Fed's Bowman Previews Stress-Test Changes and Releases Initial Findings of Outside Review of SVB's 2023 Failure
In two London speeches on Sept. 18, 2026, the Federal Reserve's top bank regulator outlined changes to annual stress tests and said an outside review found supervisors knew or should have known of Silicon Valley Bank's weaknesses by March 2022.
Two Speeches, One Question: Was SVB's Collapse a Rules Problem or a People Problem?
Michelle Bowman gave two speeches in London on September 18, 2026, and they were meant to be read together. In the first, the Federal Reserve's top bank regulator previewed changes to the annual stress test that decides how much extra capital big banks must hold[1][6]. In the second, she released the first findings from an outside review of how the Fed supervised Silicon Valley Bank before it collapsed in March 2023[2].
The two speeches share a spine. If the review finds that weak rules caused SVB's failure, the case for loosening capital requirements gets harder to make. If it finds that supervisors simply failed to act on rules that were already strong enough, the case gets easier[3][4][5]. That is why a report about a bank that failed three years ago is landing in the middle of a live fight over rules for the banks operating today.
What the Records Already Showed
Some facts here were never in dispute. About 94% of SVB's deposits were uninsured, sitting above the FDIC's $250,000 protection limit, and were concentrated among venture-backed tech companies that could all move their money at once[13]. The bank held long-term securities whose paper losses were bigger than its entire capital cushion, and it hadn't done the paperwork needed to borrow quickly from the Fed if depositors ran[2][13]. When they did run, in March 2023, the bank failed[5][13].
Bowman said the new review found that Fed examiners knew, or should have known, about these vulnerabilities as early as March 2022, a full year before the collapse[2][6]. Nobody disputes that supervisors were too slow. Even the Fed's own original 2023 post-mortem, written under then-Vice Chair for Supervision Michael Barr, agreed examiners had been too cautious[3][13].
Where the accounts split is on why. The new review, done by a firm called Starling Advisory Group, points to what Bowman described as a "culture of risk aversion" inside the Fed: staff felt it was safer to write a finding and do nothing than to act and risk being wrong[2][3]. Barr's 2023 report pointed instead at a 2018 law that loosened supervision requirements for regional banks like SVB, arguing weaker rules blunted examiners' authority and appetite to enforce them[3][5]. Bowman said the new review found that law played no role in the delays[4].
Why a Number on a Model Nobody Can See Is Worth Billions
The stress test at the center of Bowman's other speech works like this: the Fed runs a hypothetical severe recession through a bank's books and checks whether it survives. How badly a bank does sets its "stress capital buffer," extra capital it must hold on top of the legal minimum[1]. Capital held for that buffer is money that can't be lent out or returned to shareholders, so every percentage point is a real cost to a bank's bottom line[11].
The Fed has long kept the exact models and scenarios behind that test secret. Bowman argues that a requirement banks can't see, can't comment on, and can't predict from one year to the next isn't really a rule at all, it's a moving target that forces banks to hold buffers on top of buffers just in case[1][9]. Her fix, expected to be finalized in the coming weeks, is to average a bank's two most recent test results and publish more detail on the models behind them, smoothing out year-to-year swings without lowering the bar, in her telling[1][6].
Large banks and their trade groups back this hard. The Bank Policy Institute sued the Fed in 2024 to force the stress test through the normal public rule-making process[3][11]. They also point to this year's test results, which found banks could absorb $708 billion in hypothetical losses and still stay above their minimums, as evidence the system already has plenty of cushion[14].
The Reviewer Reviewing Itself
Here is where the story turns. Senate Banking Democrats, led by Elizabeth Warren, aren't just disputing the review's conclusions, they're disputing who was allowed to write them. Starling Advisory Group's own earlier study on bank supervision lists Randal Quarles, the Republican who ran Fed supervision when the 2018 tailoring law took effect, as study chair. Its advisory board includes Gary Cohn, a Trump economic adviser during his first term[5].
Warren said Bowman "wasted taxpayer resources to hire what appears to be her partisan friends" for an "autopsy that magically and conveniently exonerates her and President Trump," and she's called for an independent investigation into how Starling was hired[5]. The Fed engaged Starling around January 2026, roughly seven months after Bowman was sworn in as Vice Chair for Supervision in June 2025[10][15].
