The Slate

Kevin Warsh arrived at the Federal Reserve chairmanship in May 2026 carrying a doctrine he had spent fifteen years assembling. Its root is a diagnosis: American inflation is not an act of pandemics, oil markets or foreign wars — it is "a choice" the central bank itself makes, because "the central bank can hit any price level that it wants" — and the institution making that choice has swollen beyond its constitution, behaving, as he put it at the IMF in April 2025, "more as a general-purpose agency of government than a narrow central bank."

On July 9, shortly after 3pm, the Fed announced the instrument through which that doctrine is to become institutional fact: five task forces, co-led by fifteen outside experts, charged with examining the central bank's methods and reporting by the end of the year.

The release promises that the panels will "operate independently, with a mandate to follow the evidence." Whether they will find anything is another question, because the fifteen arrive with published records on precisely the matters they are asked to reopen. These are testable claims, and the test must begin where the announcement did not: with the doctrine itself.

Five Planks

The doctrine is not one belief but five, each documented in Warsh's own words across the decade.

First, humility over false precision. A central bank should stop publishing what it cannot know. In practice: retire forward guidance — the Fed's public promises about the future path of interest rates — and the point paths of the dot plot, its quarterly grid of anonymous rate projections; replace point forecasts with scenario ranges, above all for anything as uncertain as artificial intelligence; demote the Phillips curve and NAIRU — the models that read future inflation off today's unemployment — from operational guides to background; and size the balance sheet off market signals rather than fixed targets. The conviction is documented at both ends of his career: his 2014 review of the Bank of England argued that too much transparency degrades judgment — "creating a safe space for true deliberations is among the most critical indicia of organisations that make good decisions" — and at his first meeting as chairman he refused to file his own rate projection, calling it "not helpful in the conduct of policy."

Second, a smaller, less activist footprint. A leaner portfolio of Treasuries only, the mortgage bonds — mortgage-backed securities, MBS in the trade's shorthand — run off; bond-buying, known as quantitative easing or QE, reclassified as emergency medicine rather than a standing tool; a light footprint even in data, partnering with private sources rather than building an estate of its own; in short, a Fed that stays in its lane. He resigned from the Board of Governors in 2011 rather than support a second round of bond-buying, has said the institution "treated every day for more than a decade like it's a crisis," and called the result, in a 2024 opinion piece, a "bloated" balance sheet.

Third, credibility and independence as the paramount assets. The policy statement — not speeches, not leaks — carries the signal; members stand accountable for their own votes; the 2 per cent inflation target is held symmetrically and credibly, misses above treated like misses below; and the Fed defends itself explicitly against fiscal dominance — the pressure to keep rates low because the government's debts are expensive. The premise is the diagnosis in his own words: if "the central bank can hit any price level that it wants," responsibility cannot be outsourced, and "independence has to be earned… by delivering on the promises."

Fourth, rules-leaning, but not mechanical. A clear reaction function and credible commitment — the opposite of "the full discretion of a central banker's latest whims" — yet short of a hard formula — a Taylor rule that sets the interest rate by arithmetic, or a target for total nominal spending, NGDP; and the Fed's 2 per cent number itself ring-fenced from the review: "I tend to focus on the left of the decimal point."

Fifth, a supply-side, disinflationary worldview. Productivity — today, artificial intelligence — raises the economy's potential output and its neutral rate — written r* in the profession's notation, the interest rate the economy can bear without slowing — and is itself disinflationary: the licence for patience and restraint that high rates would otherwise have to buy. "Strong productivity-led growth is not something that we fear, but something we embrace."

Humility over false precision; a smaller, less-activist footprint; credibility and independence; rules-leaning but not mechanical; a supply-side, disinflationary worldview. One program — "restoration, not revolution," in his own framing — and, since July 9, five committees to carry it, due by year-end. Hold the five planks in mind. They are about to reappear, one at a time, wearing institutional clothes.

Five Charges

The five task forces announced on July 9 are the doctrine drawn as an organizational chart — each panel organized around one dominant plank, with the supporting planks recurring across the others. Communications, for the promises he wants stopped, will "review how the Federal Reserve conveys policy deliberations and decisions amid uncertainty": humility over false precision, made institutional. Balance Sheet Policy, for the portfolio he wants shrunk, will examine "the costs, benefits, and institutional implications of the current balance sheet regime": the smaller footprint. Data, for the gauge he distrusts, must make the Fed's signals "more accurate, relevant, contemporaneous… actionable": the evidence a humbler and more credible Fed would act on. Inflation Frameworks, for the 2020 regime he blames, will "revisit how the Federal Reserve understands and responds to the drivers of inflation": credibility, and rules without mechanics. And Productivity and Jobs, for the supply-side bet that must pay for all of it, will "assess the economic impact of new general-purpose technologies, including artificial intelligence, to inform the Federal Reserve's policy judgments."

