The Fed Hiked Into a Supply Shock. The Long End Did Not Care.

A unanimous hike, hawkish dots, and a 10yr real yield in the top 1.2% of the past 23 years. Four central banks, one question: who is still the marginal buyer of duration?

The decision that was never the variable

The Federal Open Market Committee raised the target range for the federal funds rate by a quarter point to 3.75% to 4% on 2026-09-16. The vote was 12 to 0. It was the first increase since July 2023.

The market had it at more than a 90% probability going in, and when expectations of a hike have been that high, the Fed has delivered, going back to 2008. So the decision itself carried almost no information. What carried information was everything around it.

The median projection for the end of 2026 rose to 4.1% from 3.8% in June. Sixteen officials now see at least one more increase this year, against 6 in June who saw at least two. And the committee pushed out its expectation for inflation returning to 2% by a full year, to 2029.

Chairman Kevin Warsh, in the press conference, put it in a sentence: "We removed a dose of accommodation so that financial and credit conditions would be more consistent with our ultimate objectives."

That is not a one-and-done. It is the opening of a cycle, and the market repriced to near 100 basis points of tightening by the middle of 2027.

The curve flattened, which is the opposite of what a broken bond market looks like

Short-dated yields rose and long-dated yields did not. The 2yr touched 4.74%, the highest since 2024 and more than 10 basis points above its pre-meeting level. The 10yr finished lower on the day at 5.01%. The 30yr sat at 5.35%.

The 2yr to 30yr gap narrowed 8 basis points to the tightest since April. That is a curve being flattened by policy expectations moving up the front end, not a long end repricing on fiscal risk.

This distinction matters more than the headline. Earlier in the day, with the 10yr touching 5.04% and the long end at multi-decade extremes, the natural expectation was that a hawkish Fed would break the long end further: higher term premium, a steeper curve, the bond market refusing to take the Fed's word. It did not happen.

Every tenor from 5 years out sits at the 100th percentile of its 3yr range, according to this site's own series. The 30yr is at the 99.3rd. These are not levels that leave room for a further repricing on policy news. They are levels where the question changes from "how high" to "who is left to buy."

The number the whole week turns on

The 10yr real yield, the yield on inflation-protected Treasuries, is 2.62% as of 2026-09-15.

Against the 5,930 daily observations in the TIPS series, which begins on 2003-01-02, that level has been reached or exceeded 69 times. That is 1.2% of the sample.

The averages put it in context. The full-period mean is 1.01%. The mean from 2003 to 2008 was 2.06%. The mean from 2008 to 2022, the era that shaped a generation of portfolio assumptions, was 0.43%.

So two claims are simultaneously true, and most commentary picks one and drops the other. The first: 2.62% is not a return to some ancient normal, it is a level the market has visited in 1.2% of the last 23 years. The second: the decade of 0.43% was the actual anomaly, and there is no rule that says it comes back.

The window has to be stated plainly: this is TIPS history, which begins in 2003. It does not say anything about real yields in 1985. Anyone claiming "the long-run average" without naming the window is choosing a number that suits the conclusion.

This is why the scarcity complex is under pressure, and it is the same story as the 5% nominal 10yr told from a different angle. A scarce asset has to clear a hurdle to justify holding it instead of a bond that pays a real return. When that hurdle sits at 0.43%, almost anything clears it. At 2.62%, most things do not.

The inflation the Fed is fighting is not the inflation it can fix

The drivers of this inflation are the war in Iran and the energy prices that came with it, tariffs, and an artificial intelligence capital boom. Two of those are supply shocks. Central banks do not have a tool for supply shocks.

Wolfe Research's chief economist makes the point directly: if the Fed hikes once or twice, that is unlikely to change the backdrop one way or the other.

The third driver is more interesting, because it blocks the channel the Fed normally relies on. In past tightening cycles, housing cracked first, construction layoffs followed, and demand cooled with a lag. This time residential construction employment has been falling since September 2024, but overall construction employment hit a record in August 2026 on nonresidential and engineering hiring, which is the artificial intelligence buildout.

Barclays' analysis is that hyperscalers are spending over 90% of their operating cash flow on AI infrastructure, and that a 50 to 75 basis point rise in the cost of financing a data center is not going to change the decision to build it.

So the Fed is tightening into an economy that has partially disconnected from the rate channel, and its remaining lever is the consumer, who is more sensitive to borrowing costs and less responsible for the inflation. The 30yr mortgage is 6.76% as of 2026-09-10, up from 6.66% 3 weeks earlier, without the Fed having done anything yet.

What the market is actually priced for

Inflation expectations are not de-anchoring, which is the fact that argues against the panicked version of this story. The 10yr breakeven is 2.38% as of 2026-09-15. The 5yr 5yr forward, the market's estimate of inflation five to 10 years out, is 2.35%. Both are above the 2% target. Neither is running away.

Positioning, by contrast, is stretched. JPMorgan's Treasury client survey showed short positions rising 10 percentage points in the week to September 14, with most of that shifting out of neutral. Citi's strategist called the short base "tactically extreme." Options skew continues to favour puts in the long bond while sitting closer to neutral from the 2yr out to the 10yr.

