in a conference room.
And if the argument now shifts from the bank that failed to the people who were supposed to tap the glass before it cracked, well. That argument was always coming. The only surprise is how long it took the official language to catch up with what the balance sheet had already been saying.
