Newsletter · 2026-09-24 · Ashutosh Agarwal

Five Percent Is the Floor Now and the Bond Market Cannot Agree Why - The Dollar Brief - Week of September 24, 2026

The Dollar Brief for the week of September 24, 2026 (podcasts recorded roughly September 17 to 24): the 10-year Treasury broke and held above 5% and the market split on whether that signals a boom or a buyers’ strike, the 2s10s curve flattened toward inversion, the long bond lost its natural owners, Bessent’s stablecoin plan was laid out as a three-step scheme, and the desks argued whether the hawkish hike unshackled the dollar or set up a high-conviction short.

The Dollar Brief

Week of September 24, 2026: Five Percent Is the Floor Now and the Bond Market Cannot Agree Why

Here is the one line that captures the whole week's argument. This week on The Julia La Roche Show, veteran stock-picker George Noble did a little role-play about lending money to the US government. "I have income of 5 trillion. I'm spending seven, seven and a half trillion. I owe the banks 40 trillion… and inflation is three and sticky." So when someone offers you 5% to lend to that borrower for ten years, Noble says, the honest reaction is: "wait a second, you're not terribly credit worthy… Why would you lend me the money? If you were a bank, you wouldn't."

That is the question hanging over the dollar this week. The 10-year Treasury, the single most important interest rate in the world, has now punched through 5% and stayed there, a level not sustained in nearly two decades. Last week the debate was whether the government's bond buybacks could stop the bleeding (the podcasts said no). This week the debate moved somewhere more interesting and more unsettled: nobody can agree on what a sticky 5% actually means. One camp hears a boom, a strong economy pulling in the world's capital, and a dollar that just got "unshackled." The other hears a buyers' strike, the natural owners of long-term US debt quietly walking away, leaving the Treasury to manufacture new lenders out of thin air. And in a genuine plot twist, one of the most-watched recession signals just flashed for the first time in a year and a half. Let's get into it.

(Quick plain-English glossary, skip it if you live in this stuff. The "dollar" is the US dollar; the "DXY" measures it against a basket of other rich-world currencies. A "Treasury" is a loan to the US government; its interest rate is its "yield," and yields rise when bond prices fall. "T-bills" are the short-term version (under a year); the "10-year" and "30-year" set mortgage and corporate borrowing costs. A "real" yield is the yield after subtracting inflation. The "yield curve" plots short-term against long-term yields. When long rates rise faster than short, the curve "steepens"; when the gap shrinks it "flattens"; when short rates top long rates it "inverts", historically a recession warning. The "2s10s" is the gap between the 2-year and 10-year yields. "Term premium" is the extra yield lenders demand for locking money up for years. "Convenience yield" and "exorbitant privilege" describe the discount the US has enjoyed because everyone wants to hold its debt. A "buyback" is the Treasury purchasing its own bonds. The "basis trade" is a heavily-borrowed hedge-fund strategy that's become a big Treasury buyer. The "carry trade" is borrowing cheaply in a low-rate currency, for years, the yen, to buy higher-yielding things elsewhere. A "stablecoin" is a digital dollar token backed by real dollar assets. "De-dollarization" is the world using and saving in the dollar less; "re-dollarization" is the opposite. "Fed independence" means the central bank sets rates free of political orders.)

