Indicators · The labour market
Permanent job losses
Normal Downturn signal
24.5% of the unemployed had lost their job for good in Oct 2026, down 1.9 percentage points on a year earlier.
The change in the share over twelve months is lower than 63% of the readings since 1995.
US recessions Periods when this signal was on
What it measures
People who lost their job for good, as a share of all unemployed Americans.
Why it matters
People on temporary layoff are usually recalled when demand returns; people whose job ended for good have to find a new one. If firms automate the jobs they cut in a downturn and do not rehire, this share would rise. It rose about 15 percentage points in 2008–10 and fell in 2020, when most layoffs were temporary.
When it triggers
Triggers when the share is 5 percentage points or more higher than a year earlier.
Where it fits
Unemployment starts to rise. The argument predicts: Unemployment starts to rise within a few months of a market crash, and the jobs lost in work exposed to AI are not regained.
Other possible causes. Every recession triggers these signals, and most of the jobs lost are later regained.
A trigger can count as evidence for point 5 of our view, a recession in which spending fails.
The prediction
Open Set 8 Oct 2026
- If we are wrongtriggers, then stops
- Now−1.9 pointsOct 2026
- Start−1.5 pointsSep 2026
- If we are right5.0 points or aboveby Mar 2027
- Severe16.1 pointsas in 2007–09
If our view is right, this signal triggers by Mar 2027: the change over a year in the share of the unemployed who lost their job for good reaches 5.0 points or above. The date comes from the model, in which US unemployment's rise over twelve months is by then larger than 99% of twelve-month changes from 1990 to 2019. A downturn of that size would be expected to show in this measure as well.
If it triggers and then stops being triggered on two consecutive readings, that counts against our view, which expects a downturn to deepen for several years once it begins. The deadline refers to the period the data cover. If it passes without a reading of 5.0 points or above, that counts somewhat against our view.
When the rule was set
The rule was set on 24 Sep 2026, when it was not triggered. Its threshold is chosen by the author.
How often it has been on
Applied to past data at the end of each month from Apr 1995, using only what had been published by then, it was on in 11% of months to 2019 and 14% since 2020. Each time it triggered, it was within six months of a US recession dated by the National Bureau of Economic Research. The rule was set with these years in view, so these figures cannot show how well it predicts.
Past triggers
On the data published at the time, it triggered in Jun 2001, Nov 2008 and Oct 2020.