The Reaction Function Has Broken
Bill Dudley said the quiet part on Bloomberg Television yesterday. The former New York Fed president, choosing his vocabulary carefully, warned that the central bank is "in danger of losing credibility" after running inflation above its 2% target for more than five years. The statement was reported as news. It is not. Five years above target is the diagnosis everyone already has. The interesting question is why the market chose this week to begin pricing it.
The answer is in the calendar. Kevin Warsh has been Chair of the Federal Reserve for twelve days. The Senate confirmed him 54–45 on May 13 — the most divisive Fed confirmation vote in the institution's history. Two weeks earlier, Jerome Powell presided over his final FOMC meeting and recorded four dissents on the policy statement — Governor Stephen Miran voting for a 25bp cut, and Presidents Hammack, Kashkari, and Logan opposing the inclusion of an easing-bias clause. It was the largest dissent count since November 1992. Powell had previously recorded the lowest dissent rate of any chair in 48 years. He inherited his successor a Fed where almost nobody disagreed; he handed one over where the disagreement is structural.
The macro print does not help. Atlanta Fed's sticky-price CPI ran 4.6% annualized in April; the flexible component ran 19.3%, courtesy of a Strait of Hormuz that has been under US blockade since the Iran conflict escalated in February — with negotiations now underway but the blockade still in force pending agreement. Headline inflation just printed 3.8%, the highest since May 2023. Five years of target miss did not break the Fed. What is breaking it now is the arrival of a chair whose mandate is the political opposite of what the data requires, presiding over a committee that has just spent its final pre-transition meeting making clear it will not be moved on language, let alone on rates.
Credibility is not the belief that the Fed will hit 2%. It is the belief that, given a data print, the policy response is knowable. The first can fail for a decade without breaking markets. The second cannot fail for a quarter. — ZTrader.AI Research
What the Market Is Actually Pricing
The conventional read of "Fed credibility" looks at long-dated inflation breakevens and concludes that nothing dramatic has happened. The five-year breakeven sits at 2.59% as of mid-May — elevated, but not unanchored. The ten-year sits in a similar zone. By this metric, expectations remain "anchored." This metric is asking the wrong question.
A breakeven is the market's expectation of average realized inflation. It says nothing about the dispersion of paths to that average — and the entire credibility apparatus of a central bank lives in the dispersion. When traders priced Powell-era policy, the variance around the breakeven collapsed because the reaction function was treated as a noisy but stationary kernel: given CPI x and unemployment y, the next move fell within a narrow distribution. That kernel is now bimodal. The breakeven is no longer a forecast; it is a midpoint between two regimes — a Warsh-led easing cycle that the chair has publicly endorsed but cannot deliver, and a hawkish-dissent committee that has already telegraphed its veto. The level is the same. The information content is not.
II.Two Mistakes Hiding Inside One Word
The discourse around "Fed credibility" conflates two losses that should be analyzed separately. The first is the loss of target credibility — the belief that the Fed can return inflation to 2% on some reasonable horizon. This is what Dudley referred to, and it has been deteriorating for years. Markets adapt to it. They reprice term premia, they buy gold, they shorten duration. It is a slow leak.
The second is the loss of reaction-function credibility — the belief that, conditional on a given data realization, the policy response is a deterministic function the market can solve. This is what is breaking now. Warsh has arrived with a publicly stated framework — productivity-led easing justified by AI disinflation — that is structurally inconsistent with the data the committee is staring at. He has indicated he will not abandon it. The committee has just signaled, in formal vote tallies, that it will not be moved.
The result is that the reaction function has acquired what options traders would call regime variance. There are now two FOMCs nested inside one institution: the one Warsh wants to chair, and the one that will vote against him. The market response is not to revise its inflation forecast. It is to widen the confidence interval on every policy path. Forward guidance — the lowest-cost tool in the Fed's arsenal — has had its multiplier compressed toward one. A press conference that does not bind a vote does not move a market.
Target failure is a leak. Reaction-function failure is a phase transition.