TL;DR

The 10-year broke and held above 5%, and the market can't agree if that's strength or stress. George Noble says a government with $5T of income and $7.5T of spending simply "isn't terribly credit worthy" at that price (The Julia La Roche Show, Sep 22). A Reuters host countered that 5% is "genuinely one of those glass half full, glass half empty" phenomena (Reuters Econ World, Sep 23).
The plot twist: the yield curve just flattened, not steepened. Jeff Snider says the 2s10s spread has collapsed to about 20 basis points, the narrowest in ~18 months, because the long end fell, not because the short end rose. "The front of the curve sees the Fed. The back of the curve, that sees demand" (Eurodollar University, Sep 23). That is the mirror image of last week's "steepening = inflation alarm" story.
The long bond has lost its natural owners. Insurance companies, pensions and sovereign wealth funds, "the natural buyer of the 30 year", have defected to private credit and corporate bonds. They "loved the 10 year when it was at 1.5%," and won't touch it at 4.7–4.8% (Monetary Matters, Sep 19).
Bessent's stablecoin plan got spelled out as a three-step scheme. Buy back long bonds, shift borrowing to T-bills the Fed controls, then use stablecoins, required by law to hold T-bills, as "a giant natural buyer" (Daily Crypto News, Sep 23). (Crypto show relaying third-party research; weigh accordingly.)
JPM's FX desk says the hawkish hike "unshackles the dollar." Removing the July credibility discount gives the dollar "a valuation tailwind," and the first Fed hike "doesn't necessarily have to mean the peak in the dollar" (At Any Rate, Sep 18).
David Rosenberg is on the other side, 70% short the dollar. His model portfolio "is 70% inversely correlated to the U.S. dollar," a bet on reserve managers rotating from dollars into gold, and on Washington wanting a weaker currency (Excess Returns, Sep 18).
A former Fed president says the market is pricing too many hikes. Robert Kaplan called September correct but the dot plot "muted", one more hike, then nothing in 2027, and says the market is adding "a risk premium" of extra hikes it probably won't get (Goldman Sachs Exchanges, Sep 23).
Harley Bassman wanted a bigger, cleaner hike. The fixed-income veteran argued the Fed should have gone "50. One and done", because this is about trust, not the number, and drip-feeding guidance is like quitting heroin slowly (Macro Voices, Sep 17).
The dollar direction is now a pair-by-pair call. BNY's Jeff Yu: the dollar strengthens against "these fiscally challenged colonies" (the euro and sterling) but "needs to adjust against APAC names", dollar-China is already falling (Bloomberg Surveillance, Sep 22).

What's new

5% is the floor now, and the fight is over why

For two years the story was "will the 10-year hit 5%." This week it hit it, held it, and turned a scary headline into the new baseline. What's genuinely new is that the smartest people in the room have split cleanly into a boom camp and a bust camp, looking at the exact same number.

George Noble is squarely in the bust camp. "The 10-year today, no great surprise, through 5%, after Warsh did his thing. And that by itself puts pressure on markets," he said. His lending role-play, income of $5 trillion, spending $7–7.5 trillion, $40 trillion of debt, sticky 3% inflation, leads him to a blunt conclusion: at 5% the government "is not terribly credit worthy," and it isn't even offering a good enough real return to make up for it. He's not calling the crash yet; his line is "R for rotation, not recession", "passionately bearish on consumer stocks," but "wildly bullish on gold and energy stocks," which he says are up 30–50% this year (The Julia La Roche Show, Sep 22).

The more measured version came from a Reuters markets podcast built entirely around the 5% level. The host asked whether the US, which "prints its own money" and runs "the biggest economy in the world," can simply get away with it. The answer was: for a while, yes, because "other markets just are not big enough" to absorb the money. But there are cracks. "There's this phenomenon called the convenience yield," the discount the US enjoys because its debt is so easy to trade, "and there's a decent amount of evidence that that's actually been diminished." Add a rising term premium, and "the U.S. is losing a little bit of that exorbitant privilege, but it's not likely to simply go away overnight." The show also flagged the sheer weight of the interest bill: it "now exceeds a trillion dollars a year… probably $1.2 trillion this year," the government's second-largest spending category after entitlements (Reuters Econ World, Sep 23). Its wry aside, on why bond markets terrify politicians, was worth the listen: strategist James Carville once said he'd like to be reincarnated as the bond market, "because you can intimidate everybody."

Crucially, the same podcast laid out the boom case fairly. New Fed chair Kevin Warsh pinned the rise in long-term yields on three things: the strength of the US economy, competition for capital from the enormous wave of AI data-center borrowing, and geopolitics. Nobody disagreed with those three, "but they felt he left at least one big one off the table, which is debt." The US Treasury market has ballooned from "less than $20 trillion before COVID" to $32 trillion today. So the honest read on 5% is exactly the glass-half-full/half-empty answer: partly a genuinely constructive growth story, partly a warning about supply and trust.

The plot twist: the curve just flattened, not steepened

Here is the freshest, and most contrarian, development of the week, because it flips last week's consensus on its head. Last week the vivid image was a steepening curve (long rates rising fastest) read as an "inflation and debt alarm." This week Jeff Snider pointed at the opposite.

