Kevin Warsh arrives at the Eccles Building with the most ambitious intellectual agenda of any Fed chair in a generation: a smaller balance sheet, less forward guidance, and an explicitly monetarist account of inflation that locates its cause in fiscal excess rather than consumer demand. Markets, told to expect a steepener as a politicised Fed cut into sticky inflation, did the opposite. The curve flattened. That reaction is the most important signal in the cluster, and it deserves more weight than the confirmation theatre around it.
Warsh's framework is coherent in a way the consensus reading has underplayed. He treats the 2020 FAIT revision as the proximate cause of the inflation surge, views 2021-22 as a legacy of policy errors, and describes quantitative easing as fiscal policy in disguise that lifts asset prices for the wealthier half of households. None of this is rhetorical. It implies a reaction function in which balance-sheet runoff is the primary tightening lever, the policy rate is a secondary instrument, and forward guidance — the signature tool of the Bernanke-Yellen-Powell era — is downgraded to noise. If he executes even half of it, the transmission mechanism the Street has modelled for fifteen years no longer applies.
“We need to fundamentally rethink macro, which is a fundamental rethink of the core economic models that the Fed is using—rethink what is the core theory of inflation that the Fed is using, which I think is mistaken.”
— Kevin Warsh
Why the curve flattened
The expected trade was straightforward: a Trump-aligned chair cuts the front end, term premium widens at the long end, the curve steepens. It did not happen. One reading, advanced in One Basis Point, is that traders took Warsh's monetarist signalling seriously and concluded that balance-sheet shrinkage offsets any rate-cut impulse — quantitative tightening as a counterweight to political pressure on the policy rate. Brad DeLong's reading is complementary: TIPS-implied inflation expectations sit near 2.5 percent and real borrowing costs remain below growth, so the bond market is not pricing a regime break at all. Both can be true. Neither supports the cartoon of a captured Fed.
The operational questions worth pricing are narrower than the political ones. Does the System Open Market Account portfolio shrink faster than the current runoff cap implies by mid-2026? Does the post-meeting statement drop or materially weaken its forward-guidance language within two meetings? Does the dot plot survive at all? Each is binary, observable, and resolvable. The Quoth the Raven framing — whether aggressive cuts or a restart of purchases under Warsh materially lift inflation within two years — is the wrong bet, because Warsh's stated framework treats restarting purchases as the cardinal sin. The live risk is the opposite: a chair who tightens through the balance sheet while cutting the funds rate, and a market that cannot decide which instrument to weight.
The live risk is not a captured Fed cutting into inflation. It is a Fed tightening and easing simultaneously, through different instruments.
The political overlay matters but is being overpriced relative to the framework. The Cook litigation, the dropped Powell investigation, and Warsh's refusal at his hearing to rebut the stolen-election claim are real reputational costs to the institution. They are not, by themselves, monetary policy. The Economist's judgement that Warsh could rank among the consequential American central bankers if he withstands Trump is the correct frame: the test is not whether he cuts, but whether the cuts come packaged with a balance sheet small enough, and a communications regime quiet enough, to make the cuts non-inflationary. That is the trade to monitor, and it is not yet in consensus pricing.
Briefings are synthesised by the Ledger Desk from multiple sources cited in the sidebar. They are distinct from Articles, which are written by named contributors and carry a tracked Calibration Index. The Desk does not currently carry a Brier score; this is a deliberate choice for the v0.1 editorial layer and will be revisited.