Four of the five charges revisit complaints Warsh has pressed for a decade — the over-communication, the footprint, the slow gauge, the 2020 framework. The fifth places the wager. And the wording of the charges does quiet work: a panel charged with weighing the costs of the current regime deliberates on the chairman's ground, because whether current practice is the problem has been settled by the charge before the first meeting convenes. This, too, has a precedent in Warsh's own career — the reviewer of the Bank of England in 2014 understood that the terms of a review decide most of its findings. One coherent program, distributed across five committees.

The terrain, then, belongs to the chairman. The findings belong to people — and the fifteen names are where the announcement stops describing a search.

The Fifteen

Charges set the terrain; the fifteen names decide what grows on it. Panel by panel, the roster reads as the doctrine pre-staffed — each directional coalition chosen for the course it had already put in print, each counterweight seated where its signature would be needed. What follows walks the five rooms: who sits, what they have said, why the chairman wants them there, and what each panel will most probably report.

First, the whole arrangement at a glance — the chairman at the centre, the five panels around him, the fifteen names on the rim. The colouring here uses one broad measure: how close each member's decade of published work sits to the chairman's overall program, as distinct from their position on their own panel's question, a second reading that will matter shortly. One feature deserves registering before the walk begins: across the entire map, exactly one name draws the red of outright tension — Jeremy Stein.

Fifteen Dots

Before the argument moves, the reader should see what they have just read — one strip, fifteen dots, the pattern visible at a glance.

Read the last dot in each row, top to bottom: the opponents sit where something is being torn down — the dot plot, the bond portfolio, the 2020 framework — because demolition needs their signatures for credibility; no opponent sits where something is being built — the new gauge, the supply finding — because those panels produce the deliverables the December meeting is waiting for. The colours here are each member's published position on their own panel's question. The map that opened the walk coloured the same fifteen by the other measure — distance from the chairman's whole program — and the two readings disagree on more than half the names. Fraga, Dynan and White, red here as opponents of their panels' reforms, are amber there, fellow-travellers of the wider program; Murphy, gold here as a brake on degree, is green there, the chairman's closest ally; the four builders — Chetty, McMillon, Andreessen, Sharma — are blue here and amber there, friends of their assignments rather than of the whole doctrine. Read twice, the disagreement is the design: Warsh seated opponents of his reforms, and almost none of his program — of the fifteen, only Stein is red on both readings. It is also why the panel of three optimists shows not one green dot on the map: two of them outrun the chairman's program, and the third trails it deliberately. The two readings set up the questions that remain: what the seated opponents can and cannot do, and why the panel that licenses December is the one built without brakes.

Two patterns organise the roster. The first: the seated opponents sit where something is being torn down — the dot plot, the bond portfolio, the 2020 framework — because demolition needs their signatures for credibility; no opponent sits where something is being built — the new gauge, the supply finding — because those panels produce the deliverables the December meeting is waiting for. The second: measured against the chairman's whole program rather than any single panel's question, almost no one on the roster is an opponent at all — of the fifteen, only Jeremy Stein stands in outright tension with the program itself. Warsh seated opponents of his reforms, and almost none of his program. The two patterns set up the questions that remain: what the seated opponents can and cannot do, and why the panel that licenses December is the one built without brakes.

The Function of a Dissent

The strongest objection to everything above is sitting in the rosters. A chairman engineering conclusions would not, on the face of it, hand seats to Stein, Dynan, White and Fraga — the four most credible opponents of the conclusions in question. Their presence is the release's best evidence for its own word, "independently," and the objection deserves to be taken at full strength.

Consider, then, what each opponent can and cannot do. Stein and Dynan hold two of three seats on the balance-sheet panel; they have the standing, and the memory of September 2019 — when reserves ran short and overnight lending rates spiked — to block the return to a small, pre-2008 balance sheet that Warsh has publicly favoured. What they cannot do, against their own co-lead and the charge above them, is produce a report that endorses the current regime; both are on record for gradual shrinkage of some degree. The likeliest outcome, and the one contested number in the whole review, is a portfolio that comes down without becoming scarce: the doctrine's one defeat. White can force the sharpest objection in the five panels into the record; Sargent and Mankiw still define the report's central framework. Fraga can shape what replaces the dot plot — an honest account of uncertainty rather than silence — without saving the dot plot itself.

