The people · who decides, and what the record says about speed

Who decides

Pipes: step 1 of four, Holding. World: Storm.

Step 1 of four — 3 steps remain

Who decides, and how fast they have moved before

A payment pipe breaks. Someone has to decide whether to answer it, and how quickly. This tab is about those people, what they have actually said, and what the historical record says that is worth. It exists because the question "how long will they take?" is a fair one.

People change the timing. They do not change the pipes.

The chance on the Now tab is the model as fitted, with no chair in it. This tab prices the people separately, in a band small enough that a story about a person can never outvote the evidence about the plumbing.

The person in the chair

Kevin Warsh — Chairman of the Board of Governors and of the FOMC

In the chair since 22 May 2026. four-year term as Chair ending 21 May 2030; Governor term to 2040. President Trump; confirmed by the Senate 54 to 45 on 13 May 2026.

Fed Governor 2006 to 2011, so he sat through 2008 and was warned by name about the run on tri-party repo lending to Lehman

What he has said, in his own words

Dated, and quoted rather than summarised. A person's own sentence is evidence; what a newspaper says he thinks is not.

“most do not necessitate extraordinary policy interventions”

6 Nov 2008 — his first term as a governor, on the Fed's response to the crisis. federalreserve.gov/newsevents/speech/warsh20081106a.htm

“Moral hazard in the financial system is higher than any of us should countenance”

3 Feb 2010 — a governor arguing that rescues have a cost that outlasts the rescue. federalreserve.gov/newsevents/speech/warsh20100203a.htm

“Facts, not force, should be the predominant policy response”

28 Jun 2010 — against the idea that a policymaker must overreact when markets fall. federalreserve.gov/newsevents/speech/warsh20100628a.htm

“I strongly supported this crisis-time innovation, then and now ... I disagreed with the second round of asset purchases, and resigned from the Fed soon after”

25 Apr 2025 — the distinction he draws: crisis tools yes, routine tools no. Group of Thirty, 'Commanding Heights'

“where necessary and appropriate, we will not hesitate to act ... in crisis mode, that strikes me as a very prudent policy”

29 Jul 2026 — as Chair, at his press conference - the crisis carve-out he keeps. FOMC press conference transcript

“Unconventional policies to spur economic activity may suit genuine crises but should otherwise be used sparingly, if at all”

28 Aug 2026 — as Chair, at Jackson Hole - the sentence that decides what counts as a genuine crisis. Jackson Hole remarks

What actually predicts the wait

Not who is in the chair. What kind of answer is needed. The record is blunt about this, and the two kinds differ by two orders of magnitude.

1. TURNING ON A PIPE THAT ALREADY EXISTS

A repo window, the discount window, a swap line, or an emergency cut at a scheduled meeting. These are pre-authorised and operated by the New York Fed.

standing repo window, announced to effective — 1 day (Jul 2021)

emergency cut and swap lines, March 2020 — same day (Mar 2020)

fast, and not about the Chair at all.

2. BUILDING A NEW ONE

A new lending facility needs a majority of the seven-member Board, and by law it cannot be created without the prior approval of the Treasury Secretary. That signature is the gate.

Commercial Paper Funding Facility, announced to lending — 28 days (Mar-Apr 2020)

Main Street Lending Program, announced to operational — 88 days (Apr-Jul 2020)

Term Asset-Backed facility, announced to first loans — about 3.5 months (Nov 2008-Mar 2009)

weeks to months, and the gate is a signature, not a temperament.

Who else is in the room

The Chair is one vote of seven, and cannot create a facility alone. By law a new facility also needs the signature of the Treasury Secretary.

The Chair — Kevin Warsh, since 22 May 2026

one vote of seven; sets the question and the language

The Board of Governors — seven of seven seats filled: Warsh, Jefferson, Bowman, Barr, Cook, Powell, Waller

five votes are needed to authorise a facility; Cook is also in a removal proceeding, which is a live distraction

The New York Fed — John Williams, president since June 2018; Roberto Perli runs the market desk

physically operates the repo window and the swaps. On 22 Sep 2026 Williams said the tools must 'function effectively under all circumstances'

The Treasury Secretary — Scott Bessent, since January 2025

signs off every new facility by law. On 2 Aug 2026 he publicly asked the Fed to enlarge its foreign repo backstop, and got no visible answer in three and a half months

What the record says

Five chairs sit inside the years this model is fitted on. Their events: Greenspan 0, Bernanke 9, Yellen 0, Powell 4, Warsh 0.

