Newsletter · · Ashutosh Agarwal

Treasury Auctions Cleared at Generational Highs While the Yen Rescue Leaks - The Long End & Fiscal Supply - Week of August 18, 2026

Rates and macro newsletter for the week of August 18, 2026. A soft inflation print let the Treasury sell 10-year and 30-year bonds at the highest yields since 2007 and 2001 to real buyers, while the U.S.-Japan yen rescue gave back half its gains and the bears put a number on fair value near 6 percent.

The Long End & Fiscal Supply

Week of August 18, 2026: The Auctions Held and the Yen Rescue Is Leaking

This was the week the long end got a real-time exam and mostly passed. Softer inflation landed on Wednesday, the Treasury sold a mountain of 10- and 30-year bonds at the highest yields in a generation, and buyers actually showed up. The reaction was a split screen: the bet on a September rate hike collapsed, but long-term yields barely budged, with a fresh 25-year high at the 30-year auction. Meanwhile the extraordinary rescue of the Japanese yen from two weeks ago quietly started to come undone. Underneath it all, the same argument kept surfacing on the podcasts: are these high long-term rates a slow-burning crisis, or just the plumbing of a curve that spent years upside-down finally turning right-side up?

TL;DR

  • The Treasury tested the market and it cleared, at a price. A soft inflation report let a $42 billion 10-year auction (highest yield since 2007) and a $25 billion 30-year auction (highest since 2001) get digested smoothly. Goldman's and JPMorgan's desks both said demand was fine; the buyers now are "price-sensitive" investors demanding more compensation, not foreign central banks or the Fed.

  • The yen rescue is already fading. Two weeks after the first U.S.-Japan yen intervention since 1998, the yen has given back about half its gains and is back above 159, its weakest since the operation, because big Japanese investors used the bounce to sell yen and buy foreign assets. The market's next line in the sand is 160.

  • The debate hardened into a clean fork. Bears now put a number on it, fair value for the 10-year near 6% (Darius Dale, Michael Howell), and say the Fed must hike to stop it. The sharpest bull, Jeff Snider, says the whole panic is a misread: the long end is rising because the curve is un-inverting, not because vigilantes have arrived.

What's New

1. Three Auctions, One Soft Inflation Print, and a Market That Cleared

The Markets: "How Inflation and Fiscal Policy Are Driving US Treasury Markets" (Mike Mitchell, head of U.S. Treasuries and inflation trading at Goldman Sachs, a live dealer desk, recorded on the trading floor) and The TreppWire Podcast: "The AI Capital Boom, Stubborn Treasury Yields..." (Stephen Bushbaum, head of applied research at Trepp).

For weeks this newsletter has flagged the same gap: lots of talk about supply, no actual auction color. This week the desks finally showed their work. On Wednesday, July's inflation report came in almost exactly as expected, headline consumer prices up 0.1% for the month and 3.4% from a year earlier, core up 0.2% and 2.5% (Bushbaum). Goldman's Mitchell called it "roughly down the middle, 21.5 basis points for core CPI," and noted the read-through to the Fed's preferred inflation gauge, core PCE, "was a little softer than that" because the software category, bigger in PCE, came in soft. That soft print did real work.

Right after it, the Treasury sold $42 billion of new 10-year notes, and the auction "cleared at a yield of 4.683%, with a bid-to-cover ratio of 2.53 times… indirect bidders took nearly 77% of the offering. Primary dealers were left with less than 9%," Bushbaum reported, "a very healthy demand story," even though it was "the highest it's been since 2007." (Bid-to-cover is simply how many dollars of bids came in for each dollar sold; indirect bidders are mostly foreign and institutional buyers; when primary dealers, the banks obligated to mop up whatever's left, are stuck with very little, it means real end-buyers turned up.) Mitchell, watching from the Goldman desk, said the 10-year "was reasonably smoothly digested despite the high yield level," and credited the soft inflation data: "The comforting inflation data brings buyers off of the sidelines… gives them comfort owning longer-term duration." He expected the next day's 30-year to go the same way "assuming we don't get a hot PPI print," and it more or less did, clearing around 5.22%, the highest 30-year borrowing cost since 2001.

