Bank Stocks and Bond Losses: Reading the October 2026 Sell-Off

Bank stocks fell on three continents in the first week of October 2026. US banks lost almost 10% in a month while the S&P 500 rose. The popular explanation is that rising yields have wiped value off the bonds banks hold. It fits some banks well and others hardly at all. That matters if you are deciding which bank, if any, you would be happy to own.

This post goes through the moves and the figures for each bank. The last section is our AI-assisted read. It is neither a forecast nor advice.

What happened

The trigger was the bond market. The Fed raised rates by 0.25 percentage points on 16 September, and its minutes pointed to another rise by the end of the year. On 7 October the 10-year Treasury yield touched 5.36% and the 30-year 5.73%, both the highest since 2002, as government bond yields rose in the UK, France and Italy too. The 10-year closed at 5.24% on 9 October.

Banks took it badly. Price moves to the 9 October close:

	1 week	1 month	From 3-month high
US banks (KBE)	-1.9%	-6.6%	-11.4%
US regional banks (KRE)	-2.5%	-6.0%	-10.9%
Bank of America	+1.1%	-13.2%	-15.8%
Charles Schwab	0.0%	-9.9%	-14.9%
JPMorgan	+0.7%	-5.3%	-8.4%
European financials (EUFN)	-1.6%	-8.2%	-10.9%
Mitsubishi UFJ (ADR)	-1.9%	-3.9%	-7.3%
OCBC	-8.4%	-7.6%	-10.1%
UOB	-7.3%	-2.5%	-9.2%
DBS	-5.0%	-4.5%	-6.7%
S&P 500 (SPY)	+1.2%	+3.0%	-0.1%

The KBW Nasdaq Bank Index fell 9.7% in the month to 9 October, and US banks had their worst September against the S&P 500 since 1990.

What a bond loss does to a bank

Banks hold large portfolios of government and agency bonds, much of it bought when yields were far lower. When yields rise, those bonds are worth less. The FDIC put total unrealised losses on securities at US banks at $326.7 billion on 30 June 2026: $109.8 billion on bonds available for sale and $216.9 billion on bonds held to maturity.

How much of that loss counts depends on the label.

• Held to maturity. Carried at cost. A fall in market value touches neither profit nor regulatory capital, as long as the bank keeps the bonds until they mature.
• Available for sale. Marked to market through equity. For the largest US banks the loss also comes out of regulatory capital. For mid-sized banks it is currently excluded, though a rule proposed in March 2026 would phase that exclusion out.

A paper loss becomes a real one only when the bank sells. That is what happened at Silicon Valley Bank in 2023. Depositors pulled money, the bank sold bonds to pay them, and the loss crystallised. A bank can survive bond losses on their own. The danger comes when the same bank is also losing deposits.

Higher rates also help. A bank earns the gap between what it pays depositors and what it earns on loans and bonds. The US yield curve has steepened: the 10-year now pays about 0.44 percentage points more than the 2-year. Most large US banks disclose that higher rates raise their net interest income. JPMorgan, for example, estimates +$1.8 billion for a one-point rise.

Does the bond story explain the sell-off?

Partly. Bank by bank, from each bank's second-quarter 10-Q:

Bank	Unrealised bond losses as % of CET1 capital	Net interest income if rates rise 1 point	Trend at 9 Oct	Q3 results
Bank of America	41.9%	+$1.0B	Below 200-day, falling	14 Oct
Charles Schwab	36%	+3.6%	Below 200-day, falling	15 Oct
Truist	26.9%	Slightly down	Below 200-day, falling	16 Oct
US Bancorp	26%	Flat	Below 200-day, falling	15 Oct
Wells Fargo	25.9%	+$1.3B	Below 200-day, falling	13 Oct
Morgan Stanley	11.1%	+ (wealth unit)	Below 200-day, falling	14 Oct
PNC	10.3%	Flat	Below 200-day, falling	15 Oct
Capital One	9.8%	+0.1%	Below 200-day, falling	20 Oct
Citigroup	7.3%	+$1.23B	Above 200-day	13 Oct
JPMorgan	6.9%	+$1.8B	Above 200-day	13 Oct
Goldman Sachs	1.0%	n/a (trading-led)	Below 200-day, falling	13 Oct

The loss figures are our own arithmetic: unrealised losses on both bond categories, before tax, divided by each bank's CET1 capital at 30 June. Yields have risen since, so the third-quarter numbers will be larger.

The bond story fits the banks with big bond books. Bank of America and Schwab carry the largest losses relative to capital and fell the most: 13.2% and 9.9% in a month.