"One of its most important warning signals, the spread between the two-year and 10-year Treasury yields, just flattened to only 20 basis points," he said. "That's the narrowest spread in roughly a year and a half. Inversion is now within the market's grasp." And the way it happened is the whole point. For most of this oil shock the curve flattened because short rates were climbing, the 2-year jumped "from about 4.2% to as high as 4.75%… in only a couple of weeks" as traders bet the Fed would keep hiking. "That was not an inflation forecast. It was a Fed forecast." But now, he says, the long end is moving the other way: "the two-year yield barely moved yesterday while the 10-year yield actually fell." His framing: "The front of the curve sees the Fed. The back of the curve, that sees demand." Higher oil raises inflation now but acts like a tax that destroys demand later, and the long end may be starting to price the second thing. His startling conclusion: "the next big move in interest rates may not actually be higher" (Eurodollar University, Sep 23).

He backed it with corroborating evidence rather than just the one signal. PIMCO "has begun trimming its underweight position in long-term U.S. government bonds as yields above 5% create… better value." The riskiest corner of the credit market is cracking, "triple-C rated credit spreads are approaching 1,100 basis points." And there's a strange standoff in corporate bonds: investors are desperate for long-dated paper (Aon's $2 billion 30-year deal drew ~$10 billion of orders; GSK's $500 million drew roughly ten times the supply), yet companies refuse to issue it, "only about 5% of U.S. investment-grade bonds sold during the first half of September had maturities of at least 30 years, the smallest proportion… since at least 2020." Both sides are betting the same thing: today's high rates won't last. (Snider leans structurally dovish/contrarian, but the 20bp spread is a fact, not a forecast.) Even David Rosenberg, coming from a very different place, clocked the same move: "two's tens is down to 35 basis points… I am on recession watch" (Excess Returns, Sep 18).

The buyers' strike: who's left to hold the long bond?

Underneath the yield drama is a quieter, more structural story that got its clearest telling this week: the people who used to reliably buy 30-year Treasuries have walked away, and no one obvious has replaced them.

David Busch laid out the plumbing. "The natural buyer of the 30 year is insurance companies, public pension plans, sovereign wealth funds… beyond that there's not a lot of natural buyers out there." The problem is those buyers now have better options. "There's competition, in the form of private credit funds, as well as investment-grade corporate issuance," and only so much capital to go around. Host Jack Farley put the paradox perfectly: these institutions "loved the 10 year when it was at 1.5%. And now they don't seem to like it now that it's at 4.7% or 4.8%." Why? Insurers used to buy the 30-year for its long duration to match their long liabilities; now they'd rather earn more in floating-rate private credit, trading interest-rate risk for credit risk. Busch's caution: before anyone celebrates those higher private-credit yields, "do you have access to a loan tape? Do you know who they've lent to?" His bottom line is that it will take "a catalyst that forces investors to seek the shelter of a U.S. treasury" to bring the big buyers back (Monetary Matters, Sep 19).

That's the same thread that runs through last week's point about foreigners now holding a minority of Treasuries. Insurers gone, pensions gone, foreigners retreating, the long bond is losing its patient, price-insensitive owners just as the government needs to sell more of it than ever.

The manufactured buyer: Bessent's three-step plan, spelled out

So if the natural buyers are leaving, where does the new demand come from? This week a crypto podcast relayed a striking framework, from an outside research shop, Bravo Research, that ties together three things we've been tracking separately into one coherent Treasury Secretary game plan.

"Step one is buy back long-term treasuries," to lean less on long-term investors who keep demanding higher yields, the buybacks the Treasury doubled to $4 billion a round. "Step two is replace long-term borrowing with short-term bills," because short rates are "much more closely tied to Federal Reserve policy," giving the government "a financing market that is less dependent on foreign investors." And step three: "use stablecoin growth to create a giant natural buyer of T-bills." Because the GENIUS law "requires payment stablecoins to be backed one-to-one by highly liquid reserves, including dollars and short-term Treasuries," every new stablecoin dollar mechanically buys T-bills. The host noted stablecoin issuers are already "in the top 20" holders of US debt and floated projections of "a trillion dollars held by these stablecoin companies" by 2030 (Daily Crypto News, Sep 23). (This is an openly pro-crypto show, it explicitly pitched buying Circle stock in the same breath, so treat the enthusiasm as a sales pitch, not a forecast.) A more dollar-friendly spin came from another crypto venue, which framed stablecoins as *re-*dollarization, rerouting the world's offshore dollar system onto stablecoin rails and spreading the dollar further, not weakening it (Bitcoin Magazine Podcast, Sep 23).

