Not a payrolls forecast — and treating it as one is a costly habit.
intermediate · 4 min read · 12 XP
ADP arrives two days before payrolls and is widely traded as a preview of it. That framing is out of date: ADP itself abandoned the goal of predicting NFP years ago and rebuilt the report as a standalone measure. Knowing what it now is makes Wednesday a much less dangerous session.
The ADP National Employment Report measures the change in US private-sector employment, built from the anonymised payroll records of the very large number of businesses that use ADP for payroll processing. It is released on the Wednesday before payrolls Friday, at 8:15am New York time.
Two features distinguish it from non-farm payrolls:
For most of its life, ADP's stated purpose was to predict the BLS payroll figure, and its methodology was tuned towards that. In 2022 it was rebuilt in collaboration with an academic partner and explicitly repositioned as an independent measure of employment rather than a forecast of NFP.
The practical consequence is blunt: the month-to-month correlation between ADP and NFP is weak. Over longer stretches the two series tell a similar story about the direction of the labour market. In any individual month, ADP has very limited power to tell you what Friday's number will be.
The Wednesday trap
The most expensive mistake around this release is trading it as a payrolls preview: a strong ADP, a long dollar position held into Friday, and then an NFP number that has nothing to do with it. The two prints disagree often. Trade ADP for what it is — a real but modest data point that moves the dollar for a few hours — and start Friday's decision from a clean sheet.
ADP publishes a pay insights section alongside the employment count: year-on-year pay change for job-stayers and for job-changers, separately.
That split is genuinely useful and has no equivalent in the official data with the same timeliness:
For a trader building a view on wage-driven inflation, this section is worth more than the headline job count above it.
No — at least not on the basis of ADP. The month-to-month relationship with NFP is weak, ADP excludes government hiring entirely, and the two are built from different populations. The right use of a strong ADP is as one small input into a labour-market view built from several sources: jobless claims, JOLTS, the employment sub-indices of the PMI surveys, and the payrolls trend itself.
Its most reliable value is confirmatory. When ADP, jobless claims and the PMI employment components all point the same way, the labour-market signal is much stronger than any one of them. When they conflict, the honest conclusion is that the labour market is genuinely unclear — which is itself worth knowing before Friday.
What to remember
ADP measures private-sector employment from real payroll records and, since its 2022 methodology rebuild, is explicitly not an attempt to predict non-farm payrolls — its month-to-month correlation with NFP is weak. Its most valuable content is the job-stayer versus job-changer pay comparison, a composition-free read on labour-market tightness that has no equally timely official equivalent.