Each room, restated — members, probable outcome, the swing vote, and the dissent most likely to surface in print. One term the cards use has not yet been unpacked: a penalty-priced backstop is emergency central-bank lending kept permanently on offer, but at a rate that stings — available in a crisis, unattractive on any ordinary day.

The opponents, in short, bound the reforms; they do not reverse them. And a bounded reform, signed over the recorded objection of its most credible critic, arrives at the Federal Open Market Committee carrying the one property a chairman cannot confer on his own conclusions: the appearance — and, within the bounds, the substance — of having survived resistance. The panels, on this reading, were not built to rubber-stamp the doctrine; they were built to return the strongest version of it that can survive expert opposition — a narrower thing than the chairman's wish list, and a more durable one. The defeat on the balance sheet is not a flaw in the design; it is the design's proof of authenticity, purchased at the price of the one reform Warsh wanted most. Scott Clemons, chief investment strategist at Brown Brothers Harriman, compressed the arrangement into seven words in June: "regime change, but in a velvet glove."

Which leaves the panel with no glove and no fist to resist it — the one that matters most. Why it matters most — and why its deadline is December — is the business of the machinery's first application.

The Gauge

The machinery's first application begins with a number. Core inflation — prices with volatile food and energy stripped out — was still running near 3 per cent when the FOMC met on June 17: a full point above target, with the Fed's policy rate at 3.5 to 3.75 per cent. The doctrine's endpoint — patience, restraint, and in time lower rates on the strength of supply — runs straight into that reading. Treat the gauge as a sufficient guide, the way the Fed has treated it for decades, and a near-3-per-cent print makes the cut Warsh wants difficult to justify under the inherited framework. His answer, delivered from the podium that afternoon, was to dispute the instrument rather than the reading.

The distinction carries the whole argument. The gauge counts the demand side of the economy — spending, and the prices it produces. It cannot say how much the supply side is expanding beneath it, and it is supply, in the doctrine, that decides whether today's inflation persists or dissolves. A central bank that treats the counted number as its prescription will hold rates too high while a productivity boom quietly does the work. The fork this opens is the hinge of the whole design: if the gauge is a sufficient guide, the cut waits on measured inflation returning to target; if it is not, the cut requires the committee to accept a broader reading of supply — one in which rising productive capacity makes today's inflation less persistent than the gauge alone suggests — and the task forces become the machinery for changing the decision rule built around the gauge. Warsh had flagged the consequence at his confirmation hearing in April — an interest in "improving the data used to measure inflation" and in "newer methodologies" — and had already tied the repair to his review: measuring inflation, he said, "will overlap with my data group."

His appointees supply the indictment's support from both ends of the measurement problem: Greg Mankiw, assigned to the inflation panel, has written that "disentangling shocks to supply and demand is nearly impossible in real time," and Doug McMillon, assigned to the data panel, watched Walmart's shelves report consumer stress weeks ahead of any official print. (The gauge has been contested ground before; as two earlier articles in this series, The Measure and The Paradox, documented, the Fed's own choice of inflation index has at times been the policy rather than the measurement of it.)

An inference, however, is not a licence. If policy is to act on supply the Fed can only infer, some body must put the inference in writing. That is a framework problem, and Warsh solved it in June.

The Framework of June

A doctrine cannot outvote a gauge. What it can do is dethrone the gauge — change what the committee decides on — and that is what the June 17 meeting accomplished, as the previous article in this series, The Inference, documented. The committee held rates and changed everything around them. Forward guidance was dropped from the statement as "not well suited to the current policy conjuncture." Warsh withheld his own rate projection — the refusal the first plank promised — and observed that his colleagues filled in theirs "with pencils… with the big erasers." The statement itself was cut to the facts. (An earlier article, The Game, had anticipated the method — a framework built instead of a coalition — and recorded the archive's caution about five prior attempts at framework-first policy.)

The committee beneath the unanimous vote was split evenly. "Half of my colleagues thought the policy rate… should be at this level or lower between now and year-end, and the other half thought higher," Warsh told reporters — nine members projected an increase. The disagreement was not resolved; it was scheduled. The five task forces were announced at that meeting, their reports due by year-end, and the year's final rate-setting meeting falls in December. Asked how the committee would know when expanding supply justified easier policy, Warsh answered: "we have a task force for that."

