Why the Fed Might Raise Rates in September 2026 After Markets Expected Cuts

In January 2026, the consensus forecast called for at least one rate cut during the year. The economy had come through a government shutdown in late 2025, and inflation looked like it was drifting down. As recently as June, the FOMC held rates steady with cuts still the base case for later in the year. That expectation has now flipped.

The July 28-29 FOMC meeting held rates at 3.5% to 3.75%, but the vote was 9-3. Beth Hammack of Cleveland, Neel Kashkari of Minneapolis and Lorie Logan of Dallas each dissented in favour of an immediate 25 basis point hike. By the St. Louis Fed's record of FOMC dissents, no chair since Arthur Burns in 1970 has drawn that much opposition this early in a tenure.

On August 26, the Bureau of Economic Analysis released July PCE. The price index rose 0.2% for the month and 3.7% over 12 months, a tenth above the 3.6% forecast. Core PCE, which excludes food and energy, rose 0.2% monthly and 3.3% annually. Both matched expectations, and both marked a third straight month at or above 3.3%.

Then came Jackson Hole. Warsh delivered his keynote on August 28 and the tone was hawkish. No new inflation data arrived with it; the July PCE report was already two days old. What changed was how he framed the numbers that were already on the table.

The two places you can watch that repricing disagree, and the gap is worth understanding. CME FedWatch derives probabilities from fed funds futures, where a hike moved from roughly 35% the day before the speech to about 56% after it. Kalshi runs an event contract on the same meeting, and it moved from about 30% to 47%. Futures-implied odds and prediction-market prices are built differently and are not measuring quite the same thing, so a spread between them is normal rather than an error.

Date Event Hike odds, CME FedWatch Hike odds, Kalshi August 27, 2026 Day before the keynote ~35% ~30% August 28, 2026 After Warsh's Jackson Hole keynote ~56% ~47%

Market-implied probability of a 25 basis point hike at the September 15-16 FOMC meeting, as reported August 28-30, 2026. Sources: CME Group, Kalshi. Market-implied odds measure expectations, not outcomes.

Year-over-year inflation held at 3.7% in both June and July 2026. Read only the headline and inflation looks stuck but stable. The Fed does not stop there.

Warsh put the gap in his own words at Jackson Hole: "The Fed's preferred measure of inflation, the 12-month change in the PCE price index, stands at 3.7 percent, while the six-month change is 4.1 percent." A six-month rate above the twelve-month rate says the recent pace is worse than the annual figure, not better. That single comparison is why the conversation changed.

The mechanism connecting the two, and why a central bank answers rising prices with higher rates at all, is the subject of inflation, rate hikes and cuts. Year-over-year inflation compares prices today with prices exactly 12 months ago. Everything in between, good months and bad, gets averaged together. If inflation was high a year ago and is high today, the annual number can sit flat while the recent trend accelerates underneath it.

Weight loss is the intuition. You step on the scale and compare with a year ago: down five pounds, so the annual trend looks fine. But if you have put on three pounds in the last month, the recent direction is wrong, and that is the part that should change what you do next.

Warsh gave a second version of the same point that has nothing to do with averages. Over the past 12 months, he said, 54 percent of the goods and services in the PCE basket rose more than 3 percent. Over the past six months, annualised, 49 percent did. Inflation concentrated in a handful of categories is a supply story. Inflation above 3 percent across roughly half of what people buy is harder to describe that way.

KPMG's read on the July data was that core inflation is not moving decisively lower, but grinding sideways at a pace too hot for comfort.

Measure Reading What it says PCE price index, 12-month change 3.7% The headline, unchanged from June PCE price index, 6-month change 4.1% Recent pace running above the annual figure Share of PCE basket above 3%, 12 months 54% Breadth, not a few outlier categories Share of PCE basket above 3%, 6 months annualised 49% Breadth has barely narrowed The Fed's target 2.0% Headline PCE, not core, not CPI

All figures as cited by Chairman Warsh in his Jackson Hole keynote, August 28, 2026. Source: Federal Reserve.

You have probably heard more about the Consumer Price Index than the Personal Consumption Expenditures price index. CPI is released earlier in the month and has been around longer, so it gets the coverage. But the Fed's 2% target is written against PCE.

