Founder Data ReportEvery January, Businesses Cut Hours. Almost None Cut Staff.
You can’t afford to risk your business without a Sidekick at your side.
Across 74 fortnights of Census data, more US businesses reported cutting hours than cutting people in every single one. In the first two weeks of January the gap triples, and headcount does not move at all.
Quick answer: Small and large businesses adjust hours long before they adjust headcount. In the latest Census fortnight, 14.7% of US businesses said the total hours their employees worked had fallen, against 9.5% that said the number of employees had – roughly 1.5 times as many. That gap holds in every one of the 74 fortnights Census has published since 2023. In the first two weeks of January it widens to a chasm: hours-cutting jumps to about 28% while headcount-cutting does not move at all.
- Across all 74 published BTOS cycles, more businesses reported cutting hours than cutting staff – 74 times out of 74. The two ranges do not even overlap: hours-cutting runs 13.4-28.6%, headcount-cutting 8.1-12.1%.
- In the first collection cycle of January, three years running, the share cutting hours averaged 28.3% against a normal 15.8% – a jump of +12.5 points.
- In those same cycles the share cutting headcount averaged 10.3% against a normal 9.7%. The move is +0.6 points. Effectively nothing.
- The shock behind it is real: in the same fortnights the share reporting revenue down rose from 31.9% to 39.0% and demand down from 24.3% to 29.3%. All of that adjustment landed on the hours dial.
- It is not a small-business habit. The January lift is +10.8 points at businesses with 1-4 employees and +18.1 points at those with 250 or more.
Two dials, and only one of them gets turned
Every fortnight the Census Bureau asks a large national sample of US businesses two questions that sit next to each other. Did the number of paid employees change in the last two weeks? Did the total number of hours those employees worked change? They are the same decision at two different costs, and the answers are not close.
In the most recent cycle, covering 10 August to 23 August 2026, 14.7% of businesses reported hours down and 9.5% reported employees down. On the other side, 7.9% reported hours up against 5.6% reporting employees up. Headcount is the flatter number in both directions: 84.9% of businesses reported no change in employees at all, against 77.4% reporting no change in hours.
That is not one unusual fortnight. Across all 74 cycles Census has published, the share cutting hours exceeded the share cutting staff 74 times out of 74. The two ranges do not overlap anywhere on the record: the worst fortnight for headcount, at 12.1%, is still below the quietest fortnight for hours, at 13.4%.
Source: U.S. Census Bureau, Business Trends and Outlook Survey (BTOS), national response estimates, 74 published fortnightly collection cycles from 11 September 2023 to 23 August 2026, downloaded from census.gov on 9 September 2026 and last updated at source on 26 August 2026; size-class figures from the companion Employment Size Class release. Q5 asks how the number of paid employees changed in the last two weeks and Q6 asks the same of total hours worked, so every figure here is the SHARE OF BUSINESSES reporting a direction of change over a fortnight – not the size of the change and not a level. Three cycles (202521, 202522 and 202523, covering 6 October to 16 November 2025) were cancelled by the federal funding lapse and are absent rather than zero. BTOS refreshes its sample once a year; the two series used here were tested for that break before any trend claim was made and the refresh steps sit inside ordinary fortnightly variation, unlike the survey’s subjective self-rating questions. Estimates are weighted and self-reported, and small cells in the size-class file are disclosure-suppressed.
What happens every January
The clearest view of the hours dial is the first collection cycle of January, which picks up the fortnight either side of New Year. It has now run three times – 1 January 2024, 30 December 2024, 29 December 2025 – and it does the same thing every time.
The share of businesses reporting hours down: 28.2%, 28.6%, 28.0%. The normal reading is 15.8%. The share reporting employees down in those very same fortnights: 10.6%, 10.4%, 9.8%, against a normal 9.7%.
One line nearly doubles. The other does not move. And the pressure is not imagined: in the same three fortnights the share reporting revenue down rose from 31.9% to 39.0% and the share reporting demand down from 24.3% to 29.3%. Something real hits the top line in the first week of January, and essentially all of the labour response comes out of the schedule.
