September 2026: The Sell-off in Global Bonds and the Ghost of the 2023 US Bank Failures
The Sell-off in Global Bonds and the Ghost of the 2023 US Bank Failures:
Rising yields suggest a fiscal policy fork in the road coming between elevated inflation with stagflation risk or recession – exposing familiar cracks in US bank supervision
In my last few Treasury Talk columns, I sketched out a set of cycle-end risks, pointing to a Fed late to tightening (growing divergence from a Taylor Rule), a stretched AI narrative, and expanding geopolitical challenges adding to H4L commodity prices, inflation, rising long-term interest rates globally and what that means for asset prices, including housing, CRE and equities.
As we move into the fall, pressures are mounting. At two banking conferences I spoke at this month: I noted that the main question is in light of the negative supply shock stemming from rising commodity prices and strained global shipping do we get
1) more fiscal stimulus to “compensate” the consumer and “just” more inflation with the associated rise in long-dated Treasury yields – which could trigger a selloff in other assets and result in stagflation
or
2) demand destruction due to elevated inflation and rising risks of a recession
And, in an arguably tone deaf moment, this month Federal Reserve Vice Chair of Supervision Bowman (really, bringing up SVB as the 10-year Treasury yield is well above 5%?) reminded the world to ask the question– 3 ½ years after the failures of Silicon Valley Bank (SVB), First Republic (FRC) and Signature (SBNY), are US bank regulators any better prepared on interest rate and liquidity?
Judging by her speech and the Fed’s bank supervision manual, the answer seems to be no. I also discuss the macro and banking set up in spring 2023 vs today.
Recapping the Economic Outlook: More Inflation Shocks – But How Do They End?
We are in the midst of the most significant, unfolding global commodity price shock since the 1970s. The ongoing closure and hostilities in the Strait of Hormuz, mounting conflict in the Red Sea, and the escalation of the war in Ukraine—where Ukraine continues to bravely defend itself from Russia without sufficient US support —have collectively put a chokehold on energy and agricultural exports, notably refined products. Extremely elevated diesel prices (white) will notably impact inflation as a result of its extensive use in shipping and agriculture. Headline oil prices (pink) understate the stress.
The 321 crack spread (measuring the gross margin a refiner earns from converting 3 barrels of oil into 2 barrels of gasoline and 1 barrel of diesel) shown above [one-month futures (red) and six month futures (yellow)] illustrates that refining capacity is the crucial constraint on the global economy, not oil. Simply, Europe largely closed its refineries, and this activity relocated to the Gulf and China. US gasoline inventories now are akin to 2022 Russia/Ukraine levels.
We could see US CPI “merely” rise into a 5-6% range or, with significant new fiscal stimulus, more of a Covid style inflation surge up to 8%+.
Yet, this negative supply shock is only half the story. The US economy is also grappling with other inflationary pressures that the Fed did not durably quell. Q1’s OBBBA tax refunds and Q2’s surprise SCOTUS tariff refunds have eased US fiscal policy and contributed to US inflation remaining elevated.
Consequently, service inflation (in July core PCE, table 1) and supercore inflation (August CPI, table 2) remain elevated before the peak of global commodity shocks have arrived.
Also of concern for any central bank, consumer inflation expectations, according to the University of Michigan survey (white), are rising anew.
The global inflation outlook coupled with the possibility of changes to US inflation measurement (note 30s modestly underperforming 20s) have caused a revolt in long-dated Treasuries. Simply, TIPS’ real rates have repriced to anticipate Fed tightening, except at the one-year point where the decline in the real rate reflects higher near-term US inflation expectations.
While in my August column, I had expressed skepticism about an FOMC consensus to tighten in Sept, I changed my callafter the August data labor and CPI releases to a 25 bps rate hike in Sept.
At the Sept FOMC meeting, the Committee showed an ability to take some action to prevent inflation from getting even further above target. In short, the FOMC finally prioritized the back-end of the Treasury curve which the bond market briefly appreciated.
