The September 2026 FOMC Concludes With No Increase

The first account of this failure was itself wrong, and the real cause is worse. Both the pre-settlement note below and the verdict beneath it said this call was overtaken by the Jackson Hole address of 28 August. That framing let the entry off. The call was built on an inverted fact.

What this page claimed. That the three dissents at the July 2026 meeting were for a cut, and that an increase was therefore the one outcome no committee member had argued for in public.

What actually happened in July. Beth Hammack, Neel Kashkari and Lorie Logan dissented in favour of raising the target range by 25 basis points. It was the most hawkish FOMC vote since September 2016, and contemporaneous coverage described those three dissents as signalling that a September hike was live. Three members had argued for exactly the outcome this call treated as having no constituency.

The fact was already on this site, stated correctly. Tempora's reconciliation of the July meeting, published 4 August 2026, records that the three regional presidents dissented in favour of a 0.25 point hike. This call was filed on 27 August, three weeks later, and inverted a fact the site had already checked and published.

What that changes. The stated 0.93 was not merely too confident for a three week horizon. It pointed the wrong way. The signal that would have moved it was in the record before the call was written, and a correct reading of July would have produced a call closer to a coin, or no call at all. Jackson Hole did not create this loss. It arrived at a conclusion the July vote had already flagged.

What changes in process. For any call that continues a franchise, the previous instance's own reconciled result text is now the source for facts carried forward, quoted rather than restated. The failure here was not forecasting judgment. It was research hygiene, and it is the more embarrassing of the two.

This call is likely to fail, and we are saying so before the result rather than after it. The FOMC statement lands at 14:00 Eastern on 16 September 2026. As of this morning, CME FedWatch prices a quarter point increase to 3.75 to 4.00 per cent at roughly 86 to 90 per cent. Our call requires no increase. If the market is right, it fires FAILED.

The paragraph below is superseded by the correction above: the dissent direction stated here is wrong.

What changed after we published. This entry went up on 27 August 2026. The reasoning rested on the July meeting, which held on a 9 to 3 vote with all three dissents arguing for a cut, and on the observation that no committee member had publicly argued for a hike. Chair Kevin Warsh delivered a hawkish address at Jackson Hole on 28 August, one day later. A solid August employment report followed, and the August CPI release on 14 September printed 4.82 per cent. Market pricing moved from a hold to a hike across those three weeks.

Why the entry is not being withdrawn or amended. The claim, the falsifier and the stated probability of 0.93 stand exactly as filed. A call that can be pulled once the odds turn against it is not a call. The purpose of publishing before is to be wrong in public when the world moves, and this is what that looks like from the inside, a few hours ahead of the print.

What we got wrong in the reasoning, assuming the market is right. The falsifier was designed to isolate the one outcome nobody on the committee had argued for. That design treats public committee commentary as the binding constraint on the next decision. A chair can move the distribution in a single speech, and the entry carried no allowance for that. The stated 0.93 was too high for a three week horizon that contained a Jackson Hole address.

The verdict will be recorded here after the announcement, either way.

FAILED. The FOMC raised the federal funds target range by 25 basis points to 3.75 to 4.00 per cent on 16 September 2026, the first increase since July 2023. This call required no increase and was stated at 0.93. It was wrong.

The vote was 12 to 0. That is the part worth sitting with. This entry did not lose a close committee call. It lost a unanimous one. The Summary of Economic Projections released the same afternoon showed 16 of 18 participants expecting a further increase this year, and four of those expecting two.

Superseded by the correction above.

What the reasoning claimed. The falsifier was built to isolate the one outcome nobody on the committee had argued for in public. In July the committee held on a 9 to 3 vote and all three dissents were for a cut. The inference was that a hike had no constituency. Seven weeks later that same committee raised rates without a single dissent. The reading was not slightly off. It was inverted.

What actually went wrong. The entry treated public committee commentary as the binding constraint on the next decision. It is not. It is a lagging read of a distribution a chair can move in one speech, and that is what happened: this call was filed on 27 August and the Jackson Hole address came on 28 August, one day later. A three week horizon containing a Jackson Hole address should never have carried a 0.93.

What we did before the result. A note went up on this page on 16 September, before the announcement, stating that the call was likely to fail, giving the market pricing of 86 to 90 per cent for a hike, and naming the design fault. It is above. The point of publishing in advance is that the record shows what was known and when, including the hours when a call is visibly going wrong.

What changes on this franchise. Meeting-level rate calls will either shorten the horizon to the post-blackout window, when committee speech risk is closed off, or carry a stated probability wide enough to hold a chair repricing the meeting. The instrument is not retired. Its horizon was wrong.