The Long End Has Read Williams and Declined
The Long End Has Read Williams and Declined
On Tuesday the president of the New York Fed spent the session talking the market out of an October hike, and the market obliged. CME FedWatch odds of a 25-basis-point move next month slid to 51.5%, down from nearly 70% a day earlier. In the same session the 30-year Treasury touched 5.62%, its highest since June 2002, and the 10-year pushed to 5.29%, a whisker from its 2007 peak. Read those two facts side by side, because they contain the entire week. The Fed offered a softer voice and the part of the curve it cannot instruct answered by selling.
Michael Barr, for his part, told anyone listening that more hikes are probably needed. So the central bank spent the day sending two messages to a bond market that had already stopped taking dictation.
Call it what the trade desks call it: a bear steepener. Front end calmer, long end unwell. It is the signature of investors who believe the policy rate is not the variable that matters anymore. They are pricing a war that is entering its seventh month, a Treasury that must roll a $40 trillion pile of paper into a world where Japanese life insurers are being lured home by JGB yields above 3%, and an oil price that sits near $100 Brent even on a day when Saudi Arabia resumed exports through its East-West pipeline. None of that answers to a fed funds decision in October or December.
I keep thinking about 1951.
For most of the 1940s the Federal Reserve held long Treasury yields near 2.5% so the government could finance a world war cheaply. It worked, in the way that anesthesia works. When Korea erupted and prices started running hot, the Fed wanted its balance sheet back and the Treasury wanted the peg to stay. Truman summoned the FOMC to the White House that winter to make the case in person. The stand-off ended in March with the Treasury-Fed Accord, and the terms were simple: the price of long money would be found by the market, and the central bank would be free to worry about inflation instead of the government's borrowing bill. Martin took the chair a month later and spent two decades defending the principle.
Seventy-five years on, nobody has pegged anything, and yet the argument has come back inverted. This time the market is doing what the Accord promised it would, discovering the price of long money without anyone's permission, and it is discovering a number that a government with net interest costs near a trillion dollars a year will hate. Treasury runs a buyback program through November 4 that is, at bottom, a piece of theater aimed at that number. The 30-year has treated it as a rounding error.
Now add the household, because the household reported for duty on Tuesday and the report was ugly.
The Conference Board's confidence index fell to 81.9 from a downwardly revised 88.6, against a consensus near 90. That is the lowest reading since 2014. The share of consumers calling current business conditions good has flipped negative for the first time since September 2024, and year-ahead inflation expectations jumped 0.6 points to 4.6%. Job openings in the August JOLTS report dropped 256,000 to 7.079 million, under the 7.225 million economists expected. Mortgage News Daily had the average 30-year fixed at 7.58%.
A consumer who expects 4.6% inflation, sees fewer jobs, and cannot refinance a house is a consumer behaving like the last marginal buyer in a shrinking market. Confidence surveys are soft data and they have cried wolf before. What makes this one awkward is the company it keeps. Long yields at 22-year highs are usually accompanied by an economy that deserves them. This economy printed 162,000 jobs in August, and now its own households say they are bracing for the opposite.
Equities noticed and declined to care. The S&P 500 gave back about 0.16% to land near 7,671, the Nasdaq Composite slipped 0.09% to 26,797, the Dow shed roughly 130 points to 51,350. That comes after Monday, when the Dow lost 347 points and Arm Holdings fell 8.7% on a session where the Nasdaq dropped 0.9%. Two red days, no rout. Stocks are behaving like a passenger who has been told the pilot is arguing with air traffic control and has chosen to keep watching the movie.
The movie is AI, naturally. Nvidia now trades below the multiple of the S&P 500 after announcing a $150 billion buyback, which makes it the only large bidder in the room. Micron reports Wednesday after the close and will tell us whether the memory chain believes its own hype at these discount rates. AMD spent $8.2 billion on World Labs, an acquisition with a certain 2021 flavor to it. Everyone has cash and conviction. Nobody has a duration hedge.
The calendar does not leave much room for nerves. Wednesday brings ADP, August PCE, the third estimate of second-quarter GDP and Micron. Friday brings the September payrolls, where the consensus of 84,000 is roughly half of August's tally. If that number lands soft, the Fed's dovish faction gets its December and the front end rallies. Then we find out whether the 30-year is listening. My guess is that it is not, because the reasons it is selling off are fiscal, geopolitical and global, and none of them get repaired by a labor report.
The lesson of the Accord was that a central bank buys its credibility by refusing to serve as the government's shock absorber. It took a president's tantrum and a change of chairs to learn it. Today the tantrum is optional, since the bond market is administering the lesson for free. Every basis point of the long end above the policy rate is an invoice for the war, the deficit and the doubt about who will hold the paper. The Fed can read that invoice as a mandate to hike or as a warning that its own tools have run out of purchase. Whichever it picks, it should stop pretending the bill was addressed to someone else.
Upvoted! Thank you for supporting witness @jswit.
Interesting how the Fed’s softer tone pushed the 30‑year to 5.62% while the 10‑year barely missed its 2007 peak—does this suggest the market is still pricing in more hikes despite Barr’s caution? 🚀📈💬