It does not explain Goldman Sachs or Morgan Stanley. They fell 13.1% and 11.7% over the same month with very small bond losses. Their exposure is to deal-making and trading, which higher rates slow. These are two separate effects of the same rise in yields, and the next section checks the deal-making half against the data.

Capital One is a third case. Its August card charge-offs rose to 4.16% from 4.12% in July, so the concern there is credit quality.

Is deal-making slowing?

Yes, in the biggest deals and in IPOs. Across the year as a whole, activity is still near a record.

Measure	Third quarter 2026	Compared with
North American M&A, by value	$560.4 billion	Down 23% from a year earlier; lowest since the second quarter of 2025
North American deals of $10 billion or more	6	19 in the second quarter, 12 in the first
Global M&A, by value	$993 billion	Down 41% from the second quarter; first quarter under $1 trillion since mid-2025
Global deals of $10 billion or more	10	26 in the second quarter
Global M&A, year to date	$3.9 trillion	Up 28%, the highest since 2001; deal count down 8%
US IPOs	30 deals, $32.8 billion	$26.5 billion of it was one listing (SK hynix); $6.2 billion without it

So the drop came from big deals. Deal counts have been soft all year, and the quarter fell mainly because buyers stopped signing $10 billion transactions. Renaissance Capital reported "a string of postponements" of IPOs near the end of September and named high yields, resumed rate rises and doubts about AI spending among the reasons.

Bond issuance has not slowed. Bloomberg reported in early September that companies were bringing investment-grade deals forward before yields rose further, and September was reported as the busiest month of the year for high-yield bonds. Lenders are charging more instead. High-yield spreads were reported at about 2.9 percentage points in late September, the widest since April.

Share prices have moved the same way. Over the month to 9 October, advisory firms PJT Partners, Jefferies and Lazard fell 20.1%, 17.7% and 12.2%. Private equity managers KKR, Blackstone and Ares are 36%, 28% and 32% below their one-year highs. JPMorgan, with a broader business, fell 5.3%.

Why this matters beyond banks (our interpretation):

• It is how higher rates reach the economy first. The largest leveraged deals are the most sensitive to the cost of borrowing, so they stop before anything else does. The firms paid on those deals feel it before the wider economy shows a change.
• Private equity is under the most pressure. Funds need sales and listings to return cash to their investors. With both harder and borrowing above 5%, private equity and private credit are where strain would build next.
• The AI build-out is partly paid for with borrowed money. US technology companies have sold billions of euros of bonds this year to fund data centres. If spreads keep widening, the most debt-dependent part of that spending feels it first.
• There is a precedent. In 2022, as rates rose, global deal value fell from $2.2 trillion in the first half to $1.4 trillion in the second, which Refinitiv data showed as the largest half-year swing since 1980.
• One quarter is not a trend. It is measured against a record second quarter, and Morgan Stanley's head of M&A told Reuters he was "not ready to call a slowdown". JPMorgan was reported in September to expect third-quarter investment banking fees up by a mid-to-high teens percentage on a year earlier.

The bank results from 13 October will show investment banking fees for the quarter and, more usefully, what the banks say about deals still to close.

The Singapore banks

DBS, OCBC and UOB fell hard in the week of 5 October. The STI dropped 3.5% on 8 October alone. Two broker notes drove it.

• Citi cut OCBC to sell on 7 October, with a S$27.50 target. Its reasons: flat third-quarter earnings, a narrowing net interest margin as Singapore-dollar fixed deposit rates rise, a price-to-earnings ratio up 46% this year, and a dividend yield only 0.70 percentage points above bond yields. Citi kept DBS at buy and UOB at sell.
• JPMorgan warned the same day that rising long-term yields mean larger mark-to-market losses on bank bond holdings and weaker capital-markets income, and named OCBC the most exposed. HSBC also cut Singapore equities to neutral.

In Singapore, then, bond losses were one of the brokers' reasons. Margins and valuation were the others.

The first-half 2026 results, released on 6 and 7 August, put numbers on it.

