Connect with us

BUSINESS

Treasury Yields Post Biggest Drop Since the 2023 Bank Crisis

The 10-year’s sharpest drop since SVB came from the same AI spending boom that had just driven yields to a 24-year high.

Published

on

The 10-year Treasury yield fell 19.3 basis points to 4.952% on Thursday, its largest one-day drop since the March 2023 bank crisis. Investors bought the notes as the Nasdaq sank 2.6% and Brent crude jumped to $113.17 a barrel.

A basis point is 0.01 percentage point. Bond prices rise when yields fall, and Thursday’s bid was large enough that Dow Jones Market Data counted only 15 such 19-point days for the 10-year in the past decade, a list that already includes the Covid shock, the regional-bank run, the April 2025 tariff scare, and the start of the Iran war in late February.

The 10-Year Just Posted Its Sharpest Drop Since SVB

The cash 10-year, the note traders actually buy, closed at 4.952% on the Dow Jones print, below 5% for the first time since September 23. The policy-sensitive 2-year yield dropped 18.8 basis points to 4.635%, its biggest daily decline since August 1, 2025, the session after President Donald Trump fired the head of the Bureau of Labor Statistics. The 30-year yield was more than 14 basis points lower at 5.296%.

The government’s own close is a shade different because it is a fitted par curve, not the on-the-run note. Treasury’s daily par yield curve put the 10-year at 4.96% on October 22, down 17 basis points from 5.13% on October 21. The 2-year par yield fell 16 basis points to 4.65%, the 5-year 17 basis points to 4.78%, and the 30-year 13 basis points to 5.29%. The 2-year/10-year spread finished at +31 basis points, still upward-sloping.

THURSDAY VERSUS WEDNESDAY, PAR CURVE

Maturity Oct. 21 Oct. 22 Change
2-year 4.81% 4.65% -16 bp
5-year 4.95% 4.78% -17 bp
10-year 5.13% 4.96% -17 bp
30-year 5.42% 5.29% -13 bp
2s10s spread +32 bp +31 bp -1 bp

The belly of the curve, the 5-year and 7-year, fell as hard as the 10-year, while the 3-month bill barely moved. That is a reprice of growth and the Fed path, not a dash into overnight cash. Robert Tipp, head of global bonds at PGIM Credit, said the size of the session still stood out against ordinary trading.

On a normal day, a decent-size move is a few basis points. You are getting three times that.

Robert Tipp, head of global bonds, PGIM Credit

THE TWO WEEKS THAT BROKE THE 10-YEAR

  1. October 7: The cash 10-year trades as high as 5.3645%, the highest since April 2002, after inflation and energy fears, and FOMC minutes say most officials still see another hike as likely this year.
  2. October 21: Meta reports negative free cash flow and lifts the floor of its 2026 capital-spending plan, then says it will create a non-voting Class C stock.
  3. October 22: The 10-year falls 19.3 basis points; the Nasdaq drops 2.6%; Brent settles at $113.17; weekly jobless claims print at 194,000.
  4. October 27-28: The Federal Open Market Committee meets with the funds rate in a 3.75% to 4.00% range.

From that October 7 cash high to Thursday’s 4.952% close, the 10-year gave back 41.25 basis points in 15 days. The 10-year constant-maturity yield series had still been at 5.13% as of October 21, so almost all of the official decline landed in a single session.

The Same Spending Boom That Lifted Yields

For months the AI buildout had been a reason to sell Treasurys. Data-center power, heavy corporate issuance, and sticky inflation helped push long yields to those two-decade highs earlier in October. Thursday flipped the sign on the same trade. Meta Platforms, the first of the big AI spenders to report this season, beat on revenue and still saw its shares drop 7%.

Meta’s third-quarter earnings release showed why the bond market flinched. Revenue was $64.60 billion, up 26%. Cash flow from operations was $32.79 billion. Capital spending, including principal payments on finance leases, was $34.91 billion, and free cash flow was negative $2.12 billion. Full-year 2026 capital spending was narrowed to $135 billion to $145 billion from a prior $130 billion to $145 billion range, which raised the floor by $5 billion. The board also approved a new non-voting Class C common stock, due in January 2027, as another way to fund the buildout. The company still held $82.45 billion in cash and marketable securities at quarter-end.