Why it matters: The bearish story all summer has been "who is going to buy all this debt?" This week the answer, at least for now, was: plenty of people, but only at yields not seen in roughly 20 years. Both desks were blunt that this doesn't fix the underlying problem. Mitchell: the U.S. is on "a problematic fiscal path… it's not just this country, it's a global phenomenon. It's been driving increasing term premium… it will continue." And he flagged the mechanical squeeze ahead: "The market's very focused on the Treasury's need to increase auction sizes further in the coming years. It's clear that they will need to."

2. The Buyer Base Has Quietly Changed, and That's Why Yields Are Up

Bloomberg Surveillance: "Rising Equities and Bond Yields" (Priya Misra, JPMorgan, a sell-side rates strategist).

Misra was the calmest voice of the week, and her explanation of who buys Treasuries now is the most important thing many listeners will hear this month. "The auction was actually fine," she said. "We look at bid-to-cover, we look at end-user demand… If there wasn't demand to meet this supply, I think then there would be angst. But right now the auctions were fine." Her framing of the level: "If you zoom out and you go back to the late '90s or the pre-Lehman time period, actually interest rates don't look particularly odd."

Then the key point. "Who buys our Treasury securities these days? …It's changed in the last 10 years. It used to be foreign central banks. It used to be the Fed. And now it's what I call price-sensitive buyers, and which is why these interest rates have risen. What we call term premium, which is how much more you should get paid to extend out the curve, that has risen." The new buyers, she said, are "asset managers… what the Fed calls households" (meaning ordinary people's retirement and brokerage money in bond funds), and U.S. banks; foreign investors are still there, "just smaller than it was 10 years ago." Crucially, she decomposed the steeper curve and found it's "real rate," not inflation, which to her means "the Fed is still credible. The market expects inflation to come back down." Her bottom line: "There is income in fixed income… we just have to get used to the new normal of higher interest rates."

Why it matters: This is the bull case in one sentence: the long end isn't broken, it's re-priced. The Fed and foreign central banks used to buy bonds regardless of price; the new marginal buyer demands compensation, so yields sit higher, but the demand is real and it clears. It reframes "5% is a crisis" as "5% is what a free market charges."

3. The Contrarian Bomb, This Isn't Vigilantes, It's the Curve Un-Inverting

Eurodollar University: "Something Really Strange Just Happened at the 30-Year Bond Auction" (Jeff Snider, independent analyst).

Snider took direct aim at the entire "bond vigilantes are back" narrative, and he did it with a number that stops you cold. Yes, the 30-year priced at its highest yield since 2001. But that, he argues, is not the market waking up to America's debt. Consider, he said, the gap between the 3-month Treasury bill and the 30-year bond, the compensation investors demand to lend for 30 years instead of 3 months. Back in 2011, when S&P first downgraded U.S. debt, that spread was 379 basis points, with total debt of $14.3 trillion. Today, "despite a $25 trillion increase in the debt," total debt now $39.1 trillion, "the spread… is just 134 basis points." In plain terms: the government owes almost three times as much, and the market is charging less extra to hold its longest bond, not more. "The market is not demanding additional compensation for holding Treasury paper that it sees now suddenly as getting more and more worthless."

His read: the 30-year is rising for a mechanical reason, the yield curve, historically inverted since 2022, is "un-inverting," a process he dates to September 2024. And auctions, he insists, are a sideshow: the truly "horrific" 30-year auctions after the 2011 S&P downgrade (the bond jumped 28 basis points intraday) and the 2023 Fitch downgrade "didn't mean a damn thing," and both were followed by bond rallies. He even said the quiet part out loud: "I wish it was" vigilantes, "I wish we did have bond vigilantism to get the government under some kind of control. But that's not what's going on here."

Why it matters: Every crisis take this week rests on the premise that the market is demanding more and more to fund the deficit. Snider's spread math is the single cleanest piece of evidence against that premise, and it's testable. If he's right, the long end stabilizes as the curve finishes normalizing; if the bears are right, that 134-basis-point spread should start blowing out.

4. The Bears Put a Number on It, Fair Value Near 6%

Forward Guidance: "The Growth Strategy Trapping The Fed" (Darius Dale, 42 Macro) and Macro Voices: "Michael Howell: Warsh vs. The Markets" (Michael Howell, CrossBorder Capital), both independent strategists.