That said, Warsh did not acknowledge that the long-end Treasury selloff reflected a real rate repricing of the Fed’s own policy path due to higher inflation – not higher growth.
Warsh’s de-emphasis of fiscal policy and greater emphasis on geopolitics driving inflation also was notable.
One wonders does his geopolitical focus imply that he thinks the policy path is less steep than the market (dare I say inflation could be more “transitory”) than the market which is pricing 3 more Fed rate hikes.
Questions were highly controlled at the Sept press conference and Warsh was better prepared, contributing to a better outcome than the July press conference. But communicating less also means what gets priced by the market may play an outsized role of shaping the policy path.
Turning back to geopolitics, it is no secret to Iran—and China, for whom this ill-conceived and US initiated conflict is a proxy war—that the US mid-term elections are rapidly approaching. To maximize leverage, they have a strong incentive to keep upward pressure on commodity prices, inflation and the US bond market.
Therefore, I expect that elevated commodity prices into the mid-terms will drive expectations of a 25 bps Fed rate hike in October to 100% and that the FOMC will deliver this.
Like betting markets, I do not anticipate a normalization of shipping, commodities, and price pressures much before Q2 2027 (white line is July 2027, blue is Jan 2027).
The Social Security cost of living adjustment (COLA) also will be set soon in the US and looks likely to come in at 3.6%+ which can be anticipated to create “echo” inflation in early 2027, much as it did after the Covid inflation shock. As a result, my baseline is that US inflation will remain elevated through H1 2027.
Any eventual de-escalation or normalization of the geopolitical situation may well wait until H2 2027, when China assumes the rotating presidency of the BRICs. It seems plausible to expect that China would like the “win” of stabilizing the global geopolitical situation and economy.
For bank treasury teams trying to navigate this painfully complex geopolitical/commodity/inflation/interest rate and economic outlook, the baseline of “higher-for-longer” (H4L) commodity prices into late H1 2027 points to two different macro paths but which sadly could share a similar final destination:
1. The More Fiscal Subsidy/Inflation/Elevated Interest Rate Path[1]: To shield consumers from energy shocks, governments could further increase consumer subsidies. According to IEA data, this kind of public policy response has already begun globally.
Fiscal transfers to consumers into a negative supply shock guarantees H4L inflation that drives global long-dated government bond yields higher and may pop the AI bubble and/or pressure US housing, risking a move from “just” H4L inflation to a stagflationary scenario
It would not be surprising if further increases in long-term real yields (green) deflated the AI-bubble in the US equity market (white) as happened during the Dot.combubble.
2. The Demand Destruction Path: Alternatively, elevated commodity prices without fiscal transfers could lead to outright demand destruction as consumers pull back on discretionary spending and companies defer investment, rapidly accelerating the cycle-end/recession risks in multiple countries in a synchronized fashion.
As US stagflation/recession risks are not evident in the current data due to fiscal stimulus in Q1 and Q2, the forward risks of recession/stagflation appear underpriced. There are opportunities for banks in this time to reduce any areas of potential credit concern.
Beyond these near-term risks lies a deeper structural risk arising from geopolitical developments that I note in a talk I gave to large banks in the spring at CEFPRO’s 2026 NYC bank treasury conference. There I spoke about the UK’s historical transition away from being the global reserve currency. We may be approaching a modern “Suez Canal” moment for the US. There are risks to the petrodollar system unraveling, with profound, permanent consequences that could erode the US dollar’s reserve currency status and place enormous strains on the Gulf States. The Gulf states may need to generate dollar liquidity which could further pressure long-dated Treasury yields higher.
The Ghosts of the Spring of 2023
This brings me to a second part of this month’s column – looking back to the spring of 2023 US bank failures due to excess interest rate and liquidity risk.