The two are built differently. PCE covers a broader universe of spending, including consumption paid for on a household's behalf, such as the share of medical care funded by an employer or by Medicare. CPI is narrower and focuses on what households pay directly out of pocket.

They also weight things differently. PCE's expenditure weights shift more readily as spending patterns change, so it responds faster when people trade down. CPI updates its weights too, and its methodology does account for some substitution, but it works within a more fixed structure. Neither index is more accurate than the other. They answer slightly different questions, and the Fed chose the one with broader coverage.

Housing is the other significant gap. CPI gives shelter a much heavier weight than PCE does, which makes CPI swing harder when rents move quickly.

Core inflation strips out food and energy. Not because those prices do not matter to households, which they obviously do, but because energy moves on geopolitics the Fed cannot influence and food moves on weather and global supply chains. Removing both is meant to expose the trend that interest rates can actually reach.

July 2026 is a good illustration of why you should not assume one gauge tracks the other. Headline CPI came in at 3.4% year over year and core CPI at 2.5%, released August 12. Headline PCE was 3.7% and core PCE 3.3%, released August 26. In this particular month, CPI read cooler than PCE, not hotter. The two indexes have historically diverged in both directions, and the size of the gap changes from month to month.

Measure What it tracks July 2026, 12-month Released Headline PCE All consumer spending, broad coverage 3.7% August 26 (BEA) Core PCE PCE excluding food and energy 3.3% August 26 (BEA) Headline CPI Direct household out-of-pocket spending 3.4% August 12 (BLS) Core CPI CPI excluding food and energy 2.5% August 12 (BLS)

Sources: Bureau of Economic Analysis, Bureau of Labor Statistics. The Fed's 2% target refers to headline PCE.

When Warsh says inflation is too high, PCE is the number he means. Worth remembering the next time a CPI print moves markets and the Fed appears not to react.

Trimmed mean PCE, published by the Dallas Fed, discards the largest price increases and decreases each month and reads what is left. Over the 12 months ending in July it was 2.3 percent, against 3.7% headline and 3.3% core.

That is close to target, and it is the strongest argument for holding in September. There is a catch. An alternative version of the trimmed mean, calibrated over a longer sample, runs hotter and stood at 2.6% for the 12 months ended in June. Different calibrations diverge most when the distribution of price changes is skewed, which is the current situation.

Supercore services, core services excluding shelter, rose 0.3% in July. It is often described as a wage-inflation measure, which overstates what it does. It is an attempt to isolate the more persistent part of services inflation by removing the two components that move on something other than domestic demand.

The measures disagree with each other. That is why Warsh framed the bar as confidence in the trend rather than a single number clearing a threshold.

Measure July 2026 (12-month) What It Shows Why the Fed Watches It Headline PCE 3.7% Total consumer inflation The official target measure Core PCE 3.3% Inflation excluding food and energy Less volatile month to month Trimmed Mean PCE 2.3% Central trend, outliers removed Reveals underlying persistence Supercore PCE 3.9% Core services excluding shelter The more persistent services component Alternative Trimmed Mean ~2.6% (June) Longer-sample calibration May better capture recent skew

Sources: Bureau of Economic Analysis, Dallas Fed, KPMG. Different measures highlight different aspects of inflation and are read together.

Kevin Warsh's August 28, 2026 keynote was his first Jackson Hole address as chairman. Markets wanted his reaction function, the conditions that would trigger a move. They did not get one. If the machinery behind that decision is unfamiliar, how the Fed sets rates covers it from the beginning.

Titled "In Our Time," the speech read closer to a statement of governance than a monetary policy signal. On the outline of his approach: "You can call it an outline, you can call it a trail map, just don't call it forward guidance," a practice he said "has overstayed its welcome." Forward guidance, telegraphing future moves to markets, has been a fixture of Fed communication since 2008. Dropping it returns the Fed to a pre-crisis style in which markets are left to read the data themselves.

On inflation he was direct. "While this summer's [inflation] readings were better than expected, they do not tell me that underlying trends have meaningfully improved," he said. That sentence is what moved the odds. He called July PCE at 3.7% concerning, and described 2% PCE as the firm, fixed target.

Three other passages matter as much as the inflation lines, because they are what make a hike arguable rather than merely hawkish talk.