Who sits in the chair does not predict the wait. The state of the plumbing does. On the same 13 events the spread of waits inside a single chair's tenure is 4.3 times, which is larger than any difference between chairs.

The measurements: the between-chair variance component is 0.00, the test statistic is 0.99 with a p-value of 0.46, the spread of waits inside a single chair is 4.3x in days, and the honest interval on 13 events runs 0.53x to 1.71x.

The widest spread of all belongs to the chair with the most events: Bernanke saw waits of 1 day, 125 days and 425 days for the same kind of break. His stated willingness to create facilities was the highest on record and his waits were the longest. What moved him was which market broke, not how much he believed in lending.

Both sides of the ledger

He has real arguments on both sides, and a page that only listed one would be arguing.

Why he might be faster

He already has a pipe switched on that needs no announcement.

A standing overnight repo operation exists and is priced off the policy rate, so a repo squeeze can be absorbed without a press release. That shortens the measured lag by not needing one.

His stated criticism of his predecessor is being too slow, not too fast.

He blames forward guidance for delaying the 2021 response and says the Committee will not hesitate to act in a crisis. That argues against a chair who waits for permission.

Why he might be slower

He treats a rescue as a narrow, high-threshold category.

'Most do not necessitate extraordinary policy interventions' and 'may suit genuine crises but should otherwise be used sparingly, if at all' are the same instinct, 18 years apart. A break whose severity is disputed lands on the wrong side of that line, and the first days go to classifying it.

The same week he argued against expanding the balance sheet, he voted for it.

In November 2010 he wrote that further asset purchases would have few durable benefits, and five days earlier he voted for the programme. Anyone scoring him by temperament would have got the sign wrong on the largest action of that year.

The claim, dated and signed

“A Warsh chair will take at least three months longer to create a NEW facility than the 2019 to 2023 average, while a repo or discount-window operation will arrive no slower than history.”

Written 21 November 2026 by the analyst who maintains this page. Scored on 21 November 2028. Rights and wrongs are recorded on the record tab on the grading date, whichever way it goes.

This claim is not in the printed probability. The 24.0% is the model exactly as fitted, with no chair in it.

The episode that has no precedent

The current episode is the first in which the Fed has been raising rates into market strain, and it has gone more than eight months with no new facility.

Every fast response in the record happened when cutting was defensible. This one has no precedent in the fitted data, which is a reason for the model's uncertainty, not evidence about one man.

Counted as counted as no event yet, not as a slow event.

What all of this is worth in the number

The plus-or-minus 10% band on the monthly rate is a JUDGEMENT, not an estimate. It is set once, here, and must not be widened or narrowed after a reading grades it. It is anchored to the best published validity for a single personality trait (r = 0.28-0.31, so 8-10% of variance) and is deliberately kept smaller than the plumbing channel, so that a story about a person can never outvote the evidence about the pipes.

If the Fed acts slower than its own history: 21.9% within a year, 56.7% within three.

If the Fed acts no differently than its own history: 24.0% within a year, 60.6% within three.

If the Fed acts faster than its own history: 26.1% within a year, 64.2% within three.

The whole of it — the chair, the board, the Treasury, the doctrine, the quotes — moves the one-year chance by 4.2 points in either direction. The plumbing’s own contribution is 2.1 points. That is the honest ranking of the two, and it is why the people are printed here rather than mixed into the headline.

It is not a personality forecast, and it is not in the printed probability. The headline 24.0% is the model exactly as fitted, with no chair in it. This block prices a slower Fed and a faster Fed so the reader can see the size of the judgement instead of being handed one number dressed as a measurement. Between-chair variance on the 13 fitted events: 0.00. Within-chair spread: 4.3x in days. Exact Poisson interval on 13 events: 0.53x to 1.71x. The chair hypothesis is not identifiable on this record, which is why the adjustment is capped, printed separately, and never mixed into the headline.

What this tab will not do

It will not score anyone's character and multiply the answer by it. The research run on 2026-11-21 is the reason: on the 13 events this model is fitted to, the difference between chairs is zero and the spread of waits inside one chair's own tenure is 4.3x in days. A number with no support behind it is not a measurement, it is an opinion wearing one.

If it is ever wrong about a person, the record tab is where it will be caught: the claim is dated, signed and scored on 21 November 2028, and the page prints the outcome whichever way it goes.