This month Federal Reserve Vice Chair for Supervision Michelle Bowman delivered a speech to highlight “initial” findings of a new “independent” review of SVB’s 2023 failure. As the two charts below suggest, it is remarkable that she chose to draw public attention back to this topic given renewed strains in bond markets and the H4L inflation outlook which suggest some downward repricing in US bank equity on the horizo
According to Bowman, the new SVB review confirmed that Fed supervisory staff knew about SVB’s interest rate risk and uninsured deposit concentration vulnerabilities in 2022. However, supervisors failed to take prompt action due to a long-standing, deep-seated culture of risk aversion, where staff believed it was “personally safer to take no action”. Per Bowman, this risk-averse posture was compounded by a lack of clarity regarding supervisory decision rights. It is remarkable to see Bowman in a speech criticize Fed bank examiners for a culture that her own leadership of the regional bank portfolio contributed to creating.
In her speech, Bowman highlighted some supervisory changes, such as the Statement of Supervisory Operating Principles to refocus examiners on “material” safety and soundness threats, and a new requirement for monthly escalation reports on areas of “supervisory uncertainty.”
Really, 3 ½ years after the bank failures and we still have “supervisory uncertainty” on interest rate and liquidity risks? By choosing “monthly escalation reports” over amending the supervision manual and adopting any quantitative objective triggers, the Vice Chair and Federal Reserve continue to increase the likelihood that its bank supervision remains ineffective at preventing further US bank failures.
So it seems that what has changed at the Fed since SVB’s failure is … very little
The US General Accountability Office (GAO)’s 2023 report on SVB’s failure notes that
GAO has longstanding concerns with escalation of supervisory concerns, having recommended in 2011 that regulators consider adding noncapital triggers to their framework for prompt corrective action (to help give more advanced warning of deteriorating conditions).[2]
To restate GAO’s point in my own words, outside of capital ratios generally there are too few quantitative bright lines for bank examiners with regard to prompt-corrective action to reduce likely failures. The continued absence of any Pillar 2 surcharges in US bank capital for significant levels of interest rate risk is notable.
As my coauthors and I detailed in our 2025 SVB paper, we believe that the application of Basel standards could have prevented SVB’s failure in 2023 by creating bright lines that forced ex-ante risk mitigation
– First, the interest-rate risk in the banking book (IRR-BB) standard—which US regulators failed to implement in the aftermath of the S&L crisis — would have flagged SVB’s extreme asset-liability maturity mismatch and triggered mandatory supervisory intervention ten quarters before its failure
– Second, if US regulatory tailoring in 2018 had not exempted banks of SVB’s size from the Liquidity Coverage Ratio (LCR), the bank would have been forced to hold tens of billions of dollars more in high-quality liquid assets, as its LCR would have registered at a non-compliant 75% at the end of 2022.
As we documented in 2025, at the time of SVB’s failure the Federal Reserve’s examination manual for commercial banks was highly limited in its discussion of interest rate risk. Specifically, the manual stated that Fed examiners may only focus on a bank’s compliance with its own “self-imposed parameters.” Because of this guidance, it can be very difficult for a bank supervisor to challenge a bank’s interest rate risk as being excessive if the bank is operating within the net interest income (NII) and economic value of equity (EVE) risk limits approved by its own board.
The relevant section (3300) of Federal Reserve’s commercial bank examination manual still indicates no changes or updates for examiners have taken place with regards to IRR-BB since SVB’s failure.
I am grateful to the Financial Times last week for sharing these concerns with a broader audience. You can also read more on my Substack.
But the aggregate picture of US banks’ unrealized losses at the time of SVB’s failure makes clear that weak US supervision of interest rate risk was a systemic issue, not the problem of just a few supervisory teams. While US banks’ aggregate unrealized securities loss position has improved since SVB, there nonetheless remain some banks with outsized risk (thanks to Joe Gradzki at Bankview).
· As of Q2 2026 (before the recent global bond market selloff), there were 17 US banks with a total combined assets of $12.8 billion with unrealized AFS+ HTM losses > 66% of T1 + ALLL.
