The Fed's September hike: funds rate 3.75%–4%, and the first cut waits for 2028

The front end of the United States curve is priced for a policy rate that does not come down. The federal funds target range is 3.75% to 4.00%. The effective funds rate — what banks actually pay to borrow overnight, published in the Federal Reserve's H.15 release — stepped from 3.63% to 3.88% on 17 September and has sat there since. The two-year Treasury note, the instrument that does little except price the next two years of that rate, settled at 4.71% on 22 September in the Treasury's own daily par yield curve, below its 4.76% reading on the 18th and the 21st. That 4.76% was the highest in the series since 1 July 2024.

Three numbers, one reading: the committee raised, and the market treated the raise as a floor rather than a ceiling. The two-year sits 83 basis points above the 3.88% that banks are actually paying overnight, and a note yields what the market expects the overnight rate to average over its life. Eighty-three basis points of premium is the front end saying the policy rate goes higher, not lower. The committee's own published projections agree with it, and put the first cut in 2028.

What the committee did, and how little it said

The Federal Open Market Committee (FOMC), the Fed's rate-setting body, met on 15 and 16 September and raised the target range by a quarter of a percentage point, to 3.75% to 4.00%. The statement was approved 12–0 and runs to four short paragraphs: there is no dissenter to interpret and no forward guidance in it. Inflation "remains elevated," it says, and the action "will support a timelier return to the Committee's 2 percent goal." It closes on a sentence this committee does not always use: "The Committee will deliver price stability."

Kevin Warsh chairs both the Board of Governors and the committee — the Fed's own board listing puts him first among the seven governors — and the market reads his phrasing for the reaction function the statement leaves out. At the press conference after the decision he described the move as removing "a dose of accommodation," and gave the reason: "I would be hard-pressed to describe broad financial conditions as restrictive. This view was widely shared by the Committee."

Put plainly, the Fed does not think it has tightened very much yet. A reporter put the obvious follow-up to him — when a central bank starts raising, does it generally follow with a sequence of hikes? "I'm not in the forward guidance business," Warsh said. "You heard from other people in the dots effectively, what their forecasts are. I'm not going to pre-judge any future decisions we make." The dots, then, are where the answer lives.

The path the committee wrote down

The Summary of Economic Projections (SEP) is the committee's quarterly forecast, published with the September statement. Its memo line on the projected appropriate policy path is the only place the Fed states a rate path at all. The medians, September against June:

Two lines carry the answer to the question the front end is asking. The median year-end funds rate is 4.1% for 2026 and 4.1% again for 2027. It is 3.9% for 2028. Read as a path, that is a hold through the whole of next year and a single quarter-point cut during the year after. The first decline in the median path is eighteen months away.

The ranges say the same thing with the discretion stripped out. For 2026 the projections span 3.9% to 4.4%. The midpoint of the range the committee set on 16 September is 3.875%, so the floor of the 2026 range sits just above it: on the committee's own numbers, no participant has the funds rate below today's setting at the end of this year. Nobody has a cut in 2026, and the median still has tightening in it.

What moved was not only the near year. Every line of the path was marked up, and so was everything around it: growth for 2026 from 2.2% to 2.3%, unemployment down from 4.3% to 4.1%, PCE inflation up from 3.6% to 3.7% and core PCE from 3.3% to 3.4%. A committee revising inflation up, unemployment down and growth up does not revise its rate path down. The one figure that barely moved is the longer-run rate, 3.1% to 3.2% — the committee still says it is holding policy above neutral, but by less than its own statements imply.

For context on how unusual the position is: the last increase before this one became effective on 27 July 2023, at 5.25% to 5.50%. Five cuts between September 2024 and December 2025 took the range down 175 basis points. The meetings in January, March, April, June and July of this year changed nothing. The range set on 16 September is the same range the committee last set on 30 October 2025.

What the front end did with it

The Treasury's par yield curve is a daily snapshot built from indicative bid-side quotations gathered by the Federal Reserve Bank of New York at or near 3.30pm, but it is the one series the Treasury publishes for every tenor every session. Across the meeting it reads like this:

Yields in per cent. The funds rate is the effective rate from the H.15 release of 22 September, which carries data through 21 September: the Fed's own daily release runs a session behind the Treasury's.

The two-year did its repricing before the decision, not after: 4.39% on 8 September, 4.63% on the 11th, 4.74% on the day of the announcement, 4.76% by the Friday. Thirty-seven basis points in eight sessions, seven of them on the Wednesday. That is what a thoroughly briefed hike looks like. The market had a 90% probability of it priced in on the morning of the decision, as one reporter put it to Warsh, and a priced certainty does not get repriced twice.

