The same number of people, a third fewer notices
Take New York and New Jersey in 2026. New York filed 147 notices covering 13,242 workers. New Jersey filed 114 notices covering 13,389 workers. Almost exactly the same number of people lost their jobs, reported through 23% fewer filings.
Now take California. It filed 1,096 notices in the same period — 7.5 times New York's count — but those notices covered 59,149 workers, only 4.5 times New York's total. On a per-notice basis the difference is stark: the median California filing covers 18 workers. The median New Jersey filing covers 86.
A state's WARN notice count is not a measure of how much displacement happened there. It is a measure of how that state's reporting rules and filing practice chop displacement into paperwork.
Why the medians diverge so much
A WARN notice is triggered by thresholds — plant size, how many people are affected, what share of a site the layoff represents — and states layer their own rules on top of the federal floor. A state whose rules catch smaller events produces a long tail of small notices. A state whose rules only catch larger events produces fewer, heavier ones.
The mean tells you even less than the median here, because a handful of very large filings drags it upward. In 2026 alone, 16 filings across the six states covered 500 or more workers each (151 do so across our whole table). In a state with few filings, two or three of those will dominate the average completely. Nebraska is the clearest case: 6 filings in 2026, mean 190.5 workers, median 154 — a sample far too small to compare against California's thousand-plus, however tempting the single number looks.
What this breaks in practice
If you are screening for cases, the volume ranking sends you to the wrong state first. California will always top a notice-count list, and most of what you find there will be small filings. If your threshold of interest is, say, fifty affected workers, more than half of California's 2026 filings fall below it while more than half of New Jersey's clear it.
If you are sizing a market, counting notices understates states that file in large blocks. If you are writing about a state's labour market, a year-on-year change in notice count can move entirely because a filing rule changed, with no change at all in how many people were let go.
The same trap applies to our own numbers, and we will say so
Our table holds 7,065 filings across California, New Jersey, Utah, New York, Maryland and Nebraska. It would be easy to publish that as a coverage statement and let it stand. It should not stand, because the six states do not cover the same period. Our New Jersey records reach back to 2004. California begins in 2023. New York and Maryland only begin in 2026.
So the all-time totals in our own data are not comparable between states either. New Jersey appears to have more affected workers than California across the full table — 351,411 against 288,181 — and that is an artefact of a twenty-two-year window sitting beside a three-year one. It is not a finding. Every comparison in this post is restricted to 2026, which is the only window all six states share.
That restriction is the whole point. A number without its window is not a measurement.
What to ask for instead
Three questions make a WARN dataset usable. What period does each state cover, separately? What is the distribution of affected workers, not just the total? And what is the state's own filing threshold during that period?
If a provider cannot answer those for each state, the counts they are showing you cannot be ranked against each other — and a ranking is usually exactly what they are being used for.