Like SVB in 2023, three of these banks, with total combined assets of $2 billion have unrealized securities losses > capital + ALLL and, unsurprisingly, very weak ROA.
· Turning to large US bank charters (ex-FBOs) with assets > $100 billion, there remain some large banks with significant unrealized securities losses and/or meaningful long dated loan exposures (likely jumbo residential mortgages). Much of the unrealized loss for these banks sit in HTM, implying little of it is hedged given current GAAP rules. A number of the banks with outsized exposure to low yielding assets also have weak ROA relative to peer.
Notes: Bankview, US call report, bank IDIs only, excludes FBOs and ROA calculated by multiplying H1 2026 earnings x 2.
· While it is certainly true that all these banks did weather the spring of 2023, the key question becomes whether meaningful credit losses show up on the scene in 2026/2027 in a way that they didn’t in 2023.
At the end of the day, the 2023 bank failures did not produce a full-blown US banking crisis. With hindsight, in 2023 US fiscal policy put down very significant foam on the runway and prevented credit losses from materializing even as US banks struggled with over $600 billion in unrealized securities losses, totaling ~30% of US banks Tier 1 capital + ALLL in Q4 2022.
It remains unclear whether US policymakers and bank regulators today fully understand this, but the absence of significant credit loss further straining some US banks’ capital on top of interest rate losses helped avoid a US banking crisis.
· In FY2023, the US federal budget deficit widened officially from 5.4% of GDP to 6.3% of GDP (a large 0.9% of GDP increase).
· However, stripping out the non-cash Supreme Court accounting adjustments for student loan forgiveness, the cash component of the US budget deficit actually doubled from roughly 3.7% of GDP to 7.5% of GDP—a massive fiscal expansion of nearly 3.8 percentage points of US GDP.
· The Administration’s $5000 “dividend” check proposal bears some similarities – it certainly would fuel more inflation, but perhaps also put foam down on the runway. However, such a move could be expected to fully unanchor US inflation expectations.
Conclusion: As we head toward 2027, bank treasury teams find themselves at a critical macro-regulatory crossroads. Whether the global economy ultimately steers toward “merely” an inflationary environment fueled by persistent negative commodity supply shocks, or hits a wall of rapid, synchronized demand destruction, the final destination remains both uncertain and risky.
With hindsight, the lesson of 2023 is not that the US regulatory system successfully contained a banking crisis. Rather, it is that a massive expansion in the US budget deficit put down enough “foam on the runway” to keep credit losses at banks at bay even as banks faced very large unrealized securities losses.
As long-term Treasury yields surge and US fiscal capacity faces its own structural limits, bank treasurers cannot count on US policymakers to cushion this landing. Inflation expectations are increasingly at risk of becoming entrenched.
It is unfortunate that the US bank regulation remains largely unchanged since the spring of 2023. Rather than adopting objective, quantitative bright lines of interest rate and liquidity risk we have “escalation reports.” This suggests that US bank supervision will continue to lag.
Ultimately, bank treasury teams and boards must not rely on regulators to save them. Instead, they must rigorously stress-test their own balance sheets, review self-imposed risk limits, and proactively manage/de-risk credit exposures. The ghosts of 2023 are still whispering; it is up to bank management and boards, rather than supervisors, to take heed and listen.
My sincere thanks to Treasury Talk readers.
Jill Cetina, CFA is Executive Professor of Finance at the Mays Business School, Texas A&M University, BTRM North America director, a former Associate Managing Director for US banks at Moody’s, former Federal Reserve officer and former financial economist at the OCC. The views expressed here are her own.
[1] Arguably it also is plausible that US consumers continue to dissave into H4L commodity prices (note that this still is more like path #1 with inflation staying elevated and bond markets not liking it). US consumer continued dissaving seems more likely through Q4 2026 with the holidays at hand.
[2] For non-US readers, GAO is an investigative agency that is independent of the Executive branch and conducts independent reports and studies for Congress.