All four bars are drawn on one scale, the 28.3% January hours reading, so the two rows can be compared directly against each other rather than each being stretched to its own width. Ordinary fortnights are the median of the 71 cycles outside the first cycle of January.
Source: U.S. Census Bureau, Business Trends and Outlook Survey (BTOS), national response estimates, 74 published fortnightly collection cycles from 11 September 2023 to 23 August 2026, downloaded from census.gov on 9 September 2026 and last updated at source on 26 August 2026; size-class figures from the companion Employment Size Class release. Q5 asks how the number of paid employees changed in the last two weeks and Q6 asks the same of total hours worked, so every figure here is the SHARE OF BUSINESSES reporting a direction of change over a fortnight – not the size of the change and not a level. Three cycles (202521, 202522 and 202523, covering 6 October to 16 November 2025) were cancelled by the federal funding lapse and are absent rather than zero. BTOS refreshes its sample once a year; the two series used here were tested for that break before any trend claim was made and the refresh steps sit inside ordinary fortnightly variation, unlike the survey’s subjective self-rating questions. Estimates are weighted and self-reported, and small cells in the size-class file are disclosure-suppressed.
Hours are the dial you can turn on a Tuesday and turn back on a Thursday. Headcount is a door that only opens one way.
Three years of it, in one line
Plotted across every published cycle, the two series barely interact. Hours-cutting sits in a band and spikes hard three times. Headcount-cutting is close to a flat line for 2023 through 2026, moving between 8.1% and 12.1% – a range of 4.0 points across three years and a pandemic recovery, an inflation cycle and a rate cycle.
The recovery is as fast as the spike. From the 2026 peak of 28.0%, hours-cutting fell to 21.5% a fortnight later and 17.9% the fortnight after that. Headcount over the same six weeks: 9.8%, 10.7%, 9.9%. It never noticed.
Both lines are on the same 0-32% scale. The horizontal axis is the 74 published cycles in order; the three cycles cancelled by the autumn 2025 funding lapse are absent, so the line joins early October 2025 straight to late November. That join is the one place on the chart where the horizontal spacing is not a fortnight.
Source: U.S. Census Bureau, Business Trends and Outlook Survey (BTOS), national response estimates, 74 published fortnightly collection cycles from 11 September 2023 to 23 August 2026, downloaded from census.gov on 9 September 2026 and last updated at source on 26 August 2026; size-class figures from the companion Employment Size Class release. Q5 asks how the number of paid employees changed in the last two weeks and Q6 asks the same of total hours worked, so every figure here is the SHARE OF BUSINESSES reporting a direction of change over a fortnight – not the size of the change and not a level. Three cycles (202521, 202522 and 202523, covering 6 October to 16 November 2025) were cancelled by the federal funding lapse and are absent rather than zero. BTOS refreshes its sample once a year; the two series used here were tested for that break before any trend claim was made and the refresh steps sit inside ordinary fortnightly variation, unlike the survey’s subjective self-rating questions. Estimates are weighted and self-reported, and small cells in the size-class file are disclosure-suppressed.
Not a small-business habit
The obvious reading is that small firms cut hours because they cannot afford to lose people they would struggle to replace. The size-class data does not support it as a small-business story. Cutting hours in January is what businesses of every size do, and the biggest ones do it hardest.
Businesses with 1-4 employees lift their hours-cutting by +10.8 points in the January cycle. Businesses with 250 or more lift theirs by +18.1 points, from a lower base. Every band in between sits between 10.8 and 18.1 points.
What the size cut does say is that the smallest businesses run the flattest year. Their ordinary-fortnight hours-cutting is 15.8% against 12.2% at the largest, and their January peak is the lowest of any band at 26.6%. A four-person business has less schedule to take out in the first place.
The track runs from 0 to 34%, so the length of each bar is the size of that band’s January jump. Read the printed figures rather than the bar lengths when comparing two bands. Employment size is the size of the business as reported to Census, and the smallest band here still covers the large majority of US employer businesses.