On the current stance of policy: "I would be hard pressed to describe broad financial conditions as restrictive," pointing to credit spreads near historical lows and bank lending standards on the easier end of their range. A Fed that does not believe current policy is holding the economy back has less reason to wait before tightening further.

On employment: "I believe the labor markets are consistent with full employment." The Fed has a dual mandate. When the employment half looks satisfied, the inflation half carries the weight.

On expectations, cutting the other way: medium-term inflation expectations "by and large, look stable" and "right now they are well anchored." That is the strongest reason in his own speech to be patient, and part of why September is a genuine coin flip.

He also drew a line about the Fed's role in markets. "We should not indulge a regime in which market participants are looking primarily to the Fed for their next trade," he said, reflecting a view that prices should be driven by fundamentals rather than central-bank signals. For background on why rate decisions are structured to sit apart from short-term political and market pressure, see Fed independence.

He did not rule out a September hike and did not commit to one. No thresholds, no timeline. Markets are left to do their own analysis, which appears to be the point.

There is an energy component to the 2026 inflation problem. A conflict involving Israel and Iran disrupted shipping through the Strait of Hormuz earlier in the year, and oil ran from around $80 a barrel toward $120 at the peak. We covered that chain in detail in oil prices during war.

Central banks typically look through temporary supply shocks by focusing on core inflation, which excludes energy. The reasoning is simple: the Fed cannot drill for oil or settle a conflict, and raising rates does not add a barrel of supply.

The limit is duration. If energy prices stay high long enough, they stop being a line item and start shaping expectations. Workers ask for wages that cover the pump. Businesses price in shipping. A shock becomes a trend.

Warsh's own read is that this has not happened: medium-term inflation expectations, he said at Jackson Hole, still look stable and well anchored. The concern is prospective rather than realised. A prolonged disruption to Middle East oil supply would raise the cost of goods and services that use it as an input, on top of the direct effect on headline inflation, and would keep the skew of price changes tilted upward.

A 25 basis point hike would move the federal funds rate to 3.75% to 4.00%.

The mechanics are the same in every hiking cycle. Higher rates raise the discount applied to future profits, which weighs most on companies whose earnings sit years away. Newly issued bonds pay more, which makes existing lower-coupon bonds worth less and gives income investors an alternative to dividend payers. For banks, higher rates can increase what they earn on loans, but whether profits improve depends on how quickly deposit costs follow, how fast assets reprice, and what the yield curve does. Higher short-term rates can squeeze a bank as easily as help it. We walk through why the textbook mechanics do not always show up in share prices in why stocks go up when interest rates are high, and how rates affect markets takes the same ground more slowly.

How markets take it depends less on the hike than on what comes with it. A move paired with language suggesting the tightening is finished reads very differently from the same move paired with a warning of more. And much of it is priced in already; what is already priced in does not move anything.

Separately from the market reaction, inflation at 3.7% is doing arithmetic to savings whatever the Fed decides. The inflation return calculator shows what a given rate does to a real return over time.

Three data releases land before the decision, and each one can change it.

Date Release Why it matters September 4 August employment situation (BLS) Warsh called the labor market consistent with full employment; visible weakening would undercut that September 10 Producer Price Index (BLS) Pipeline costs, an early read on where consumer prices head next September 11 Consumer Price Index (BLS) The last major inflation print before the meeting September 16 FOMC decision (Federal Reserve) Statement at 2:00pm ET, projections the same afternoon, press conference at 2:30

On the day itself, three things carry the information, and understanding Fed announcements walks through how to read each of them. The statement shows whether the target range moved. The dot plot shows where each official expects rates to end the year, which tells you more than the single decision does. And the press conference is where Warsh either repeats that he is committed to a discipline rather than a decision, or does not.

The underlying data is public and free. PCE inflation comes from the Bureau of Economic Analysis on its published monthly calendar rather than a fixed weekday: July's release was Wednesday, August 26, and the next is Wednesday, September 30, after the meeting. The Dallas Fed publishes trimmed mean PCE the same day. CME's FedWatch shows market-implied odds, which measure expectation rather than forecast. Roughly 56% means the market is close to evenly split, not that a hike is coming.

StockCram is not affiliated with any of these institutions. They are public data sources.