The Heat in Bonds Isn't Inflation. It's the Term Premium.

The 30-year Treasury yield brushed 5.31% in mid-August — its highest level since June 2007 — and the 10-year has climbed to roughly 4.95%, the most since October 2023. If you've been hearing that "bonds are selling off," the instinct is to blame inflation. That instinct is the part of this story the market is actually getting wrong. On the far end of the curve, the heat isn't inflation at all. It's the term premium — and that distinction matters more for your stock portfolio than anything the Fed does next.

Here's the difference in plain terms. Lending to the government for three months is barely a decision; you know what's coming quickly enough to change your mind. Lending for thirty years means locking your money up across wars, energy shocks, whoever runs the country, and the decades nobody can forecast. The extra pay you demand to lend longer is the "term premium" — the market's price tag for long-dated risk. When it rises, it isn't a bet on next month's consumer price index. It's a vote of no-confidence in the borrower's credibility and the path of policy years out.

The events of this summer are unmistakably a term-premium event. Inflation expectations — the "breakevens" the market bakes into bonds — stayed comparatively stable even as nominal yields climbed. That's the signature of a move in real yields, not a re-acceleration in inflation. Rising term premium does the work. And the reasons it's rising are structural, not cyclical.

The simplest way to see it: the marginal buyer of long Treasuries is stepping back. Foreign central banks used to be the price-insensitive backstop of the market — the buyer that absorbed supply without fussing over price. They are retrenching. Japan's 10-year government bond yield has blown past 3% for the first time since 1996, and the Bank of Japan has been selling Treasury securities to defend the weakened yen — a top holder of US debt effectively selling it instead of buying it. Meanwhile corporations are crowding onto the same side of the table; AI-related companies alone have issued more than $1.5 trillion of debt, much of it long-dated, competing with the Treasury for the same pool of money. Add a US federal debt that has doubled in a decade to over $40 trillion, and you have more paper chasing fewer willing hands.

Now for the mechanical detail that I think tells you the most about where this is headed. On August 19 the Treasury surprised the market by expanding its buybacks of longer-dated debt — its quiet attempt to support the long end and, frankly, to push back against the selloff. The street called it a "holding action," a signal rather than a change in the supply-demand reality. Then the first expanded operation was put to the test, and the result was revealing: the Treasury came to buy up to $6 billion, sellers offered $10.5 billion, and the government took just $5.2 billion. It left money on the table at precisely the moment the market wanted to see it absorb supply. The disappointment was read immediately, contributing to the fresh push higher in yields.

That is the part most commentary misses. Everyone wants to hang this on the Fed and whether it hikes next week — markets were pricing roughly two-in-three odds of a hike by month-end. But the Fed only controls the short end, the plumbing it steers directly. It has signaled through the new chair's hawkish resolve that the balance sheet is not coming back to rescue long-dated paper. So the long end — the part whose entire repricing is a term-premium vote — is left to find its own buyer. The heat ratchets higher precisely because the one obvious safety valve is no longer attached.

This is where the bond story becomes a stock story, and where I'd urge some skepticism about what the index is telling you. The cap-weighted S&P 500 has held up — still up more than 11% this year, only about 2% off its record. But an index that strong on a handful of the longest-duration names is not proof of broad health; it's the concentration mirage. The market's leadership this year is dominated by the AI complex — the very stocks whose value sits furthest in the future. A higher long-term discount rate is exactly the thing that hits those cash flows hardest. That's why a 10-year yield at 5% is treated as a "potential trouble spot": it's the level that, the last time the market saw it in 2023, coincided with broad stock weakness and forced de-risking. The earnings season that gave equities a buffer is over; second-quarter profit growth was enormous, but with that shield gone, the discount rate becomes the story.

To be clear about the one reading that would break my logic: not all rising yields are created equal. If the long end were climbing because the economy is genuinely booming and growth is strong, real yields rising could be a benign signal that stocks can ride. This cycle doesn't look like that — growth has been stalling even as yields climb, and the move is being driven by supply meeting fewer price-insensitive buyers. That's the repricing that hurts. The distinction to keep an eye on is whether the long-end buyers return and whether the Treasury's buyback underdelivery becomes a pattern rather than a one-off. If the term premium keeps ratcheting up against a weakening economy, the 5% line on the 10-year stops being a headline and becomes a regime test for the entire equity complex — and the cap-weighted index won't tell you it's happening until the leaders stumble.