There's a strongest case for Bowman here too, even setting the Quarles connection aside. Her argument is that no amount of capital fixes a culture where the safe career move is to identify a problem and not force a fix. On that view, tightening rules again would treat the wrong disease, examiners had the numbers in March 2022 and didn't act, regardless of what the 2018 law technically allowed[2][3].
A Second Use for the Same Report
There's a use for this review that has nothing to do with bank capital. Michael Barr stepped down as Vice Chair for Supervision in February 2025 but remains a sitting member of the Fed's Board of Governors[8][13]. Governors can only be removed "for cause," a legal bar no administration has ever tested against a supervisory failure.
Bloomberg reported in July 2026, ahead of the findings' release, that Trump allies were eyeing the SVB review as a possible path to argue Barr should be removed, though the White House was reportedly cool to the idea[8]. A documented finding that warnings were missed on Barr's watch would be the closest thing to a factual basis anyone has assembled for that argument, whatever it ultimately means for bank rules[8].
Coverage split largely along these same lines. ABA Banking Journal, published by the bank lobby, led with the tailoring exoneration as its headline finding[4]. CNN led with "epic failure" of Fed staff before pivoting to Warren's conflict-of-interest charge[5]. Reuters covered the stress-test mechanics with little mention of the political fight at all[6], while PYMNTS isolated the finding that social media didn't cause the bank run, the most counterintuitive detail, without mentioning the tailoring dispute that's actually driving the argument[10].
What Nobody Has Actually Read
Every characterization in this fight, from every direction, rests on the same thin foundation: Bowman's spoken summary of the findings. As of September 21, 2026, the Fed had not published the full Starling report, its methodology, or the documents and interviews behind it[2][5]. The praise for the tailoring finding and the attacks on Starling's ties to Quarles are both aimed at a text nobody outside the Fed has actually seen.
The stress-test rule, by contrast, is not waiting on anyone. Bowman said the Fed expects to finalize the averaged, more transparent framework within weeks[1]. The Fed itself has said that rule, combined with other planned changes to international capital standards, is expected to lower the amount of capital large banks are required to hold[9]. Whether the full SVB review changes that timeline, or gives Barr's defenders new grounds to push back once the underlying documents are public, is still open.
Summary
On Friday, Sept. 18, 2026, Federal Reserve Vice Chair for Supervision Michelle Bowman gave two speeches in London. In the first, she previewed changes to the Fed's annual bank stress test. In the second, she released initial findings from an outside review of how the Fed supervised Silicon Valley Bank before it collapsed in March 2023[1][2].
The review was done by Starling Advisory Group, hired by the Fed around January 2026 — roughly seven months after Bowman was sworn in as Vice Chair for Supervision in June 2025[10][15]. Per Bowman, it found that Fed supervisors knew, or should have known, about SVB's key weaknesses as early as March 2022 — a year before the bank failed — but did not force the bank to fix them[2][6]. The review blamed a long-standing internal "culture of risk aversion": staff felt it was safer to do nothing than to act and be wrong[3][13]. Two other findings are the politically explosive ones. Bowman said the review found the 2018 law that loosened rules for large regional banks did not cause the supervisory delays[4]. And it found no evidence that social media caused the run[10][13].
Those last two points cut against the Fed's own 2023 post-mortem, written under then-Vice Chair for Supervision Michael Barr, which pointed at the 2018 rollback[3][5]. Democrats reject the new account. Sen. Elizabeth Warren, the top Democrat on the Senate Banking Committee, said Bowman "wasted taxpayer resources to hire what appears to be her partisan friends" for an "autopsy that magically and conveniently exonerates her and President Trump," and called for an independent investigation of the hiring[5]. Starling's own earlier supervision study lists Randal Quarles — Bowman's Republican predecessor as the Fed's top supervisor — as study chair[5].
The single sharpest dispute is not whether Fed examiners were slow. Everyone, including Barr's 2023 report, concedes they were. It is why: whether the cause was weaker rules after 2018, or a supervisory culture that saw the problem and would not act. That answer decides whether the right fix is more regulation or less — and, separately, whether the findings hand Trump allies a legal argument to remove Barr, who remains a Fed governor[8].