	DBS	OCBC	UOB
Net interest margin, 1H26 (1H25)	1.88% (2.08%)	1.73% (down 0.25 pts)	1.78% (1.96%)
CET1 ratio	16.6%	15.7%	15.4%
1H26 losses on bonds held at fair value	-S$136m	-S$263m	-S$418m (including hedges)
Capital return	81 cents 2Q dividend	47 cents interim; S$2.5b capital return by FY26	88 cents interim; S$2b buyback 40% done
Q3 update	5 Nov	6 Nov	4 Nov

1. The first-half bond losses were small. DBS reported a record S$3.08 billion profit for the second quarter alone. Against that, a S$136 million first-half loss on fair-value bonds is modest. Long yields have risen since June, which is JPMorgan's point, and the third-quarter updates will show by how much.
2. The bigger pressure is margins. All three banks' net interest margins were already lower than a year earlier. Citi's case against OCBC rests mainly on deposit costs rising faster than lending rates.
3. DBS's guidance assumed no US rate rises this year. Its August outlook said it was "not expecting US rate hikes this year". The Fed raised rates in September. DBS's margin depends on Singapore rates as much as US ones, so the guidance may still hold. The assumption behind it is out of date, though, and the 5 November update is where DBS will say what that changes.

If you have a home loan, your rate is set by SORA, Singapore's own benchmark, not by US Treasury yields. SORA is the number to watch.

Our read

This section is an AI-assisted view, not financial advice.

For most banks the bond-loss fear looks like sentiment, with little sign of a solvency problem. Held-to-maturity losses only become real if a bank has to sell, and a bank only has to sell if deposits leave. Credit markets show some strain but no panic. US high-yield spreads were about 3.0 percentage points in early October. That is the widest since April, though they eased slightly in the first week of October. Regional banks have fallen in an orderly way. A selling wave driven by a paper loss can reverse once the third-quarter results show deposits holding steady.

The banks fell together, but their risks differ. In this table JPMorgan and Citigroup are the cleanest examples of low bond exposure plus income that rises with rates, and they are the only two still above their 200-day averages. Bank of America is the purest test of whether the fear is overdone: the largest paper loss relative to capital, and net interest income that still rises with rates. These are worked examples of how the numbers sort the group. They are not a list of banks to buy.

What would change this view:

• Third-quarter results showing unrealised losses up sharply and deposits falling at the same bank.
• Regional bank shares (KRE) breaking to new lows while credit spreads widen.
• A bank selling held-to-maturity bonds at a loss to raise cash.
• Large deals and IPOs staying shut into the fourth quarter while high-yield spreads move past their April levels.

For Singapore, margins matter more than the bond losses. The 4 to 6 November updates are the next hard data.

If you sell puts, every US bank in the table reports between 13 and 20 October, so a put sold now carries the results. The levelbox screener updates each name's trend and premium daily. A stock below a falling 200-day average carries the risk covered in our post on value traps, and the S&P 500 breadth post covers how rising yields have shaped past sell-offs.

Method

• Prices: daily closes to 9 October 2026, our own calculation.
• Bank figures: each bank's second-quarter 2026 10-Q filed with the SEC (28 July to 7 August 2026). "% of CET1" is our arithmetic: available-for-sale plus held-to-maturity unrealised losses, before tax, divided by CET1 capital. Net interest income sensitivities are each bank's own disclosure; definitions differ between banks. Results dates are from company announcements.
• Aggregate losses: FDIC Quarterly Banking Profile, second quarter 2026 (released 25 August 2026).
• Yields: US Treasury daily par yield curve; NBC News (7 October 2026) for intraday highs. Fed decision: Federal Reserve statement, 16 September 2026.
• Bank index: The Motley Fool (9 October 2026).
• M&A: Mergermarket / ION Analytics third-quarter and nine-month 2026 reviews (1 October 2026) for North America; Reuters via Zawya, using LSEG data (October 2026), for global figures, megadeal counts and the Morgan Stanley comment.
• IPOs: Renaissance Capital, third-quarter 2026 US IPO market review (1 October 2026).
• Bond issuance and spreads: Bloomberg via Advisor Perspectives (3 September 2026) on investment-grade issuance; the high-yield issuance and spread figures are from a secondary report of Goldman Sachs research (28 September 2026), so treat them as approximate.
• JPMorgan fee guidance: secondary report of comments at the Barclays financial services conference (15 September 2026).
• 2022 comparison: Refinitiv data as reported by the Financial Times, via PYMNTS (29 December 2022).
• Advisory and private equity share prices: daily closes to 9 October 2026, our own calculation.
• Singapore: DBS, OCBC and UOB 2Q/1H 2026 results releases and presentations (6 and 7 August 2026); The Business Times via AsiaOne (7 October 2026) on the Citi note; The Edge Malaysia (8 October 2026) on the JPMorgan note and the 8 October moves.

This article is analytical and educational and is not financial advice. It names banks as worked examples only. Figures are a point-in-time snapshot and can contain errors. This article was generated with AI assistance. If you spot a mistake, tell us.