The one-year and the three-month bill went with it, adding 28 and 22 basis points respectively between 8 and 22 September, and the front end was flat to slightly lower at the end: the two-year gave back five basis points on Tuesday the 22nd. The ten-year did not move at all across the same run — 4.96% on 8 September and 4.96% on 22 September — and the long end's own round trip through 5%, and the buyers at its auctions is a different question from this one. The gap between the ten-year and the two-year closed from 41 basis points to 25.

Separating the arithmetic from the forecasts

A two-year at 4.71% against a policy midpoint of 3.875% is arithmetic: it is the average overnight rate the market expects over the next two years, and it sits above today's rate because the market expects the rate to rise. What sits above that arithmetic is opinion, and the loudest opinion this week was louder than the market.

Bank of America's rates strategists argued in an 18 September note that the market is placing the terminal point of this cycle too low and that the policy rate could exceed 5%; they raised their year-end two-year yield forecast from 4.5% to 5.0%, as reported by Yonhap Infomax on 21 September. The same bank's economics team forecast quarter-point increases in both October and December, taking the range to 4.25% to 4.50% by year-end. Swap rates were pricing about three more increases, which lands the policy rate in the mid-to-high 4 per cent area. The market has not been pricing a 5% policy rate. One bank has been forecasting one.

The distinction is worth keeping because the two things behave differently in a portfolio. A price is something a holder can lose money against. A forecast is something a holder can disagree with, and be right.

What the two-year auction paid

The front end states its own price once a month, at the two-year note auction, and September's was on the 22nd. The Treasury sold $69,000,071,900 of new two-year notes at a high yield of 4.787%, with bids covering the offer 2.63 times. The high yield was the highest at any two-year sale in the Treasury's published results file, a run of 25 auctions going back to September 2024. It was 58.3 basis points above August's 4.204%, and 133 above February's 3.455%, the lowest of the nine sales this year.

One convention, because the raw file will not yield these figures unaided. The published results show a subtotal of $69,000,071,900, a separate SOMA line of $10,387,942,900, and a total of $79,388,014,800. The published bid-to-cover of 2.63 divides the tendered $181,263,862,700 by the subtotal, not the total; divide by the total and the cover reads 2.28. The add-on is there because the FOMC's directive to the Open Market Desk instructs it to "roll over at auction all principal payments from the Federal Reserve's holdings of Treasury securities." The Fed is a standing bidder at the front end, and the Treasury prints its size on a separate line so that the public takedown can be read on its own.

What it means for a Hong Kong book

The peg takes Hong Kong's short rates from the Fed's range rather than from the local economy, and the first link in the chain has already moved. The Hong Kong Monetary Authority's base rate — the discount window price, published in its daily monetary statistics — stands at 4.25% on 22 September, a quarter-point above the 4.00% this desk read on 13 September when it set out how a 5% long bond transmits to Hong Kong. That piece traced the long end through the peg. This is the same chain with the policy path at the top of it.

The second link has not moved, and that is the part worth noticing. The Hong Kong Association of Banks' one-month fixing was 2.84339% on 22 September against 2.88% on 11 September, and the overnight rate was 2.00% against 2.28% — both slightly lower, in the week the Fed raised. The peg adjusts through the aggregate balance, the clearing balances banks hold with the HKMA, which stood at HK$53,975 million on 22 September. Only when the Hong Kong dollar is pushed to the weak-side convertibility undertaking at 7.85 does the HKMA buy it, drain that balance and force interbank rates up. At 7.8434 on this desk's ticker, the currency is nowhere near the undertaking, nothing is being drained, and the overnight rate is still being set by local liquidity rather than by the Fed's range.

What the policy path changes is the direction of travel underneath. The base rate is now tied to a range whose median, on the committee's own numbers, does not come down before 2028. A mortgage priced off the cap — the ceiling under the borrower, set as a fixed discount to the banks' prime rate, which follows the Fed with a lag — therefore has only one way to move for as long as that path holds. The cap has been doing the work on most new Hong Kong mortgages for months. The path says it keeps doing it.

What to watch

The FOMC next meets on 27 and 28 October 2026, its published calendar shows, and it will not issue a new set of projections then: the next SEP comes with the meeting on 8 and 9 December. The minutes of the September meeting are due about three weeks after it, as every set this year has been. Before any of that, the October two-year sale, which this year has fallen between the 22nd and the 27th of the month.

The levels to carry are the ones just set. A two-year reading above 4.76%, the high of the 18th and the 21st, would say the front end still has increases to price; a close back below 4.63%, where the run began on 11 September, would say the market has stopped adding to the path and is waiting for the committee to speak. On the committee's side the number is the 2026 median, 4.1%. It sits above the midpoint of the range the committee just set, and if October passes without a move, December is the next place the Fed has to say what it means by it.