The Event
On Sept. 18, 2026, Fed Vice Chair for Supervision Michelle Bowman spoke twice at Mansion House in London. In one speech she said the Fed will consider a final version of its revamped stress-testing framework in "coming weeks," including averaging a bank's two most recent test results and disclosing more of the Fed's models and scenarios[1][6]. In the other she released initial findings from the Starling Advisory Group's independent review of Fed supervision of Silicon Valley Bank, which the Fed engaged around January 2026[2][10][15]. Both speeches were posted on the Federal Reserve Board's website the same day; the full Starling report had not been published as of Sept. 21, 2026[1][2].
Undisputed Facts
- Silicon Valley Bank failed in March 2023 after a run by depositors[5][13].
- About 94% of SVB's deposits were uninsured — above the $250,000 FDIC limit — and were concentrated among venture-capital-backed tech companies[13].
- SVB held long-term securities whose unrealized losses exceeded its capital, and had not done the paperwork and collateral prep needed to borrow quickly from the Fed's discount window[2][13].
- The Fed engaged Starling Advisory Group around January 2026 to review its own supervision of SVB, roughly seven months after Bowman's June 2025 swearing-in as Vice Chair for Supervision[10][15].
- Bowman said on Sept. 18, 2026 that the review found supervisors knew or should have known of SVB's vulnerabilities as early as March 2022[2][6].
- Bowman said the review found that the 2018 regulatory tailoring mandate did not contribute to the delays in supervisory action[4].
- The Fed's own April 2023 post-mortem, produced under then-Vice Chair for Supervision Michael Barr, reached a different conclusion about the role of the 2018 rollback[3][5].
- Michael Barr stepped down as vice chair for supervision in February 2025 but remains a member of the Fed's Board of Governors[8][13].
- Bowman said the Fed will average a bank's two most recent stress-test results to set its capital requirement, and will publish more detail on the models and scenarios behind the test[1][6].
- As of Sept. 21, 2026, the Fed had released Bowman's summary of initial findings; the full Starling report was not public[2][5].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Causation decides the policy
- Every side's position on SVB's cause maps exactly onto its preferred rule change. If weak rules caused it, you tighten. If slow examiners caused it, you fix management and can safely loosen the rules. The historical question is not being argued for its own sake[3][4][5].
- Capital is money not earning a return
- Capital held against a stress scenario is money that cannot be lent or returned to shareholders. Every percentage point of the stress capital buffer is a direct, measurable cost to large banks, which is why industry funded a 2024 lawsuit over the test's opacity rather than just lobbying[11].
- A legal predicate for removing a Fed governor
- Fed governors can only be removed "for cause," and no administration has tested what that means for a supervisory failure. An official finding of missed warnings on Barr's watch has value to the administration independent of what it says about bank regulation[8].
- The evidence is still sealed
- As of Sept. 21, 2026, the public has Bowman's spoken summary of initial findings, not the Starling report, its methodology, or its document set. Both the praise and the attacks are aimed at a text no outside party has read[2][5].
Material realitySVB's balance sheet facts are not in dispute and were visible in filings before it failed: about 94% of deposits uninsured, concentrated in one industry, funding a portfolio of long-dated securities whose unrealized losses exceeded the bank's capital, with no ready access set up at the Fed's discount window[13][2]. Fed examiners had that picture by March 2022 and the bank failed in March 2023[2][6]. Nothing in the current fight changes those facts. What is genuinely unsettled is whether examiners lacked authority or lacked will — and that question turns on internal documents and interviews the public cannot yet see. Meanwhile the operative decision is forward-looking: the Fed will finalize a stress-test framework in the coming weeks that averages two years of results and discloses more of its models, and the Fed itself expects the combination of that rule, Basel implementation and GSIB surcharge changes to lower required capital at large banks[1][9].
Narrative as a weaponBowman wants you to read SVB as a people-and-culture failure, because that story supports transparency-and-predictability reforms and drains the case for re-regulating regional banks. Bank industry groups amplify the tailoring finding hardest, because it retires the strongest objection to the capital cuts they have sought since 2023. Warren and Senate Banking Democrats want you to disqualify the messenger — the Quarles and Cohn links to Starling do the work of a rebuttal without requiring one, since the report itself is not public. Trump allies have a third use for the same document: as a paper trail for removing Michael Barr from the Fed board, a goal unrelated to bank capital. Barr's 2023 report is the counter-text, and it agrees with Starling that examiners were too slow — the two accounts diverge on cause, not on conduct. Readers should hold all characterizations loosely until the full Starling report is published.
