Four ways to measure US wage growth

The composition trap

Average hourly earnings is an average, so it rises when low-paid jobs disappear even if no one gets a raise. That is exactly what happened in spring 2020: average hourly earnings jumped nearly 8% year on year while almost nobody received an increase, because the jobs lost were concentrated in low-wage leisure and hospitality. The ECI, which holds the job mix fixed, showed nothing like that.

The Atlanta Fed tracker avoids the trap differently: it follows the same people, so it answers "how much more is a given worker paid than a year ago". Its split between job switchers and job stayers is the clearest available evidence on how tight the labour market is, because the switcher premium widens when employers must bid workers away.

Which one to use

Use the ECI to judge inflation pressure from labour costs. It is the measure the Federal Reserve cites most often precisely because composition cannot distort it, and it includes benefits, which are a third of what employers pay.

Use the Atlanta Fed tracker for labour market tightness and for the switcher-stayer gap. Use average hourly earnings only for timeliness — it arrives first, on the first Friday of the month, and should be read as a rough signal.

Use QCEW when accuracy matters more than speed: it is a near-census rather than a survey, and it is the only one that captures bonuses and equity compensation properly. It arrives about five months late.