The mechanics of step one are cleverer than they look. Busch explained that the Treasury can buy back old, low-price bonds "at like 60 to 70 cents on the dollar" and reissue at par, which "incrementally reduces the total notional outstanding", a headline cut to the debt number even though "the interest cost doesn't necessarily change." His dry read on the whole exercise: "it's all headline risk," and remember the Secretary's toolkit is thin, "the only tools that he really has is treasury issuance and buybacks."

The dollar itself: unshackled, or a high-conviction short?

For all the bond drama, the dollar had a firm week and sits near 100 on the DXY, and this week produced three sharply different views on where it goes next.

The bull case came from JPMorgan's FX strategists, and it's a clean one. They argued the "dollar discount", a risk premium the currency carried ever since the July Fed meeting raised doubts about the new chair's resolve, widened after July and has now been erased by this week's hawkish hike. "This perhaps unshackles the dollar a little bit in terms of getting that valuation tailwind," one said, adding that the Fed "shouldn't be an obvious drag to the dollar from here." History helps them: "the first Fed hike doesn't necessarily have to mean the peak in the dollar." Their message to nervous clients was literally "keep calm and carry on," and their warning to the bears was sharp: "humility is in order when we look at curves… saying that too much is in the price," because all year the market has been wrong about that (At Any Rate, Sep 18). (This is a sell-side desk that has been bullish the dollar, it's talking its call.)

David Rosenberg is the mirror image, and he's put real money behind it. "We are fundamentally bearish on the U.S. dollar," he said, the model portfolio he now runs as an ETF "is 70% inversely correlated to the U.S. dollar." His reasoning: central banks are steadily rotating reserves out of dollars and into gold (gold's share of reserves has climbed from about 10% in the late 1990s to roughly 25%; he'd trim only at "35 to 40"), and there's still a yawning gap between the dollar's ~40%+ share of global reserves and America's ~20% share of global trade, a gap "in the process of mean reverting." On top of that, he thinks Washington wants a weak dollar: "a weak U.S. dollar fits in very well" with the mercantilist, bring-factories-home agenda (Excess Returns, Sep 18). (Rosenberg runs a fund built on this view, weigh accordingly.)

The most practical framing split the difference. BNY's Jeff Yu refused the single-DXY call: "I always say against what?" The dollar "will strengthen against the euro… against sterling, let's just say these fiscally challenged colonies. But the dollar needs to adjust against APAC names. Dollar-CNY is falling," because that's where America's new AI-related exposures sit (Bloomberg Surveillance, Sep 22). In other words: firm against Europe, softer against Asia.

The Fed: is the market pricing too many hikes?

The dollar is firm largely because Warsh raised rates and signaled more. But this week a credible, calm voice argued the market has gotten ahead of itself.

Former Dallas Fed President Robert Kaplan, now at Goldman, said September was the right move, month-over-month inflation was running "closer to 3," not the hoped-for 2.5%. But he stressed how restrained the Fed's own forecast was: "the dot plot… was pretty muted," a median of "one more increase" and then "no action in 2027." Why so mild? Because the booming parts of the economy, AI, defense, don't respond to interest rates, while "the interest-sensitive parts… are already not overheated." His clearest line: "the one tool I have, the Fed funds rate… is not going to slow down the AI infrastructure build." So the market pricing extra hikes is really "putting in, in effect, a risk premium", partly because it doesn't yet understand Warsh's reaction function, partly for oil risk. He also noted the hike "didn't really have much of an effect on the curve because the curve had already adjusted." And on the independence question that dominated recent weeks, he was unequivocal: "I'm extremely confident that the folks around the FOMC table are doing their level best without political considerations" (Goldman Sachs Exchanges, Sep 23).