WHAT META TOLD THE BOND MARKET

  • Cash burn: Free cash flow swung to a $2.12 billion outflow from a $10.63 billion inflow a year earlier.
  • The spend: Third-quarter capital outlays of $34.91 billion ran hotter than operating cash flow of $32.79 billion.
  • The raise: A new Class C share class gives Meta stock it can issue without handing out votes.
  • The floor: The low end of 2026 capital spending moved up by $5 billion, to $135 billion.

Mitch Goldberg, president of ClientFirst Strategy, called the Treasury bid a flight to safety and said buyers were suddenly wary of how much debt AI-related companies may need to sell. Nvidia chief executive Jensen Huang said demand for specialized AI cloud providers, the so-called neoclouds, was “off the charts,” then added that capital was scarce. An index of semiconductor stocks fell 6.7%, its largest daily drop since June 23.

Dustin Reid, chief fixed-income strategist at Mackenzie Investments, said the tape had moved past a garden-variety wobble. “We are clearly in a theme around concerns of an AI overextension,” he said, with investors taking chips off the table this late in the year and rotating into the $32 trillion Treasury market. FactSet still had Sandisk up about 430% for 2026 and Micron up 200%, which is a lot of gain sitting in taxable accounts. Reid also pointed to short covering in Treasurys after a brutal backup in yields, a mechanical bid that can look like panic when it is partly a squeeze.

Charlie McElligott, managing director for cross-asset strategy at Nomura Securities International, wrote that U.S. Treasury bonds are working as a hedge again as the market reprices the odds of a U.S. growth scare off the AI reset. That is the irony in one line. The boom that made duration toxic in early October is the boom that made duration the shock absorber on October 22.

Oil Climbed Past $113 and Bonds Still Rallied

A clean inflation scare would have sent yields higher with crude. Thursday did the opposite. West Texas Intermediate futures rose 2.39% to $99.47 a barrel. Brent, the global benchmark, jumped 4% to $113.17 after reports that the United States and Israel were weighing a short, intensive round of strikes on Iran after the November 3 midterms.

Gregory Faranello, head of U.S. macro strategy at South Street Securities, said there was no single risk point, then put geopolitics first anyway. He also warned that if energy prices stay elevated, the consumer gets pinched and the Fed may have to do more, “which certainly wouldn’t be a good thing for the bond market.” The session ignored that second step. Treasurys were bought as oil rose, which is what traders do when they start to care more about weaker demand than about the next inflation print.

France offered a control case. Ten-year French OAT yields jumped 7.6 basis points to 5.077%, levels not seen since 2002, as a fiscal fight and street protests kept selling pressure on Paris. U.S. notes rallied in the same hours. Larry Holzenthaler, senior portfolio manager for fixed income at Catalyst Funds, said that for the first time in a while, Treasurys were enjoying haven status on a relative basis: weak tech, a hotter Middle East, and higher oil, and the U.S. curve still looked like the place to hide.

It was not a blanket dash into every so-called safe asset. Gold and silver sold off, and bitcoin slipped even as the 10-year broke back under 5%. Most U.S. stocks actually rose; the damage was concentrated in the AI complex while utilities, staples, and the equal-weight S&P held up. Sam Stovall, chief investment strategist at CFRA Research, noted that tech is about 40% of the S&P 500 by market value and asked whether this was “the big one” in a market selloff. Defensive shares and broad funds that are not stuffed with megacap AI names were, he said, holding up well.

March 2023 Hit the 2-Year Much Harder

The comparison everyone reached for is Silicon Valley Bank. It is the right date and the wrong mechanism. On March 10, 2023, the 10-year yield fell about 23 basis points, from 3.923% to 3.689%, as depositors ran and the Fed’s next hike suddenly looked optional. In the days after, the 2-year yield posted drops of 39 basis points and more, the largest since 2008, and the 2-year/10-year curve was deeply inverted.

Thursday’s 10-year move was the largest since that week, and it was still smaller than March 10 itself. The 2-year and 10-year fell almost in lockstep. The curve stayed positively sloped at +31 basis points. Regional banks were not the story; several of them even caught a bid from lower yields. Tipp’s stressors were a stack of growth and political risks hitting a market that had been positioned for more Fed hikes and higher long rates, not a hole in the deposit base.

HOW THE TWO FLIGHTS DIFFERED

  • The trigger: March 2023 was a bank-funding run. October 22 was an AI-spending scare layered on oil and French fiscal stress.
  • The curve: 2023 inverted. Thursday’s 2s10s spread stayed at +31 basis points.
  • The 2-year: The 2023 panic was a front-end event measured in tens of basis points per day. Thursday’s 2-year drop of 18.8 basis points matched the 10-year rather than dwarfing it.
  • The labor tape: Initial jobless claims printed at 194,000, below a 200,000 forecast and the lowest since July, which is not how a classic recession scare usually arrives.