The first costs you basis points. The second costs you the option value of the entire forward curve. — Z Note · 27 May
III.The Asymmetric Reprice
Notice what is and is not happening in the cross-asset tape. Gold corrected from its all-time high of $5,414 on January 28 to roughly $4,490 today — down 17% — which on the surface looks like a vote of confidence in dollar institutions. It is not. The pullback was mechanical: a stronger dollar after the Iran shock, forced liquidations into a global risk-off, and Treasury yields drifting higher on supply concerns.
The structural bid is intact. Q1 official-sector net buying was 244 tonnes, up 3% year-over-year. Gold's share of global official reserves crossed 30% in late 2025 and has widened its lead over the euro. The PBoC has now extended its buying streak to 17 consecutive months. On a year-over-year basis, gold remains up roughly 36%. Central banks are not selling. They are tactically pausing.
The Treasury market tells the same story in a different dialect. The 2s10s spread reached its widest level since January 2022 in early February and has held a steepening bias through the Warsh transition. Bear-steepening into a Fed that the futures market believes will not cut at all this year — sub-3% implied probability of any cut at any remaining 2026 meeting, with Bank of America's first-cut call now pushed into the second half of 2027 — is not what markets do when they trust the chair to engineer a soft landing. It is what markets do when they conclude that the policy function has become uninterpretable and therefore must be priced as wide.
IV. The Honest Question Nobody Is Asking
The standard counter to this thesis is that Fed credibility has survived bigger insults than this. Burns survived Nixon. Volcker survived political fury. The institution endures. Anthropomorphizing the Fed as a single agent with a single reputation is exactly the mistake. The Fed survives. The pricing of its policy does not. Volcker's pain trade worked precisely because his reaction function was visible from orbit: rates would not come down until inflation did, regardless of what the political layer demanded. Markets repriced violently but coherently around a known kernel.
The deeper counter is that every Fed transition produces reaction-function uncertainty: Greenspan to Bernanke, Bernanke to Yellen, Yellen to Powell. Markets always reprice during the handoff. What makes this handoff different is not the transition itself — it is the structural inconsistency between the chair's stated framework and the committee's voting math. Prior chairs arrived broadly aligned with the committee they inherited. Warsh has not. The chair is one vote of twelve, and no Fed chair has ever lost an FOMC vote on monetary policy. The reason Warsh will not become the first is that he will not bring policy preferences to a vote he cannot win. The path that resolves both ends of the constraint is the path he is on: do as little as possible while appearing to be in charge. This is not credibility loss. It is credibility evaporation through optionality compression. The committee retains every option; the chair retains none.
A chair without a credible policy path becomes a press-conference asset, not a monetary one. The forward curve has noticed. The breakevens haven't yet.
V. Position Implications
The first-order trade is variance, not direction. Implied vol across the front end of SOFR is biased to stay rich — not because the Fed will move, but because the dispersion of conditional paths around any given data print has widened. Receivers of front-end vol have been underwater for a year; the regime that justified that trade may be ending. The cleanest expression is to be long convexity in 3m–1y SOFR options rather than picking a directional rate view.
The second-order trade is in the curve. Bear-steepening into a hawkish FOMC and a dovish chair is the textbook expression of reaction-function uncertainty: the back end prices the inflation risk the chair will not fight, the front end prices the political constraint on hiking. 2s10s widening remains the cleanest expression of the thesis, with a kicker if the 5y5y forward breakeven begins to drift meaningfully above its post-2022 range — that is the moment target credibility joins reaction-function credibility on the same downward path, and the trade compounds.
The third-order trade is gold, and it is the most misunderstood. The May correction is a head-fake. Official-sector buying does not respond to six-week price action; it responds to multi-quarter institutional signals. A Fed that cannot price its own policy is the single largest such signal in two decades. Dollar-cost into weakness; do not chase strength.
The trade that pays the most if the thesis is right and almost nothing if it is wrong is equity downside skew. The S&P realized has been suspiciously calm for a market staring at 3.8% headline inflation, $100 oil, and a Fed in transition. 3-month 90/100 put skew has rarely been cheaper on a percentile basis. The structural story for credit spreads and growth multiples requires a reaction function the institution no longer reliably produces.
Dudley said the Fed is in danger of losing credibility. He chose the wrong tense. The institution lost target credibility years ago and adapted. What it is losing this quarter — quietly, in the spread between a chair's prepared remarks and a committee's vote tally — is the thing that made the first loss survivable.