Labour market data: why the unemployment rate is the least useful number in the report
The headline everyone quotes is a ratio with a moving denominator. The numbers that shift rate expectations sit further down the release.
Key takeaways
- – The unemployment rate can fall because people found work or because they stopped looking
- – Wage growth and participation carry more policy signal than the headline rate
- – Revisions to prior months routinely matter more than the current print
Employment reports are among the highest-impact scheduled releases in any market. They are also among the most misread, because the number that gets the headline is a ratio whose denominator moves for reasons unrelated to labour demand.
The denominator problem
The unemployment rate counts people without work who are actively looking, as a share of the labour force. Someone who gives up searching leaves the labour force entirely and stops being counted as unemployed. The rate falls, and nothing good has happened.
This is why the participation rate belongs next to the headline. A falling unemployment rate alongside falling participation is a weak report dressed as a strong one. A rising unemployment rate alongside rising participation is often the opposite – people returning to look for work because they believe jobs exist.
What central banks actually watch
Wage growth is the component with the most direct line to policy. Services inflation is dominated by labour costs, so accelerating earnings makes the inflation problem harder to solve and pushes rate expectations up. Decelerating earnings does the reverse, and can matter even when headline job creation looks robust.
Hours worked is another quiet indicator. Firms cut hours before they cut people, so a decline in average hours often precedes outright job losses by months.
Revisions
Employment estimates are built from samples and revised as more responses arrive. A strong current print alongside large downward revisions to previous months can be a net negative for the trend, and markets sometimes take several minutes to work that out. This is one of the clearer cases where the first move after a release is unreliable.
Two surveys, one report
In several countries the job-creation number and the unemployment rate come from different surveys – one of employers, one of households – with different methodologies and different sample sizes. They can disagree for months. Neither is wrong; they measure different things. When they diverge sharply, treat the direction of the trend across both as the signal rather than either single figure.
Trading around the release
Spreads widen and depth disappears in the seconds around publication. The initial spike frequently reverses once the detail is digested, which makes it a poor entry and a common place to be stopped out at a price that existed only briefly. Reducing size beforehand is a position; so is standing aside.