Harley Bassman came at it from the opposite direction, he wanted more, done faster. The fixed-income veteran (he invented the MOVE index) said the whole point is trust, not the size of the move: "the Fed needs to regain the trust of the market." He'd been "advocating strongly for the Fed to hike 50. One and done, get it done," and thinks the drip-feed of guidance is counterproductive, weaning markets off forward guidance "is like coming off… heroin," and "cold turkey would have been the better idea." He also put his finger on why the Fed may be losing control of long rates regardless: hyperscaler AI borrowing now rivals Treasury issuance, "we're getting close to 30, 35% competition for money", and "the Fed can't stop this… these guys are going to borrow money at any rate. Six, eight, 10, 12. It doesn't matter." His one hopeful note: as long as Fed chair Warsh and Treasury Secretary Bessent "work in a symbiotic manner," markets will believe there's "an adult in the room", and he thinks that's happening (Macro Voices, Sep 17).

The carry-trade plumbing: Japan's dovish hike and Bessent's yen game

One under-covered but important thread this week was Japan, because the yen is where US rate policy, the carry trade and Treasury demand all collide.

The Bank of Japan raised rates to 1.25%, but the vote was a dovish 7-2, with two board members appointed by the reflation-friendly Takaichi government dissenting. The result: the yen weakened despite the hike, and an October follow-up is now "kind of out of the question," pushing the next move to December. The deeper point is why the US cares. Treasury Secretary Bessent bought yen back in July, and Bloomberg's Paul Jackson explained the real motive: if Japan has to prop up the yen by selling its huge pile of US Treasuries (Japan is the largest foreign holder), that pushes Treasury yields up, "exactly what Scott Bessent objects to." So Bessent's preferred fix is for the BoJ to keep hiking, propping the yen up on its own. "The US now has some kind of skin in the game," Jackson said, having intervened jointly, Washington won't let the yen spiral (Big Take Asia, Sep 18). It's a reminder that America's borrowing costs now partly depend on a central bank in Tokyo.

The midterms are a rates story, but maybe a smaller one than feared

With the November 3 midterms six weeks out, this week brought a useful counter to the "election will blow out the deficit" worry. JPMorgan Asset Management's strategists laid out the likely map, a probable split Congress, with Republicans holding the Senate and Democrats potentially taking the House, and argued the fiscal impact is limited either way: "in either outcome, you probably have limited room… for deficit expansion over the next two years." The reason is the very thing this whole issue is about, the deficit already sits at a "war time" 6–7% of GDP "with an expansionary economy," and now that financing is expensive, "debt service is expensive," so the bond market has already priced the risk (Insights Now, Sep 17). That's a gentler read than last week's "the election puts a hard floor under long rates." The louder, angrier version came from George Noble, reacting to the administration floating a $5,000 check "for every man, woman and child": "Are you serious? That is emerging market banana republic type of behavior" (The Julia La Roche Show, Sep 22).

The debate

Where the podcasts genuinely disagreed this week.

Is 5% strength, or stress? This week's central split. Boom camp: Warsh's three drivers (growth, capital competition, geopolitics), JPMorgan's FX desk continually re-rating US growth higher, and the "5% is historically normal" crowd. Bust camp: Noble ("not terribly credit worthy"), Snider (demand destruction, near-inversion) and Rosenberg ("recession watch," real rates "way too high"). The Reuters host's honest verdict, "glass half full, glass half empty", is where the evidence actually sits (Reuters Econ World, Sep 23). Referee: the next few months of growth and inflation data.

Is the curve steepening or flattening, and what does it mean? Last week's near-consensus read the long-end move as steepening = an inflation-and-debt alarm. This week Snider says the 2s10s has flattened to ~20bp because the long end is falling, a growth scare, with the next rate move possibly down (Eurodollar University, Sep 23). Rosenberg has 2s10s at 35bp and is "on recession watch" (Excess Returns, Sep 18). Referee: whether the 2s10s inverts and stays inverted.

Is the market pricing too many hikes? Kaplan says yes, the dot plot points to one more and nothing in 2027; the rest is "a risk premium" (Goldman Sachs Exchanges, Sep 23). Bassman effectively says the Fed under-delivered, should have done 50 and been done (Macro Voices, Sep 17). Referee: the next inflation prints, and whether October is a skip (Kaplan's base case, and pre-election).

