There is a rule every bank analyst learns in the first month on the job: rising rates are good for banks. They lend at higher yields, they pass only part of the increase on to depositors, the margin widens, profit follows. On Monday, Brian Moynihan, chief executive of Bank of America, stood on the stage of the Barclays Global Financial Services Conference in New York and demonstrated that the rule has a condition the textbook leaves out. Investment banking fees, he said, would come in at 1.6 to 1.8 billion dollars in the third quarter — down 10 to 20 percent from the 2.0 billion the bank earned a year earlier. Sales and trading would be flat on the year. Net interest income was tracking “in line,” but revenue growth would lag expense growth this quarter. The stock fell 5.14 percent to 59.47 dollars on 92.6 million shares, 86 percent above its three-month average volume.
The timing is what turns the number into a story. On the same Monday, the yield on the 10-year Treasury crossed 5 percent for the first time since October 2023, touching 5.014 percent before easing back to around 4.94 by the afternoon. The 2-year yield set a new cycle high near 4.68 percent. Futures markets price a roughly 92 percent chance that the Federal Reserve raises the funds rate by a quarter point on Wednesday, to 3.75–4.00 percent — the first hike since 2023. If the textbook were right, the second-largest bank in the United States should be winning this week. Instead, it was the weakest of the large financials on Monday: Goldman Sachs lost 3.96 percent, Morgan Stanley 3.64, Citigroup 1.90, JPMorgan 1.71, Wells Fargo 1.75. The first bill for 5 percent money is not paid by the borrower. It is paid by the bank that lives off the borrower.
What Moynihan said — and what is inside the number
The wording is unremarkable; the arithmetic is not. “Flat year over year” in trading means roughly 5.4 billion dollars, the figure from the third quarter of 2025. In the second quarter of 2026, reported on July 21, the bank booked 7.2 billion — up 33 percent, the seventeenth consecutive quarter of growth. Flat on the year therefore means down 25 percent on the quarter. Investment banking fees follow the same math: 2.1 billion in the June quarter, up 50 percent; now 1.6 to 1.8 billion, which is 15 to 25 percent less than three months ago. Moynihan chose the year-over-year comparison because it sounds gentler. The market ran the sequential comparison because it shows the direction.
The second sentence that stuck on Monday was a self-assessment: “We’re not as well-positioned in some of the businesses that have more activity.” That is the polite version of a market-share loss. And the third sentence is the most important, because it narrows down the problem: financing and prime brokerage — the loans to hedge funds against securities collateral — had weakened as investors reduced risk, above all in Asia. Anyone who has followed the past two weeks knows the address: the Kospi, SoftBank, the memory-chip makers. The bank that supplied those clients with leverage earns nothing on their deleveraging.
What Moynihan did not say is just as instructive. He did not raise the net interest income outlook. In the second quarter the bank reported 16.2 billion dollars of net interest income, up 9 percent, driven by fixed-rate assets repricing to higher coupons. With the 10-year above 5 percent, that effect should be accelerating. That the bank says only “in line with expectations” means the deposits are getting more expensive at roughly the pace the assets are yielding more. The margin is not widening; it is holding. And a business that holds while costs grow is not a winner.
The June quarter was an exception, not a trend
To understand why the drop looks so steep, one has to see how high the bank was standing in the summer. Net income of 9.1 billion dollars in the second quarter, up 27 percent; 1.21 dollars a share; revenue of 31.6 billion, up 15 percent. None of that came from the lending book. The jump came from the capital markets, and within the capital markets it came from one event above all: the SpaceX initial public offering on June 12. Seventy-five billion dollars raised, the largest IPO in history, priced at 135 dollars and closing its first day at 161. The syndicate’s fees came to roughly 500 million dollars, 0.7 percent of the deal — among the lowest rates ever paid on a mega-listing. Goldman Sachs and Morgan Stanley took about 100 million each; Bank of America, Citigroup and JPMorgan about 75 million apiece.
Seventy-five million dollars is not much against 2.1 billion of quarterly fees. But SpaceX was not alone. The second quarter was the quarter in which the AI industry filled its balance sheets with equity, in which hedging flows filled the trading books, and in which the July burst of volatility — the first time the AI trade cracked — was still booked as client activity rather than client flight. Jefferies has run the numbers across eight major global banks: its investment banking revenue proxy for the current quarter is down 15 percent from a year ago and 27 percent from the second quarter. Dealogic has the overall fee pool on pace for a decline of about 10 percent. Bank of America did not announce a problem of its own. It was merely the first to say out loud the number everyone has.