Where the bulls see a re-pricing, the bears see a long way further to fall in price (and rise in yield), and this week they got specific. Dale: "The fair value for the 10-year, according to our model, is 5.8%. And we don't do any mental gymnastics to get there." His method is simple: term premium, the extra yield for lending long, is currently about 78 basis points, versus a pre-2008 average of 188 basis points. Add that gap back and "you wind up with a 10-year treasury that's somewhere close to 6%… 30-year treasury yield somewhere well north of 6%. That's the risk if the Federal Reserve does not appease the bond vigilantes with tighter monetary policy."

Howell got to a similar 6% from a different door. His model ties the 10-year to nominal GDP, real growth plus inflation, which he now pegs at "somewhere between a 6% to 8% range," driven by "huge fiscal spending, the AI boom, and the effects of deglobalization." On that math, "it's not impossible to see the 10-year bond testing 6% yields in the not too distant future." His warning about how the Treasury copes is the important part: to avoid selling long bonds at those yields, it keeps funding at the short end. Treasury bills are already about 22% of federal debt, above the Treasury's own stated 15–20% comfort range, and Howell thinks it "may go up to levels we last saw in the early 2000s, which were nearer 30%. That would… be really bad for the dollar, in my estimation, and it will cause the gold market to shoot up." He calls heavy bank buying of short-term government paper "monetization… writ large," effectively money-printing by another name.

Why it matters: "Yields are high" is vague. "Fair value is 6% and the Fed has to hike to defend it" is a falsifiable call, and two respected, independent strategists arrived at it the same week from opposite starting points (term premium versus nominal GDP). That's the crisis camp's strongest week yet.

5. Two Weeks On, the Yen Rescue Is Unwinding

RiskReversal Pod: "Peter Boockvar: Inflation Is The Core Disease" (Peter Boockvar, CIO of BFG Wealth Partners, buy-side), The KE Report: "Marc Chandler..." (Marc Chandler, Bannockburn Global Forex, currency strategist), and Patrick Boyle On Finance: "The Hidden Risk in the US-Japan Yen Rescue".

Last week's blockbuster, the first U.S. yen intervention since 1998, engineered so Japan wouldn't have to dump Treasuries, is already losing its grip. Boockvar: the intervention "worked, but it's not really working now." Patrick Boyle put it vividly: over two days at the end of July, Japanese and American authorities threw "something like $88 billion" at the market and the yen surged about 5%, from a 40-year low near 164 toward 155. "And then the laws of financial gravity reasserted themselves. Within two weeks, the yen had surrendered about half of that, weakening back past 159." Chandler had it at "159.55… the highest it's been since the intervention," with the market now eyeing the "old thread at 160" as the next line in the sand.

The reason it's fading is the most interesting new detail of the week. This is not a carry-trade blow-up. Chandler explained that in the first week after the intervention, "Japanese investors bought the most amount of foreign bonds and stocks… in two years. They took advantage of the little bit of a bounce that the intervention caused… and with that stronger yen, they could buy more foreign assets." In other words, real-money Japanese institutions used the government's rescue as an exit, selling the stronger yen to buy foreign assets (much of which, Chandler suspects, flowed right back into U.S. stocks and bonds because of how their benchmarks are built). Boyle cited Berkeley's Barry Eichengreen calling $88 billion "small potatoes… a rounding error" in a currency market that trades trillions a day, and Peterson's Maurice Obstfeld dubbing the whole strategy "cakeism," the belief America can have a weaker dollar with no inflation, low borrowing costs with huge deficits, and Japan both buying its bonds and strengthening its own currency. "You can want all of those things, but you can't have all of those things."

Why it matters: The rescue bought time, not a solution. As nearly everyone agreed, it only truly "works" if the Bank of Japan raises rates, and until it does, the pressure that forces Japan to consider selling Treasuries is still building.

The Debate

The podcasts split cleanly again, but the ground shifted. Last week the bull case was carried by a heavyweight buy-sider (BlackRock's Rick Rieder). This week it's carried by two very different voices: a contrarian analyst and a sell-side desk.