Source: U.S. Census Bureau, Business Trends and Outlook Survey (BTOS), national response estimates, 74 published fortnightly collection cycles from 11 September 2023 to 23 August 2026, downloaded from census.gov on 9 September 2026 and last updated at source on 26 August 2026; size-class figures from the companion Employment Size Class release. Q5 asks how the number of paid employees changed in the last two weeks and Q6 asks the same of total hours worked, so every figure here is the SHARE OF BUSINESSES reporting a direction of change over a fortnight – not the size of the change and not a level. Three cycles (202521, 202522 and 202523, covering 6 October to 16 November 2025) were cancelled by the federal funding lapse and are absent rather than zero. BTOS refreshes its sample once a year; the two series used here were tested for that break before any trend claim was made and the refresh steps sit inside ordinary fortnightly variation, unlike the survey’s subjective self-rating questions. Estimates are weighted and self-reported, and small cells in the size-class file are disclosure-suppressed.
What to do with this if you are running one
Budget the January hole before it arrives. Three years of data say the fortnight around New Year takes +7.1 points more businesses into falling revenue than a normal fortnight does. It is the most predictable soft patch in the calendar, and the cheapest time to have already decided how you will respond.
Know which dial you are actually turning. Cutting hours is reversible and cutting people is not. The whole economy behaves as though that is obvious, and then individual owners under pressure reach for the irreversible one first because it feels decisive. 74 fortnights out of 74 say the decisive move is the schedule.
Watch hours as your own early-warning signal. Headcount barely moves in this data because it is a lagging measure of trouble by the time it moves at all. If your own weekly hours are drifting down and your headcount is flat, you are not stable – you are in the first half of the pattern everyone else is in too.
Do not read a quiet headcount number as a healthy business. Across three years the share of businesses cutting staff moved within a 4.0-point band while revenue, demand and hours all swung. A number that never moves is not telling you the economy is calm. It is telling you it is the wrong number to watch.
Frequently Asked Questions
Do businesses cut hours before they cut staff?
In US Census data, consistently yes. Across all 74 fortnightly Business Trends and Outlook Survey cycles published between 2023 and 2026, the share of businesses reporting a fall in total hours worked exceeded the share reporting a fall in employee numbers in 74 of them. The two ranges do not overlap: hours-cutting runs 13.4% to 28.6% of businesses, headcount-cutting 8.1% to 12.1%.
How many businesses cut hours in January?
In the first collection cycle of January the share reporting total hours down averaged 28.3% across 2024, 2025 and 2026, against a median 15.8% in every other fortnight. The share reporting employee numbers down in those same cycles averaged 10.3%, against 9.7% normally – a difference of +0.6 points, which is within ordinary variation.
Is cutting hours instead of staff a small-business thing?
No. The January lift in hours-cutting is +10.8 points at businesses with 1-4 employees and +18.1 points at businesses with 250 or more. Larger employers swing their schedules harder, not less. What is distinctive about the smallest businesses is a flatter year overall: their ordinary-fortnight rate is 15.8% and their January peak 26.6%, both the lowest of any size band.
What does the Census Business Trends and Outlook Survey actually measure here?
BTOS asks a large sample of US businesses, every fortnight, whether a series of conditions increased, decreased or did not change over the previous two weeks. So 14.7% means 14.7% of businesses said hours fell – not that hours fell by 14.7%. It is a count of direction, not a magnitude, and a business that dropped one shift answers the same as one that halved its rota.
How quickly does the January hours cut reverse?
Within about a month. From the 2026 peak of 28.0% of businesses reporting hours down, the reading fell to 21.5% in the next fortnight and 17.9% in the one after. That is the point of the hours dial: it goes back. Headcount over the same period moved from 9.8% to 9.9%.
Published by The Lonely Entrepreneur — the community and coaching platform for entrepreneurs who are building alone. lonelyentrepreneur.com
This article is for educational purposes and is not a substitute for professional financial or legal advice.