Dollar: unshackled, or a high-conviction short? JPMorgan's desk says the hawkish hike removes the risk premium and gives the dollar a tailwind (At Any Rate, Sep 18). Rosenberg is 70% short it on reserve-diversification and deliberate policy (Excess Returns, Sep 18). BNY's Yu splits it: firm vs Europe, softer vs Asia (Bloomberg Surveillance, Sep 22). Referee: whether July's risk premium stays gone, and how the US–China/APAC adjustment plays out.

The trades in play

These are speakers' own stated positions and views, not advice. Treat openly-interested sources accordingly.

Short the dollar; own gold; buy the bonds nobody wants. David Rosenberg's portfolio is "70% inversely correlated to the U.S. dollar," long gold as a bet on central-bank reserve flows, and he says his next move is to "rotate further into treasuries at a time when nobody wants them. They're for sale. I want to buy what's for sale" (Excess Returns, Sep 18). (Runs an ETF built on this view.)
Rotation, not recession: long energy and gold, short consumer and tech. George Noble is "passionately bearish on consumer stocks" and "wildly bullish on gold and energy stocks," and thinks Bessent should "resign" because "Mr. Market holds all the cards" (The Julia La Roche Show, Sep 22). (Stock-picker CIO, Noble Capital Advisors.)
Stay long the dollar; keep calm and carry on. JPMorgan's FX desk sees the hawkish Fed as a dollar tailwind and favors FX carry over equity risk in a tightening world (At Any Rate, Sep 18). (Sell-side FX strategy.)
Lean long duration, the next big move may be down. Snider's implicit trade is that a flattening, near-inverting curve means the long end is the place to be, and he notes even PIMCO is nibbling back into long bonds (Eurodollar University, Sep 23). (Research/newsletter; structurally dovish.)
High yields are attractive, but vet the private-credit competition. David Busch likes locking in "higher for longer" Treasury yields, but warns anyone chasing private-credit yields to demand a loan tape first (Monetary Matters, Sep 19). (Advisor.)

Read-throughs

5% is the regime, not the scare, and the market is genuinely split on what it means. The question has shifted from "will the 10-year hit 5%" to "what does a sticky 5% signal," with credible voices reading it as both a growth story and a solvency warning. That disagreement is itself the story.
The most important chart right now is the 2s10s. If it inverts and sticks, Snider's growth-scare read is winning and the long-end move is not an inflation alarm at all, bullish for duration, bearish for risk assets. If the long end keeps selling instead, it's the buyers'-strike/debt story. Watch which end of the curve is doing the moving.
The long bond has lost its natural owners, and the replacement plan is jury-rigged. Insurers, pensions, sovereigns and foreigners are all stepping back. The proposed replacements, basis-trade hedge funds, stablecoins, and the Treasury's own buybacks, are clever but fragile. That fragility, more than any single yield print, is the real dollar risk.
"The dollar" is now a pair-by-pair question, not one DXY call. Firm against a fiscally-stretched Europe, contested against AI-exposed Asia, and a high-conviction short for the reserve-diversification crowd. Beware anyone giving you a single directional dollar view without saying "against what."
Watch October, then November 3. A Fed skip in October (Kaplan's base case) versus another hike changes the whole "how many" math, and the midterms sit right behind it. The good news from this week: even a change in Congress probably can't expand the deficit much from here, because the bond market has already done the disciplining.

What changed this week

The 10-year decisively broke and held above 5%, turning a two-year-old threshold into the market's new working floor.
A brand-new signal appeared: the 2s10s flattened to about 20bp (Snider), the mirror image of last week's steepening story, reframing the long-end debate as a growth scare rather than only an inflation-and-debt alarm.
The "buyers' strike" got specific. The named natural owners of the 30-year, insurers, pensions, sovereign wealth funds, have defected to private credit and corporate bonds, and companies themselves won't issue long-dated debt.
A credible "too many hikes priced" voice arrived in former Fed president Robert Kaplan, pushing back on the December-and-beyond hiking consensus.
Bessent's stablecoin push was laid out as an explicit three-step debt-management plan tying buybacks, a shift to T-bills, and stablecoin demand into one scheme.
The midterm framing softened. Last week the read was "the election puts a floor under long rates"; this week JPMorgan Asset Management argued deficit expansion is constrained either way, and the bond market has already priced it.