Three regional Fed presidents voted to hike against a quarter annualizing at 0.49%

Decompose the 3.4% ^[1]. If the three months since May annualize to 0.49% ^[8], they added roughly 0.12% to the trailing twelve, which leaves about 3.28% to come from the nine months before them, a pace near 4.4% annualized ^[2]. That older stretch is what the trailing window is actually reporting, no forecast required, and it leaves only as nine more monthly prints push it out one at a time ^[3]. A committee keyed to that measure has a hawkish case available to it for most of a year regardless of what the current run rate does.

Which is why the repricing looks brave. Consensus on a September increase went from 80% yes a month ago to about 67% no ^[13], an implied hike probability falling from 80 to roughly 33, a 47 point swing ^[6], and the S&P 500 set a record the day after the release ^[12]. The three-month number carrying that move is thin: with the last quarter totalling about 0.12%, one 0.40% monthly print in place of a 0.04% one puts the three-month annualized rate back near 1.9% ^[4]. The Cleveland Fed's own annualized-quarterly nowcast sits at 1.05% ^[11], about 2.1 times the 0.49% figure ^[8], so two short-window measures of the same thing already disagree by a factor of two while both stay under target.

The shorter window is not a fringe construction. Krugman has said a year is too long a lag in a turbulent economy and that many economists now use three- or six-month changes ^[14], and Furman has made the same argument for a three-to-six-month view ^[15]. The producer side reads the same way and worse: 4.7% year over year [c9a] against 1.6% on three months [c9b], a 3.1 point gap ^[5], with negative monthly readings in June and July [c9b] and pass-through to consumer prices that varies by industry ^[17].

This is probably wrong, but the asymmetry favours being positioned for one more hike than the strip carries, driven by a reaction function that reads a window mechanically unable to show the recent quarter ^[3], regardless of whether prices are hot, while Hammack talks about broadening pressure ^[5], Kashkari projects multiple hikes ^[6] and Warsh has promised delivery of 2% ^[7]. The counter-thesis is arithmetic too: expectations from both the five-year breakeven and the Cleveland one-year model sit around 2.3% ^[10], which is 1.1 points below the headline and 0.3 above target ^[7], and a committee weighting those plus the nowcast leaves the dissenters a minority and duration cheap at these levels.

What breaks the tighter-for-longer read is two consecutive monthly prints near 0.1% paired with a September hold, or a headline that falls through 3% while the three-month stays under 1%; at that point the trailing window stops supplying cover and the hawks are arguing from conviction rather than from data that already happened.