Why 5 percent hits the fee before the loan
Investment banking sells exactly one product: access to other people’s capital. Every fee — for an IPO, a bond, an acquisition financing, an advisory mandate — is a percentage of a transaction whose existence depends on the price of capital. With a risk-free yield of 5 percent, an IPO has to offer a buyer more than 5 percent of expected return, or the buyer takes the Treasury. A leveraged buyout that penciled at 4 percent does not pencil at 5. A bond deal a treasurer planned in the spring gets postponed because he is hoping for lower coupons. The rate that helps the bank in its loan book kills the transaction it would need in its advisory book.
That it is happening so fast this time is because the rate is coming from two directions at once. Headline inflation ran at 3.4 percent in August; gasoline rose 3.9 percent in the month because Brent sits near 107 dollars after the shutdown of Saudi Arabia’s East-West pipeline. Core inflation, at 2.4 percent, is the lowest since March 2021 — but the Fed is hiking against the headline, not the core. At the same time, the Treasury has stepped up buybacks of its own bonds and urged Japan to limit its sales, because supply is overwhelming the market. A rate hike and a supply glut together: the short end rises because the Fed raises, the long end rises because nobody absorbs the issuance. For a bank that is the worst combination — deposits reprice at the short end, and the bonds in its own portfolio lose value at the long end. Bank of America knows this better than anyone: its held-to-maturity portfolio carried more than 100 billion dollars of unrealized losses at points in 2023. With the 10-year at 5 percent, that ledger is open again.
The pipeline that is not there
Over the weekend, Sam Altman said OpenAI would not go public in 2026. For Wall Street that was not only an AI headline. It was the cancellation of the fourth quarter’s single largest fee event. A listing at a trillion-dollar valuation, even on SpaceX’s 0.7 percent terms, would have delivered several hundred million dollars to the syndicate — on a deal size that alone could have carried the year’s issuance statistics. Goldman Sachs had forecast in the summer that U.S. IPO proceeds could quadruple to 160 billion dollars in 2026. That forecast lived on two names. One is listed. The other is now a date next year.
Corning showed on Monday what equity raising looks like at 5 percent. The glass and optical-fiber maker, a supplier to the AI data-center build-out, announced a 2-billion-dollar at-the-market program — not an IPO, not a marketed placement with a roadshow, but a facility under which Goldman Sachs dribbles shares into the regular tape. The stock fell 13.7 percent, the worst in the S&P 500; Coherent, Lumentum and Fabrinet fell with it. That is the capital markets business in September 2026: it happens, but on terms where the bank barely earns and the issuer pays dearly. An at-the-market program brings the bank a fraction of what a fully marketed deal does — and the dilution the market priced in makes the next issuer more cautious.
The third piece of the pipeline is the quietest. Prime brokerage, which Moynihan named explicitly, is the business that grew fastest in the second quarter and is now shrinking fastest. Hedge funds had leveraged the AI trade — long memory chips, long semiconductor equipment, long SoftBank. Since Seoul and Taipei began to turn a week ago, they have been unwinding. A bank earns on leverage, not on the unwind. And because the unwind happened in Asia, where Bank of America is in any case smaller than Goldman and Morgan Stanley, it hit the bank at its weakest point.
Who shows what across the sector
Monday’s price action is a ranking of dependence. Goldman Sachs and Morgan Stanley lost the most after Bank of America because capital markets make up the largest share of their revenue. JPMorgan lost the least because its advisory business grew 36 percent in the first half and its backlog bridges a downturn for longer. Citigroup fell only 1.9 percent, and for a reason of its own: the bank had shortly before guided to mid-single-digit growth in markets revenue and a low-single-digit increase in investment banking. Two banks, the same conference week, two opposite forecasts. Either Citigroup is taking share from Bank of America — which would fit Moynihan’s self-assessment — or Citigroup has not yet run the number.
Backlog is the word everything hinges on. Goldman said in July that its deal backlog was at its highest level in five years. But a backlog is not revenue; it is a list of transactions planned at a particular price of capital. Every rate hike deletes a line from that list. What the Fed decides on Wednesday therefore determines not only what households pay for credit, but how much of Goldman’s five-year high still exists in October.
What it means for the U.S. bank complex — and for the Europeans
For U.S. investors the read-across runs in three tiers. The first is the money-center group itself. JPMorgan at 350 dollars, Wells Fargo at 88.71, Bank of America at 59.47: all three were bought this year as rate winners, and all three now carry a version of Moynihan’s sentence. JPMorgan has the most diversified fee base and the largest backlog; Wells Fargo has the least capital-markets exposure and therefore the least to lose from a fee recession, but also the most to lose from a deposit-cost squeeze. Bank of America sits in between, with the largest low-cost deposit franchise in the country and the least reason, on paper, to worry — which is exactly why its warning carried so far.
