Half the capacity use gained after 2020 has since been given back

The interesting thing about manufacturing capacity utilisation over the last six years is not the level. It is the asymmetry. The series climbed 4.7 points between its 2020 trough and its 2022 peak, and it has since given back 2.3 of them. The climb took two years, and almost all of it took one. The give-back has taken three years and is not obviously finished.

The six readings, in order, are 72.9 percent in 2020, 77.3 in 2021, 77.6 in 2022, 76.6 in 2023, 75.4 in 2024 and 75.3 in 2025.

The climb was a single year

Between 2020 and 2021 the measure moved 4.4 points, from 72.9 to 77.3. Between 2021 and 2022 it moved 0.3, from 77.3 to 77.6. The peak year added almost nothing. If you were tracking the recovery in how hard American factories were being worked, it was effectively over by the end of 2021, and 2022 is the top of the series by three tenths of a point.

Scroll sideways to read the whole chart

Notes: Annual observations, manufacturing only. A change in this ratio can come from the output side or the capacity side; the series does not separate them.

Source: The Cefa Brief

Chart description

A line chart showing utilisation at 72.9 percent in 2020, jumping to 77.3 in 2021 and edging up to a peak of 77.6 in 2022, then falling to 76.6 in 2023 and 75.4 in 2024 and ending at 75.3 in 2025. The single steepest segment by far is the first one.

A line chart showing utilisation at 72.9 percent in 2020, jumping to 77.3 in 2021 and edging up to a peak of 77.6 in 2022, then falling to 76.6 in 2023 and 75.4 in 2024 and ending at 75.3 in 2025. The single steepest segment by far is the first one.

This is worth stating because 2022 is the year usually cited as the high-water mark, and the citation carries an implication that something was still building through that year. In this series nothing much was. The distance between 2021 and 2022 is smaller than the distance between any two consecutive readings in the decline that followed, apart from the last pair.

The give-back took three

From the 2022 peak the series fell 1.0 point to 76.6, then 1.2 points to 75.4, then 0.1 to 75.3. The three falls together come to 2.3 points, just under half of the 4.7 points gained on the way up. No single year of the decline comes close to the 4.4-point move that started the window. This is a slow unwind, not a break.

The level that leaves is worth naming plainly. At 75.3 percent, 2025 is the lowest reading in the series apart from 2020 itself. It sits below 2021, below 2022 and below 2023. A reader who anchors on the 2022 peak sees a modest retreat. A reader who anchors on 2021, the first year after the trough, sees the whole of that year's position surrendered and then some.

What one ratio cannot tell you

Utilisation is a share: the part of manufacturing capacity actually in use. Every movement above is a movement in a ratio, and a ratio can change from either side. Output can move, or capacity can. The series does not say which, and the capacity estimate that would answer it is not held on this site. I am reporting the shape of the line, not a diagnosis of what produced it.

That limit is a reason to read the levels rather than the direction. The direction has been the same for three years and tells you almost nothing on its own. The levels tell you that the factory base is being worked less hard than at any point since 2020, and that the gap has widened gradually rather than suddenly, which is a different problem from a shock and usually a slower one to notice.

The last observation deserves its own caution. From 75.4 to 75.3 is a tenth of a point, the smallest move this series can report at the precision it is published to. It may be the floor of the decline or it may be a pause in it. I would not call it either until there is another reading, and neither should anyone quoting these six numbers to argue that the line has turned.