The Inference drew the conclusion in June: the December cut would be licensed not by the inflation data, which points the wrong way, but by a supply-side finding timed to arrive on the eve of the meeting. In June that was an argument about machinery still unbuilt. The machinery has now been walked, room by room — which leaves only the room the finding must come from.

The Missing Brake

The productivity panel is the review's keystone, because its finding is the input the December framework is waiting for. It is also the only panel constructed without an opponent. Andreessen's optimism is doctrinal; Sharma's is commercial; Jones's is calibrated — and the calibration is the point. The one dissent the room risks runs the other way: Andreessen declining to sign a document he finds too hedged — an objection from the accelerator, not the brake. Jones is a serious scholar whose own research counsels patience about the very thing he is asked to certify: his work with co-authors found that sustaining Moore's Law — the steady doubling of computing power — now requires more than eighteen times the researchers it took in the early 1970s, and that general-purpose technologies historically take decades to surface in productivity statistics. Nothing measurable between July and December can establish that artificial intelligence has raised the economy's speed limit. The panel can assemble adoption figures — the $254bn of AI infrastructure spending in 2025 is counted demand, visible in any national account — but the supply-side payoff it is charged to assess will still be an inference.

The Fed has faced this exact question before, and answered it differently. In the late 1990s, as The Inference set out, the supply-side minority — Robert McTeer of the Dallas Fed most prominently — argued that the technology boom was disinflationary, and prevailed only as years of data accumulated; Alan Greenspan conceded at a 2000 meeting that his staff had been "continuously forecasting decline in multifactor productivity growth… that slowdown hasn't happened." The vindication was earned at the pace of the evidence. The current version is scheduled: the report is due by year-end, and the year's last meeting is in December. A finding produced by three optimists on a deadline, however carefully Jones hedges its language, will function as what the framework requires it to be — a licence.

That is the whole of the mechanism, assembled: a five-plank doctrine, five charges on the doctrine's ground, directional coalitions that set their courses in print years ago, constraints positioned where credibility is needed and none where the deliverable is made, and a framework that dethrones the gauge and waits, in December, for the deliverable.

Checkable by New Year

The reading above is falsifiable, and on a short clock. If it is right: the communications report will retire or gut the dot plot, downgrade guidance and move the burden of signalling to the statement itself, with Fraga's hand visible in whatever replaces the projections — or in a published objection. The balance-sheet report will recommend further shrinkage, running the mortgage bonds off toward a Treasuries-only, shorter book, but stop short of the scarce-reserves regime Warsh has favoured — Stein and Dynan's boundary holding. The data report will stand a real-time gauge, built on private data, beside the official statistics rather than in place of them. The inflation report will scrap FAIT for a symmetric 2 per cent stated with humility, demote the Phillips curve, revive the monetary aggregates, and acknowledge White's financial-cycle critique without adopting it — White being the likeliest of the fifteen to dissent in print. The productivity report will hedge its evidence in scenario language, decline to claim a measured boom, and be cited in December all the same as grounds for the cut. And the committee that split evenly in June will deliver that cut at the December meeting.

The predictions are not equally firm. The deaths of the dot plot and of FAIT are the firmest, because no one in either room defends them — the contests there are over what replaces them; the balance-sheet boundary rests on a divided room whose internal median sits with the defenders of ample reserves; and the December cut — hostage to a committee that divided evenly in June — is the least certain of the chain and the most consequential.

A documented stance-and-influence audit — every member scored on their panel's contested question, every room averaged with the chairman's pull included — produces the same rankings and the same order of confidence; the tables sit in an appendix after the conclusion, for readers who want the arithmetic.

A documented stance-and-influence audit — every member scored on their panel's contested question, every room averaged with the chairman's pull included — produces the same rankings and the same order of confidence; the full tables are published with the written article.

Each prediction can fail, and any failure would count against the argument. If a panel surprises its charge — if the balance-sheet report demands scarcity, if the productivity panel declines to certify, if December passes without the cut — this reading was wrong, and the next post will say so. The record of framework-first projects cautions in both directions: as The Game documented from five historical precedents, conclusions installed ahead of their evidence have tended to end badly for the institutions that installed them — a warning that applies to the doctrine, and equally to any confident reading of it, including this one.

The Slate

Collapse the five probable reports into what recurs across them, and the residue prints on five cards. They are the planks of the doctrine, returned as findings.

Collapse the five probable reports into what recurs across them, and the residue is the doctrine itself: humility over false precision; the smaller footprint; credibility and independence; rules short of mechanics; the supply-side worldview — the planks, returned as findings.