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asBowman's case is that supervision failed as a practice, not as a rulebook. Fed examiners had SVB's numbers in front of them in March 2022 and did not act. On her telling, no capital ratio fixes a culture where the safe career move is to write a finding and not enforce it[2][3]. On the stress test, her argument is about the rule of law as much as about capital. The stress test sets a bank's "stress capital buffer" — extra capital it must hold on top of the minimum, sized by how badly it does in the Fed's hypothetical recession. The Fed long kept the models and scenario details secret. Bowman argues a requirement you cannot see, cannot comment on, and cannot predict from year to year is not a real rule; it swings capital targets for reasons banks cannot trace, so they hold buffers-on-buffers instead of lending[1][9]. Averaging two years of results, she says, smooths that swing without lowering the bar[1][6].
WhyShe wants to finish and defend a deregulatory program — stress-test reform, Basel changes, a lower GSIB surcharge — and a finding that 2018 tailoring was not to blame removes the strongest argument against it[9][11]. She also has an institutional interest in locating the SVB failure in career staff conduct rather than in Republican-era rule changes[4][5].
Impact on themThe stress-test rule is hers to finalize in the coming weeks; the SVB findings are the evidentiary base she will cite for it[1]. The Starling hiring is now itself under attack, with Warren seeking an independent investigation of it[5].
Frames it asTheir argument is that an independent review is only as independent as the firm you pick. Starling's own recent bank-supervision study lists Randal Quarles — the Republican who ran Fed supervision when the 2018 tailoring rules were written — as study chair, and Gary Cohn, a Trump first-term economic adviser, sits on its advisory board[5]. A review that clears tailoring and clears Quarles, commissioned by Quarles's ideological successor, is not evidence, they argue — it is a commissioned verdict[5]. Their substantive point is separate and stronger: SVB crossed $100 billion in assets and, because of the 2018 law, escaped the toughest liquidity and stress-testing regime. Whatever examiners felt culturally, the tools they would have needed had been taken off the table[3][5]. Warren's frame is that the report exists to justify the next round of cuts[5].
WhyThey want to block the stress-test rewrite and the broader capital rollback, and to protect the 2023 Barr report as the official account of SVB[3][5]. They also want to deny the administration a paper trail for removing a Fed governor[8].
Impact on themIn the minority, their leverage is oversight letters, hearings and an investigation request — not votes[5]. If the tailoring finding sticks, the post-2023 case for re-tightening rules on regional banks loses its anchor.
Frames it asBanks say the stress test has been run as a black box. BPI sued the Fed in 2024 to force the scenarios and models through public notice-and-comment, the ordinary process for a binding rule[3][11]. Their argument is procedural first: if the Fed can change a model and move your capital requirement by billions without showing the math, the requirement is unreviewable. Their second argument is that the year-to-year swings are not risk signals but artifacts of scenario design, so averaging two years measures the bank rather than the Fed's latest scenario[1][6]. They also note the 2026 test found large banks could absorb $708 billion in hypothetical losses and stay above minimums — evidence, they say, that resilience is not the binding problem[14].
WhyLower and more predictable capital requirements free up money for buybacks, dividends and lending. Predictability also lets them plan capital returns a year out instead of guessing[9][11].
Impact on themThe Fed expects Basel implementation and GSIB surcharge changes, alongside the stress-test rewrite, to lower the capital large banks must hold[9].
Frames it asBarr's 2023 report argued that supervisors were slow and that the 2018 tailoring changes plus a shift in supervisory tone had weakened both the rules and the appetite to enforce them[3][5]. Defenders of that view note the Starling summary and the Barr report agree on the central fact — examiners were too cautious — and differ only on the cause, which is the part that cannot be settled by documents alone[3][13]. For career staff, the "culture of risk aversion" finding is a charge against people who cannot answer publicly, based on a report readers cannot yet read[2][13].
WhyBarr has a personal stake beyond the historical record: Trump administration officials have privately discussed whether the review's findings could supply "cause" to remove him from the Board of Governors, though the White House has reportedly been cool to the idea[8].
Impact on themFed governors can be removed only "for cause," a term never tested in court for a supervisory-judgment failure. A documented finding of missed warning signs on Barr's watch is the closest thing to a factual predicate anyone has assembled[8].
Like this article?
