The Fed Chair Spoke and Rate-Hike Odds Jumped From 35% to 57%

The Dow Jones barely moved in New York on Friday: it closed at 53,559, down 0.02% on the day. In the same session the Russell 2000, which tracks smaller companies, fell 1.39% to 2,972. The S&P 500 lost 0.25% and the Nasdaq 0.52%.

When one index sits still while another drops nearly a percent and a half, the day's news was not in the equity market. It was in the bond market.

Fed Chair Kevin Warsh spoke at the annual symposium in Jackson Hole. After the speech, the probability investors assigned to a rate hike at the September meeting rose from 35% to 57%. Accounts do not line up exactly on those two figures; some put the pre-speech level at 33% and the post-speech level at "a little above fifty."

For all of 2026 the argument in the market had been about when the Fed would cut. On Friday the same market started pricing a hike.

What Was Said, and What Was Not

There is only one question worth asking when reading a central bank chair's speech: is this a preference, an intention, or an action?

Warsh said the committee has to be confident that underlying inflation is moving to the target clearly and at sufficient speed, and then added:

"Otherwise, we have work to do."

That is a preference, stated with a condition attached. Not "we are raising rates," not even "we will raise rates." One detail from the speech: more than half of the goods and services the Fed tracks saw price increases above 3% over the past year.

Is there an instrument that makes this preference binding? No. Only an FOMC vote sets the policy rate, and there was no vote on Friday. The target range has stood at 3.50%-3.75% since July 29.

The scope of that instrument matters too. A Fed chair does not set rates alone; he holds one of twelve votes. On July 29 the committee held rates by a vote of nine to three — and those three dissents were in favor of a hike, not a cut. What Warsh did on Friday was not decide anything. It was to show which side of an existing three-member minority he stands on.

July: The Same Market Was Pricing the Opposite

Three weeks ago the picture was entirely different.

The July employment report, released on August 7, showed the U.S. economy lost 23,000 jobs. Forecasts had ranged from a gain of 83,000 to 95,000. June's initially reported gain of 20,000 was revised down. The monthly average over the previous twelve months was a gain of 34,000. The unemployment rate fell to 4.1%; when payrolls shrink and unemployment falls at the same time, the labor force is contracting too.

Odds of a September hike dropped sharply that day. The logic was simple: you do not raise rates on an economy shedding jobs.

On Friday Warsh rejected that logic. He said the inflation data were more concerning than the labor market data, where the unemployment rate remains low, and that prices are the Fed's predominant focus right now, with short-term interest rates its predominant tool.

The Mechanism: 57% Is Not a Survey, It Is a Division

Reading the sentence "the market puts the odds of a hike at 57 percent" brings a poll to mind. There is no poll. The figure is backed out of the price of federal funds futures, and knowing how it is derived also explains why it moves so much.

This does not make the number meaningless. But the difference between "the market expects a hike" and "the market is undecided" is often a few basis points of price. The relationship between rates and bonds works directly here.

This Time the Curve Broke at the Short End

The most instructive part of the move was which maturity did the work.

The two-year yield rose into the 4.30%-4.35% band, the ten-year reached 4.728%, and the thirty-year stayed at 5.21%. Accounts differ on the two-year move, ranging from 8 to 12 basis points; all of them agree the short end moved most.

To see why that matters, look at the two episodes in July and mid-August. In both, the move was at the long end: the Fed was not moving and the 30-year yield was climbing. The sentence that described that period was: the central bank sets the short end, the market sets the long end.

On Friday that reversed. The short end moved four to six times as much as the long end. Short-term yields are set largely by expectations of what the Fed will do; long-term yields reflect the deficit, bond supply and an uncertainty premium — things the Fed does not control directly. A move concentrated at the short end means the market no longer thinks something is happening beyond the Fed's control. It thinks the Fed is going to act.

The distance between the two ends narrowed as well: the gap between the thirty-year and the two-year closed by 6 to 10 basis points in a single day.

Why Small Companies Fell Hardest

The Dow going nowhere while the Russell 2000 lost 1.39% was not a coincidence.

Large companies finance themselves mostly by issuing fixed-rate bonds. Those bonds were sold years ago, most of them in a lower-rate environment; the coupons do not change if the Fed raises rates today. Smaller companies borrow mostly from banks at floating rates: when the policy rate rises, their interest bill rises in the next payment period.

Then there is profitability. According to an analysis cited by Morningstar, roughly 40% of Russell 2000 companies report no earnings at all; two decades ago that share was 17%. For a company that does not earn anything, the cost of debt is a question of survival.

Other rate-sensitive groups fell in the same session: utilities, telecoms and real estate investment trusts. Gold dropped 2.88% to $4,529.90 an ounce. What held the indices up was large-cap technology.

The charts below are live; they show not Friday itself but where these two indices stand today.

The large-cap side over the same window, for comparison. How far the two charts have diverged over the past month shows whether Friday's single-session gap stood alone or has become a pattern.

Two Readings

What to Watch: Two Countable Things

A calendar is more useful here than a forecast.

September 4: The August employment report. This is where July's 23,000-job loss is revealed as either a one-month deviation or a trend. Why employment data matters this much: it is one of the Fed's two mandates, and it currently conflicts with the other one, price stability.

September 15-16: The FOMC meeting, which will also publish the projections showing where members expect rates to go. The thing to count is simple: if four more names join the three who voted for a hike in July, the vote changes direction.

There is also a note that follows from the speech itself. On communication, Warsh said:

"A quieter Fed... is better able to meet its objectives."

A single speech by a chair who says he is not offering forward guidance produced a 22-point shift in probability in one day. The cost of silence is that words weigh more.

This piece is based on reporting from Bloomberg, PBS, NPR, CNBC, Yahoo Finance and Quartz as of the August 28, 2026 close, and on the Federal Reserve's published meeting calendar. Figures such as the hike probability and the change in the two-year yield are given as ranges because sources differ. The futures prices are representative values calculated from the reported probabilities using the standard method.