Indicators · The labour market
Office jobs not regained after a downturn
Normal Direct evidence
Signals of a market fall and of a labour-market downturn have not been on together in the past three years.
US recessions Periods when this signal was on
What it measures
Whether, two years after signals of a market fall and of a labour-market downturn are first on together, real US business output is above its level at that date while jobs in information and in professional, scientific and technical services are below theirs.
Why it matters
In most recessions output and jobs recover together. If output recovers while jobs in work exposed to AI do not, firms may be automating that work and not rehiring. The same pattern followed the dot-com bust, before AI.
When it triggers
Triggers 24 months after signals of a market fall and of a labour-market downturn are first on together, if output is then above its level at that start and those jobs are below theirs.
Where it fits
How severe the downturn becomes. The argument predicts: A market crash and rising unemployment at the same time; then job losses that last; then policies that deepen the downturn, such as tariffs and spending cuts.
Other possible causes. Any recession brings falling markets and rising unemployment at the same time, and a bust in one industry, such as the dot-com bust, can leave its jobs below their earlier level while output recovers.
A trigger counts only if, during the period it measures, the prices US businesses charge were not rising faster than in 95% of quarters from 1990 to 2019.
When the rule was set
The rule was set on 8 Oct 2026, when it was not triggered. It triggers on a gap of any size: output higher, and those jobs lower, than when the two signals were first on together. A trigger can count as evidence only if, on data dated after the rule was set, the signal first reads normal, neither triggered nor close to triggering, and then triggers, and if the change it measures began after that date.
How often it has been on
Applied to past data at the end of each month from Jan 1990, using only what had been published by then, it was on in 4% of months to 2019 and 0% since 2020. More than six months from any US recession as dated by the National Bureau of Economic Research, it triggered once, in Dec 2002. 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 Dec 2002.