The deficit was widest in 2020 and the policy rate at its floor in 2021, while both construction series made their largest single-year move in 2023

Between the 2020 and 2021 observations the M2 money stock rose from 17.7 trillion dollars to 20.54, an increase of 2.84 trillion. Across the same two observations, construction spending on manufacturing plant rose from 75 billion dollars a year to 82, an increase of seven. The first of those is a book entry. The second is money laid out on construction. Plant spending did move later, and the dates are the whole point: its largest single step, 77 billion, runs from 2022 to 2023, and its peak of 244 lands in 2024. By then the federal funds effective rate had gone from 0.08 percent in 2021 to 5.14, and the deficit had narrowed from 3,096 billion dollars in 2020 to 1,374 in 2022 and widened again to 1,815.

Where this is written from

I am an analyst writing inside The Cefa Brief. I work from published statistics and public documents and from nothing else: annual national series redistributed by FRED and retrieved for this site on 19 September 2026, together with the agencies' own statements of what each series counts. That position sees the sequence of a response from the outside. It can say in which year a policy rate moved, in which year a money stock moved and in which year spending on plant moved, and it can subtract one published figure from another and show the working. It cannot see a decision. There is no series here for a bill, a meeting, an authority, a lead time, or an option that was considered and set aside. Nothing in these files records what anyone intended, what was available or what was refused. So I cannot tell you that crisis response is designed around speed, and I do not. I can tell you which instruments in this window moved first, which moved last, how far apart the two ends sit, and then say plainly that a sequence is not a motive.

The instruments that move by an entry in an account

Three series in this window set a six-year extreme inside its first two observations. The federal funds effective rate ran 0.38 percent in 2020 and 0.08 in 2021, the lowest of its six readings. The M2 money stock ran 17.7 trillion dollars in 2020 and 20.54 in 2021, a step of 2.84 trillion and the largest of its five annual steps. The federal deficit was 3,096 billion dollars in 2020, the widest of the six readings, and 2,774 in 2021. Each of those changes is executed by an entry in an account, and nothing has to be built, hired, permitted or delivered for the number to move. The measures such instruments are conventionally pointed at moved on a similar clock. The unemployment rate ran 8.1 percent in 2020, 5.4 in 2021 and 3.7 in 2022, a fall of 4.4 points across two observations, and total nonfarm payrolls added 4.1 million jobs and then 6.2 million, the two largest annual additions of the window. Whether the instruments produced those movements is not something these files can settle. What they show is that each of these numbers can change without anything being built, and that each of them changed between the first two observations.

The instruments that have to be built

Construction spending on manufacturing plant ran 75 billion dollars a year in 2020, 82 in 2021, 125 in 2022, 202 in 2023, 244 in 2024 and 215 in 2025, a sequence of steps of 7, 43, 77, 42 and minus 29. Nonresidential construction as a whole went the other way at the front: 856 billion dollars a year in 2020 and 845 in 2021, so it fell before it rose, reaching 1,309 in 2025. Its own largest step, 225 billion, runs from 2022 to 2023, the same interval as plant spending's. Whatever that spending added to capacity is difficult to locate in the output series. Manufacturing output, an index with 2017 set to 100, ran 93, 97.6, 98, 97.3, 96.4 and 97.3 across the six years and does not reach its own base in any reading. Industrial production ran 95.1 in 2020 and 101.2 in 2025, a gain of 6.1 index points across the whole window. Manufacturing capacity in use peaked at 77.6 percent in 2022 and stood at 75.3 in 2025.

The exclusion is arithmetic, not preference

This is the part that requires nobody to have decided anything. A response is assembled inside a window, and an instrument that cannot complete inside that window cannot be assessed inside it either, whatever its merits; the constraint binds before the merits are reached. Labour supply is the limiting case. The labour force participation rate ran 61.7 percent in 2020, 61.7 in 2021, 62.2 in 2022, 62.6 in 2023, 62.6 in 2024 and 62.4 in 2025. The entire six-year range is 0.9 points, the net change from end to end is 0.7, and the largest single step is 0.5, from 2021 to 2022. A series that moves half a point in its fastest year has little to offer a response assembled inside a window this short. None of this is visible as a deliberation in these files, because no deliberation is recorded in them. The sequence in the numbers is consistent with the argument and does not demonstrate it.

Reversal is the second asymmetry

The account-entry instruments came back. The policy rate went from 0.08 percent in 2021 to 5.14 in 2024, a rise of 5.06 points, and then to 4.21, which is 0.93 below the peak. M2 reached 21.61 trillion dollars in 2022, fell to 20.87 in 2023, and stood at 21.96 in 2025. The deficit narrowed by 1,722 billion dollars between 2020 and 2022, then widened by 441 between 2022 and 2024. Each of those instruments can be run in either direction, and each was. Two of the twelve series cited here never move backwards. Total nonfarm payrolls is one, adding jobs across each of its five steps. The other is the consumer price index, on a 1982-84 base of 100, which ran 258.9, 271, 292.6, 304.7, 313.7 and 322 across 2020 to 2025 and does not fall between any two observations; its largest single step, 21.6 index points, runs from 2021 to 2022. The reversible instruments have been reversed and the price level has not come down. These files establish no link between those two facts. They place both in the same six years and stop.

What the record can score, and what it cannot

A measurement asymmetry sits on top of the timing one. A policy rate, a money stock and a price index arrive as a single number. Plant construction is published as a flow of spending. Nothing in these files counts a finished factory, a commissioned line or a unit of installed capacity, so I cannot say what the 244 billion dollars of 2024 bought. The construction series does not only move later in this window. It has no outcome column at all, which means it cannot be defended inside the window on the terms the account-entry instruments are judged on.

What this licenses

The record supports a sequence and not much more. The policy rate reached its floor and the deficit its widest reading in the first two observations of the window, and M2's largest annual step runs between the same two. The two construction series registered their largest steps between the third and the fourth. The series that tracks labour supply moves by at most half a point in any single year of the six. A sequence is not a motive, and I have found no decision anywhere in these files. What can be said is narrower than a claim about intent and more useful than one. Anyone assembling a response inside a window shorter than a construction cycle is choosing from a list that has already been shortened, and the shortening happened before the meeting. The list is the part worth examining, and what fell off it before anyone looked.