Tipp called the session a take-back trade, not an all-out fear-of-recession episode. That reading fits the claims number and the still-steep cash 30-year above 5.2%. It also fits a market that had, in his words, been caught off-guard in its positioning for several more rate increases.

What Thursday’s Collapse Means for Mortgages

Homebuyers will feel this only if the 10-year stays down. Freddie Mac’s Primary Mortgage Market Survey, published Thursday for the week ending October 22, put the average 30-year fixed rate at 7.32%, down from 7.40% the prior week. That was the first weekly decline since late August and it snapped a seven-week climb. A year earlier the same average was 6.19%.

Sam Khater, Freddie Mac’s chief economist, said the decline offered some relief to prospective buyers. The 15-year fixed rate eased to 6.68% from 6.71%. The survey averages quotes across the week, so it only partly captures Thursday’s 19.3-basis-point cash crash. If that close holds into next week’s sampling, posted mortgage rates have more room to follow the 10-year lower. They will not close the gap with last year. From the brief 5.98% low in late February, the seven-week run-up to 7.40% had added about $355 a month to the payment on a $400,000 loan.

Auto loans and credit-card pricing also reference this part of the curve, with a lag. A one-day collapse does not reprice a household’s existing loan. It does change what a new 30-year quote looks like if lenders believe the 10-year is done making 24-year highs.

The 2-Year Is Already Voting Against Another Hike

The Federal Reserve’s target range is 3.75% to 4.00% after a quarter-point increase in September, the first hike since 2023. The next meeting is October 27-28. CME FedWatch had put the odds of another 25-basis-point move at that meeting at 14.9% as of October 22, down from 33.2% a week earlier and 57.6% a month earlier. December had still been treated as live. Thursday’s 18.8-basis-point drop in the 2-year is the bond market’s way of asking whether even that later hike survives an AI growth scare.

The Fed’s H.15 selected rates will print the official constant-maturity close with a one-day lag; the October 21 10-year in that series was 5.13%. Minutes released on October 7 still had most participants judging that another increase would likely be appropriate by year-end. Tipp’s point is the collision: a market that had priced a few more hikes is now being forced to ask whether the central bank can deliver them if the stock market’s AI engine seizes up or if diesel and crude keep squeezing the consumer.

Faranello’s oil warning still sits on the other side of that trade. A 10-year at 4.952% with Brent at $113.17 is an unstable pair. Either crude comes in, and the growth-scare bid in notes can last, or energy stays high enough that inflation expectations wake back up and the 24-year yield high starts to look like a pause rather than a peak. Tipp’s own verdict on Thursday was narrower than a recession call, and it is the one that matches the curve.

It’s a take-back trade. Not an all-out fear-of-recession type of situation.

Robert Tipp, head of global bonds, PGIM Credit

The 10-year is back under 5% because the spending boom that made Treasurys look dangerous made equities look expensive in the same week. The Fed sits on October 27 with hike odds already in the teens and a 2-year yield that just had its largest drop since a political shock in August. Oil, Meta’s cash flow, and that 14.9% probability now have to share the same room.

Disclaimer: This article is news reporting and market analysis for information only. It is not investment advice, a recommendation to buy or sell Treasurys, mortgage products, equities, or any other security, and it is not a forecast of Federal Reserve policy. Readers who are considering changes to a bond portfolio, a home loan, or an equity position should consult a licensed financial adviser or mortgage professional who can review their own facts. Yields, mortgage averages, index closes, and policy odds are taken from the sources named in this piece as of the dates given and can move in the next session.

Harry is the editor of BUDGY APP, an independent title he owns and runs after ten years in journalism that began on a reporter's desk and ended up at the editor's. Numbers get particular attention here. A percentage in a business story is recomputed from the underlying figures before it goes live, a benchmark in a technology or gaming review is quoted with the conditions it was measured under, and a transfer fee or a lap time in the sports and auto pages is traced back to the club, the league or the timing sheet that published it. The same rule covers news, science, entertainment, lifestyle and travel: if a figure cannot be tied to a filing, a dataset, a transcript or a test Harry ran himself, it does not appear. Readers around the world see prices in the original currency with a conversion alongside. Errors are corrected in the open under a published corrections policy, with the change noted on the article. Questions about any figure reach him at support@budgyapp.com.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending