The Dots Nobody Trusted

Run the stratigraphy.

2003: "considerable period." 2009: "extended period." 2011: an actual calendar date — rates near zero "at least through mid-2013." 2012: the Evans rule, which swapped the calendar for numeric thresholds, 6.5 percent unemployment and 2.5 percent projected inflation. 2014: "patient," a word chosen specifically because it meant less than the thresholds it replaced. 2020: outcome-based guidance under flexible average inflation targeting, formally scrapped in the August 2025 framework revision, makeup strategy and all.

Six instruments. Each introduced to reduce uncertainty, each read by markets as firmer than a forecast, each quietly disowned when the world declined to cooperate, each replaced by a successor with a new name and the same structural defect. The dot plot is the seventh, and only the version that lasted longest: nineteen anonymous projections of the appropriate policy rate, plotted as a scatter since January 2012. Never a vote. Never binding. Not even a consensus — nineteen people's private guesses about their own future behavior, rendered precise enough to look like a commitment.

Markets traded it as a commitment anyway. That is not a misreading on their part. That is what happens when you hand a trading floor a number.

In December 2015, the median dot implied four rate increases across 2016. The Fed delivered one, in December, having spent the year discovering that the projection had been an aspiration wearing a decimal point. In December 2021, the dots implied three increases for 2022. The Fed delivered 425 basis points across seven meetings, off a chart that had projected a gentle stroll.

Off by fourfold one way and nearly sixfold the other, six years apart.

The easy conclusion is that the instrument was imprecise. The easy conclusion repeats the error it diagnoses. A conditional projection that moves when conditions move is not broken — that is the function. And the Fed did not just say so in the disclaimer: since 2017 the Summary of Economic Projections has carried fan charts built from twenty years of forecast errors, intervals wide enough to cover roughly seventy percent of outcomes. The Fed published its own margin of error, same page, same hour.

Nobody traded the fan chart.

So the defect is not in the forecasting. It is in the format. Nineteen discrete points, each at a specific decimal, read as nineteen decisions; a shaded band reads as a shrug. Offered both, the market took the dots and left the uncertainty on the page — which means no volume of footnote was ever going to fix it. Retiring the object is the only intervention that operates on the object.

That makes Warsh's move defensible, and the coverage has stopped there, at housekeeping. The part it has not reached is less comfortable.

Retiring the dots does not remove the forecast. It privatizes it. Whatever else the scatter was, it was a leveling instrument: a pension board in Ohio saw the same nineteen dots at the same minute as a primary dealer's rates desk. Withdraw it and the demand does not evaporate — it gets met by whoever can pay. Fed-whisperer newsletters, dealer research, speech-parsing NLP, former-governor consultancies at six figures a year. The signal survives. It acquires a price.

Which is the cut the commentary misses, because "markets want a number" treats markets as one actor. They are not. The question about any withdrawn public good is who can afford to rebuild it privately, and here the answer is the parties who already had the best seats. Epistemic honesty at the center becomes epistemic concentration at the edge. A reform that reduces distortion for the institution while raising it for everyone who cannot afford a Fed whisperer is not obviously a reform. It is a transfer.

And it will not hold anyway, because the binding constraint was never the Fed's willingness to be honest. It is the demand. Markets want a number, politicians want a commitment they can attack or defend, financial media want a headline that fits above the fold. All three will read the absence of a dot plot as a vacuum, and vacuums in this business do not stay empty. Something will fill it — a scenario matrix, a policy-rule reference range, the fan chart promoted out of the appendix, some construct with a technically impeccable disclaimer bolted to the front.

And then it will be traded as a promise.

That is the loop. Not because central bankers are dishonest, but because an institution asked daily to be certain, and deriving its power from being believed, will keep generating certainty-shaped objects. Retiring the dots removes one. It does not touch the appetite that produced it — it only changes who gets billed for feeding it.

So: the reform is correct on its own terms, and I would give the replacement about eighteen months before somebody on a trading desk starts calling it "the new dots."

I would like to be wrong. The stratigraphy suggests I will not be.