The second tier is the pure capital-markets names. Goldman and Morgan Stanley are where the fee cliff is steepest and where the backlog matters most; Morgan Stanley’s wealth-management arm, like UBS in Switzerland, charges on assets rather than transactions and cushions the fall. The third tier is the regionals and the brokers. The KBW regional index does not depend on IPO fees, but it depends entirely on the two things Moynihan said were merely “in line”: deposit costs and the value of the bond book. The 2023 episode is the template — the 10-year crossed 5 percent then too, and the regional banks, not the money centers, were where the held-to-maturity losses became a solvency question. Charles Schwab, whose bank subsidiary carried the same kind of portfolio, is the name to watch if the 10-year closes above 5 percent rather than merely touching it.
The European mirror is worth a paragraph because it shows the same exposure with less cushion. Deutsche Bank reported its best second quarter ever on July 29 — 2.7 billion euros pretax, 1.9 billion after tax — and it was the investment bank that carried it: pretax profit there rose almost 60 percent to 1.3 billion euros, origination and advisory grew 36 percent to 559 million, fixed income and currencies 16 percent to 2.6 billion. That is precisely the revenue mix Moynihan wrote down on Monday, with a 4.1 percent share of the European market that offers less buffer than a U.S. leader’s. Barclays itself, the conference host, has lost 8 percent in a month as European houses keep ceding equity market share to the Americans. The European Central Bank raised rates on September 10 for the second time this year, to a 2.50 percent deposit rate from September 16; the German 10-year yield is at its highest since 2009, and the French-German spread widened to 98 basis points intraday on Monday, against a highest close of 88 since the sovereign debt crisis. Bank bond books are under water on both sides of the Atlantic.
For a U.S. taxable account the housekeeping is simple: Bank of America yields 1.9 percent at 59 dollars on a price-earnings ratio of about 13.6, the dividend is qualified and the payout is not in question. The question is the reason for owning it. Anyone who bought a money-center bank as a rate winner owns a business whose margin, by its own account, is not widening — and whose growth engine is shrinking.
The counter-argument
One can call Monday’s reaction excessive, and the numbers support that. Investment banking fees are roughly 6 percent of Bank of America’s revenue. A 300-million-dollar shortfall against a consensus near 2 billion equals 1 percent of quarterly revenue of 31 billion. One percent of revenue cost about 20 billion dollars of market value — more than sixty times the shortfall, on a market capitalization of 416 billion. The market did not price the number; it priced the signal that the quarter in which the Fed hikes is not a good quarter for the bank.
Second, Moynihan was explicitly upbeat on the economy: “We feel very good about the underlying U.S. economy”; loans and deposits were growing as planned. A bank whose loan book grows and whose charge-offs do not rise has no balance-sheet problem, only a revenue-mix problem. Third, Wednesday’s hike may be the only one: core inflation is falling, the headline is rising because of oil, and if the Saudi pipeline comes back the arithmetic flips. A one-and-done would preserve the backlog that a cycle would destroy. And fourth, Citigroup’s opposite guidance may simply be right — in which case Monday is a story about Bank of America, not about the sector.
The strongest counter-argument is the comparison with 2023. The 10-year crossed 5 percent then as well, investment banking was on the floor, and Bank of America still earned — because the net interest margin carried the rest. The difference today: in 2023 deposits were still cheap and the securities book had already taken its losses. Today the deposits have become expensive, and the losses in the bond book are being created anew.
What to watch now
The first date is this afternoon, when Doug Petno, JPMorgan’s co-president and head of the commercial and investment bank, speaks at the same conference at 2:45 p.m. Eastern. If he confirms the decline, Bank of America’s number is a sector number; if he does not, it is a share loss. The second date is Wednesday: if the Fed hikes and signals more, it deletes the backlog; if it hikes and signals calm, it preserves it. The third marker is the 10-year yield — a close above 5 percent, not just an intraday high, would reprice every bank’s bond book. And the fourth is third-quarter earnings in mid-October, which will show for the first time how much of what was called growth in June is still there in September.
Bank of America did not say on Monday that its business is going badly. It said that the business which drove profit up 27 percent in the summer will shrink by a quarter in the fall, while the business that was supposed to benefit from higher rates is merely keeping pace. The textbook is right: rising rates are good for banks. But only for the ones that sell loans — not for the ones that broker capital. And the second-largest bank in America has made a great deal of money over the past three years being the second kind.
Try TradingView Free for 30 Days
Plus get a $15 discount on your first subscription through this link.