The crisis camp (Dale, Howell, Gromen, Schiff, Iuorio). Fair value is roughly 6% and rising, term premium is far below its historical norm, and the Treasury is papering over it by flooding the front end with bills. Luke Gromen on The Meb Faber Show: "The Bull Market That Loses You Money" gave the darkest long-run version: the U.S. "can't afford [10-year rates] above 4.7%," with almost 100% of tax receipts now consumed by interest and entitlements that are growing faster than revenue. His historical analog: the last time debt was this high, after World War II (110%+ of GDP in 1946), it was worked down to roughly 55% by 1951 through "capital controls, significant inflation, and real rates that bottomed at negative 13%." His 2021 study found that cutting today's debt load the same way would need "real rates… negative 12 to negative 16% for five straight years," brutal for anyone holding long bonds ("thank you for your service, long-term Treasury holders"), rocket fuel for gold and Bitcoin. He argues we're only at the "bargaining" stage of accepting this, "maybe Bessent can intervene in the yen and do stablecoins and T-bills and that'll lower interest," with "depression and acceptance" still to come. Jim Iuorio on Soar Financially: "Gold Is Front-Running New Money Printing" flagged that bills are now 21.7% of marketable debt, above the usual 15–20%, and called it a "Ponzi scheme," targeting 4.95% on the 10-year before the Fed steps back in to buy.

The it's-not-a-crisis camp (Snider, Misra, and, in his own way, Eisman). Snider's 379-versus-134-basis-point spread (above) is the intellectual core: the market is charging less to hold long bonds even as debt has tripled, so this can't be a debt panic. Misra's "auctions were fine, term premium is real-rate, the Fed is still credible" is the practitioner version. And Steve Eisman, on The Real Eisman Playbook: "The AI Trade, Rising Rates...", delivered the week's best reality check on how "explosive" this move actually is: "If you look at the 400 days since Trump was inaugurated, this is the narrowest range in 10-year yields that we've ever seen. The range… is 85 basis points over the last 400 days. It's been a remarkably narrow range… I just don't think this move in yield is as explosive yet as it's going to need to be" to hurt stocks. His team hasn't "found the rate of interest that gets money to leave equities"; they thought it was 4.5%, "now it seems higher." And, importantly, he agrees the backup "has largely been in real rates… it hasn't been in inflation," which "argues for not tightening, even though I think the Fed will."

Where they actually agree: the whole thing hinges on Japan and the Fed. Nobody thinks the yen rescue holds without a Bank of Japan rate hike, and the September Fed meeting sits right on the fault line between the two camps.

Trades and Positioning in Play

  • Bessent's bill bet is "underwater." Patrick Boyle laid out the wager buried in Treasury policy most clearly: by funding overwhelmingly with short-term bills for two years, Treasury Secretary Scott Bessent is effectively betting long-term rates will fall so he can issue long bonds cheaply later. "This is the kind of directional bet you'd expect from a hedge fund," and Bessent ran one. But "rates have not been falling," so the bet "would look very underwater indeed." Boyle drove home the irony: Bessent made his name shorting the Bank of England in 1992 and the Bank of Japan under Abenomics, and is now "sitting in Alexander Hamilton's old chair, using public money to defend a currency against 30-year-old hedge fund analysts running his exact old playbook." This is commentary rather than insider policy detail, but the framing to watch is concrete: the Treasury has abandoned its decades-old "regular and predictable" doctrine, and rolling short-term bills "works beautifully right up until the time you have to roll those bills over at a higher interest rate."

  • Gold as the fiscal-dominance trade. Both Dale and Gromen are structurally long gold as the debasement hedge. Dale said his firm pivoted its model portfolio's 30% bond allocation into gold back in the fall of 2024, and pointed to gold's rising share of global central-bank reserves as "frightening stuff [that] doesn't happen very often," a signal of eroding confidence in the dollar as money. Gromen: the "debasement trade" is "not close to dead… it has to run for the next five to 10 years minimum" because reshoring is inherently reflationary.

  • Scale back beta. Howell's takeaway for equities: with rising bond yields and potentially rising oil, "one ought to be scaling back beta exposure within equities and risk asset markets." He thinks the path resembles late 2021 and early 2022, a period in which the S&P fell 25% and Bitcoin fell 75%.

Read-throughs

  • Mortgages and housing. The through-line from the desks: even if the Fed pauses or cuts, the long end may not follow. As Trepp's Bushbaum put it, "a Fed pause or even an eventual decline in short-term rates does not guarantee the 10-year Treasury or CRE borrowing costs will fall by the same amount." The Remnant Finance: "How Interest Rates Actually Work" episode made the same core point for homeowners: the Fed controls only the short end, while "the bond market controls the long end where mortgages are priced."

