How the shares moved
Not on average. Across 237 job cuts at 145 companies listed on the New York Stock Exchange or Nasdaq, announced from January 2023 to November 2026, the shares' typical move against the S&P 500 in the three trading days around the announcement was 0.0% (the average was +0.3%), and 50% of the stocks beat the market and the rest fell behind it, close to an even split.
Not in these cuts. When the employer itself named AI as a reason for the cut (99 cuts), its shares moved +0.4% against the S&P 500 on average in the three trading days around the announcement. When it gave no AI reason (138 cuts), they moved +0.2%: a difference too small to tell apart from chance (the 95% range of the difference runs from −2.5 to +2.9 percentage points).
Every cut's move, before, around and after the announcement, near results or not: Do stocks go up after layoffs? and the file of every cut's share move (CSV).
What a row holds
- The cut: the employer, the day it announced the cut, the country, the number of roles where one was put on the record and whose number it is, and a link to the source anyone can open.
- What the employer said about AI: its own words, graded. Linked to AI when the employer named AI as a reason or credible reporting tied the cut to AI; not linked when it gave another reason and nobody tied it to AI, or denied AI had a part. The methodology sets out the rule.
- The stock: the ticker and exchange of the employer's primary listing, or of its listed owner, named in a column of its own (Aer Lingus carries IAG on the London Stock Exchange; Frito-Lay carries PepsiCo on Nasdaq).
- The day it first appeared: when the row entered the published register, beside the day the employer announced the cut.
Point in time
The register's files carry their full version history from Sep 16, 2026, and each row's first-published day is read from it. Rows already in the register on that day carry it, meaning on or before: the history cannot show the day they were added. Cuts announced before then were mostly added later, some of them in bulk, so they are history filled in afterwards, never signals that were available at the time.