And yet credit is not confirming stress. High-yield spreads are 2.76% as of 2026-09-15, from 2.65% on September 11: 3 sessions of widening, from a floor, not a break.

The honest read of that combination is a market positioned for a hawkish Fed that got one, in a credit market that has not decided whether the hiking cycle matters. Risk appetite on this site's composite is neutral, with the headline at 0.024, rates and policy the dominant input, and the scarcity component at just 0.16.

The same question, four different answers

The Fed is the only major central bank tightening into this. Everyone else is solving the same long-end problem with different tools, and the divergence is where the trade is.

The Bank of England has the most acute version of the problem and the fewest options. Long-dated gilt prices touched their lowest since 1998 on 2026-09-14. The Bank has spent 4 years reducing its gilt portfolio, and unlike the Federal Reserve and the European Central Bank, it has done so partly through active sales rather than letting bonds mature. It is the marginal seller of the exact paper the market no longer wants.

The reported plan, ahead of Thursday's decision, is to stop selling 20- and 30yr gilts. That is not stimulus and it is not a rescue. It is a seller standing down because it can see the bid is not there, and it frees up cash for the government by removing a drag on prices at the long end.

Japan is the purest case. The 10yr yield is 2.94% as of August 2026, its highest since the 1990s, while the policy rate sits near 0.98%. The Bank of Japan is the largest holder of Japanese government bonds and has been running down its balance sheet into a market with the developed world's worst duration profile. If it slows that runoff to protect the curve, it is the British playbook exported, at a much larger scale.

The euro area has the opposite problem. Harmonised inflation was 1.9% year on year, below target, and the 30yr yield is 3.80% against a 5yr at 3.28%. The European Central Bank has no reason to tighten and no reason to be in this conversation. That gap between a Fed at 4% and an ECB watching 1.9% inflation is the FX trade, and it is why the dollar rose half a percent on the day.

The yen, and what the carry market is actually pricing

The dollar rose against the yen to about 156 after the decision, up roughly 0.70%. For a day, that was read as the carry trade being given a stay of execution: a hawkish Fed widening the rate differential against a Bank of Japan that cannot respond, which supports the funding currency staying cheap.

The level is worth keeping in perspective. Dollar-yen was already at 156.17 in this site's data before the meeting, at the 62nd percentile of its 3yr range. The move is a return to where the day started, not a new regime.

What the carry market is pricing is that the Bank of Japan remains the constraint. The unwind risk was not eliminated by a Fed hike. It was deferred to the moment the Bank of Japan either moves or loses control of its long end, and the second of those is the one that does not give notice.

Energy is the joint between the inflation and the fiscal story

Brent closed at 130.80 a barrel on 2026-09-15. That is the number the Fed is reacting to, and it is not a monetary phenomenon.

It is also where the fiscal story starts. The United States is running a naval blockade of Iranian commerce that began on 2026-07-14. Twenty-eight Navy ships have rotated through it. The operating cost of the ships alone has topped 7.1 billion dollars since January, at roughly 32.5 million dollars a day, and that figure excludes the assets escorting tankers through the Strait of Hormuz.

The Congressional Budget Office put the total cost of the war at roughly 38 billion dollars as of 2026-08-01, with a burn rate of 2 to 3 billion dollars a month if the intensity stays low. Over 30% of the US Navy's warships are committed. Naval Support Activity Bahrain, the largest US Navy base in the region, is unusable after Iranian strikes. The CBO also judged that the United States has probably used as much as two-thirds of its missile defence interceptor inventory since June 2025.

Iran's economy is roughly 300 billion dollars and its inflation has already run above 80%. A state that has operated under sanctions for decades can absorb a blockade for a long time. The question is which side has the clock, and the honest answer is that the cost is being carried by the side with the Treasury market.

That is the joint. The war is the supply shock that put the Fed in this position, and it is simultaneously a spending commitment that lands on the long end of the same curve. The Fed tightening makes the funding side of that harder, not easier, and it cannot do anything about the supply side at all.

What would change the picture

The bull case for duration is not dead, it is just early. It needs the energy shock to resolve, and the mechanism for that is political rather than economic: a war that becomes cheaper to end than to continue, ahead of US midterms where the cost of living is the ballot question.

If that happens, the supply shock eases, the Fed's justification for further hikes weakens, and the 2.62% real yield starts to look like a peak rather than a plateau. That is the sequence that would let the scarcity complex and the long end both rally.

The bear case is that none of it resolves. The oil stays above 100 dollars, the Fed delivers at least one more hike, the long end holds at these levels on the same fiscal supply, and the real yield grinds into territory that has no modern precedent to anchor it.

What would settle it, in the observable data: the 10yr real yield. Above 2.62% with gold and bitcoin both falling is a discount-rate repricing. Back below 2.00% with both holding is the scarcity bid reasserting itself against a falling discount rate. The level is doing the work that narrative has been doing all week.