The Bias Ledger average rating 4
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| Reuters | U.S./international wire, center | 2 | "Fed's Bowman says changes to bank stress test coming soon" — mechanics of the stress-test overhaul, with Bowman's "opaque and unnecessarily unpredictable framework" line quoted[6]. | Covers the two speeches as separate stories; the wire piece on stress testing barely touches the SVB review or the political fight over who hired Starling. |
| Banking Dive | U.S. trade press, center | 3 | "3 takeaways from stress test changes and a new SVB review" — treats the two speeches as one coordinated policy event[3]. | The takeaways format gives the tailoring finding, the staff-culture finding and the BPI lawsuit equal billing, which flattens the contested question of causation into a list. |
| Bloomberg | U.S. center, financial | 4 | "Trump Allies Eye SVB Review as New Path to Oust a Fed Governor" — reported in July 2026, ahead of release, framing the review as a removal vehicle aimed at Barr[8]. | Frames the review by its political use rather than its content; notes the White House was unenthusiastic, which cuts against its own headline premise. |
| PYMNTS | U.S. payments-industry trade press | 4 | "Fed's Bowman Challenges Social Media Story Behind SVB Collapse" — the finding that social media did not drive the run[10]. | Isolates the most counterintuitive finding for clicks and drops the tailoring dispute entirely, which removes the reason anyone is arguing about the report. |
| CNN | U.S. center-left | 5 | "Report faults Fed staff for epic failure of Silicon Valley Bank" — leads on staff blame, then pivots to Warren's conflict-of-interest charge and Starling's Quarles and Cohn ties[5]. | The word "epic" is CNN's, not the report's. Placing the Quarles link high signals the reader should discount the finding before reading it. |
| ABA Banking Journal | U.S. bank-industry trade press, published by the American Bankers Association | 6 | "Bowman: Independent review finds regulatory tailoring had no role in SVB failure" — the tailoring exoneration is the headline[4]. | Selects the one finding most useful to its members and puts it in the lede; the "culture of risk aversion" and the unread full report are secondary. Trade association ownership is the orientation, not a hidden one. |
References
- The Final Chapter on Modernizing Bank Regulatory Stress Testing — speech at Mansion House, London, Sept. 18, 2026 — Federal Reserve Board · U.S. central bank; primary source, speaker is the Fed's Republican-appointed vice chair for supervision
- Speech by Vice Chair for Supervision Bowman on initial findings from independent review of Silicon Valley Bank — Federal Reserve Board · U.S. central bank; primary source, the Fed summarizing a review of itself
- 3 takeaways from stress test changes and a new SVB review — Banking Dive · U.S. banking trade publication owned by Industry Dive/Informa; ad-supported, center
- Bowman: Independent review finds regulatory tailoring had no role in SVB failure — ABA Banking Journal · Published by the American Bankers Association, the main U.S. bank lobby; industry-aligned
- Report faults Fed staff for epic failure of Silicon Valley Bank — CNN · U.S. cable/digital news, center-left
- Fed's Bowman says changes to bank stress test coming soon — Reuters · International wire service; center, read here on Investing.com
- Rethinking Silicon Valley Bank's failure — Axios · U.S. digital news, center
- Trump Allies Eye SVB Review as Path to Oust a Fed Governor — Bloomberg · U.S. financial news, center; owned by Bloomberg L.P.
- Fed's Bowman previews further stress test reforms ahead of finalization — American Banker · U.S. banking trade publication owned by Arizent; subscription-funded, center
- Fed's Bowman Challenges Social Media Story Behind SVB Collapse — PYMNTS · U.S. payments-industry trade site; sponsor- and vendor-funded
- BPInsights: September 19, 2026 — Bank Policy Institute · Lobbying and research group funded by the largest U.S. banks; plaintiff in the 2024 stress-test lawsuit
- Independent Review of the 2023 U.S. Bank Failures — Federal Reserve Board · U.S. central bank; primary source landing page
- Independent SVB Review Finds Fed Supervisors Knew Of Risks, Failed To Act; Social Media Didn't Cause Run — CU Today · U.S. credit-union industry trade publication; competes with banks, so not bank-lobby aligned
- Federal Reserve stress test: U.S. banks can withstand $708B in losses — CNBC · U.S. business news, center; owned by Comcast/NBCUniversal
- Report finds 'culture of risk aversion' hindered oversight of SVB, Fed official says — Reuters · International wire service; center