  • AI credit is now openly competing with Treasuries. This is the loudest read-through of the week. Trepp's Bushbaum: four major hyperscalers issued roughly $195 billion of bonds in the first half of 2026; issuance from the five largest is projected at roughly $250 billion in 2026 and potentially $400 billion in 2027; NVIDIA unveiled a $500 billion third-party AI-infrastructure financing ecosystem, and Bank of America announced another $250 billion fund, "close to a trillion dollars in capital this week for AI infrastructure." That is a second, private-sector wall of long-duration borrowing landing on top of the Treasury's, and both Goldman's Mitchell and the Trepp team named it as a structural reason term premium stays elevated.

  • Gold and the dollar. The dollar has been slipping, and the intervention's euro-for-yen mechanics only reinforced the debasement narrative (Dale, Gromen, Howell). Central-bank gold buying "remains robust despite higher yields," per the Saxo Market Call: "Treasury market reaction to US CPI the next key" desk (John Hardy).

  • Equities' pain threshold is higher than everyone thought. Eisman's history lesson: in Japan in 1989, JGB yields went from 4% to 8% while the Nikkei melted up; in the U.S. in 1999, 10-year yields went from 4% to 7%; in 1987, long rates went 6% to 9% while stocks rose 30% for most of the year. His warning: "When stocks are going straight up… at the same time long-term interest rates are going up, and the stock market seems impervious, that's the point at which you have the most risk."

  • Cross-sovereign correlation, and a Japan tell to watch. Howell noted Japan's own long end is repricing for the same reason as America's: the 10-year JGB is "catching up with nominal GDP growth" of over 4% and "should be closer to 4% than 3%." Boockvar's key question if the Bank of Japan hikes: at what JGB yield do Japanese investors start repatriating their enormous overseas holdings? Higher Japanese long rates bring that repatriation, a potential drain on U.S. and European bonds, "closer."

What Changed vs. Last Week

  • The intervention flipped from triumph to test. Last week's frame was "Washington bought yen to defend its bond market," and it read as a decisive win. This week the yen has surrendered roughly half its gains and sits back above 159, its weakest since the operation, with 160 the next line in the sand (Chandler, Boyle, Boockvar). The mechanism of the fade is new and important: it's not a carry-trade unwind but big Japanese institutions using the bounce to sell yen and buy foreign assets, the opposite of the panic selling the rescue was meant to prevent.

  • The persistent auction gap finally closed, with a clear verdict. For weeks there was no auction desk color; this week we got three live auctions and dealer commentary from Goldman and JPMorgan. Verdict: heavy supply cleared at generational-high yields (10-year 4.683%, highest since 2007; 30-year around 5.22%, highest since 2001) on genuinely healthy demand (indirect bidders roughly 77% of the 10-year). The bearish "failed auction" scenario did not happen.

  • The front end and the long end split. Last week's fear was a September Fed hike. This week soft inflation plus a weak jobs report (July payrolls −23,000, prior two months revised down roughly 103,000) and soft retail sales collapsed those odds: Goldman's Mitchell said September was pricing just "9 basis points," and Chandler put a hike near 25%. Yet the long end barely eased and the 30-year still printed a 25-year auction high. So the "the data will bail out the Fed" trade partly happened, but only at the short end. The supply and term-premium story at the long end did not get the memo.

  • The bears now have a hard number. Last week's bear case was directional (neutral rate rising 50–75bp). This week it's a level: fair value roughly 5.8% to 6% on the 10-year, 30-year "well north of 6%" (Dale via term premium; Howell via nominal GDP). That's a sharper, more falsifiable claim.

  • The September Bank of Japan hike went from coin-flip to near-consensus. Last week markets priced roughly a 50% chance. This week Boockvar and Chandler both put it at 75–80% (Chandler: from roughly 5–6 basis points of tightening priced pre-intervention to about 20 basis points now), and Japan's Prime Minister Takaichi is reportedly, via a Bloomberg leak, now endorsing a hike, having realized intervention alone "never lasts" without a rate increase. The master variable is firming toward "hike."

  • A record fiscal print entered the story. July produced the largest single-month U.S. budget deficit on record, cited around $432 billion by pundits Peter Schiff and Craig Hemke, though Chandler noted it was inflated by roughly $100 billion of refunds for tariffs later deemed illegal. Either way, it put a fresh, concrete fiscal data point behind the supply narrative.