A key use case for Point-in-Time data nonfarm payroll revisions (and surprises)

We're revisiting controversies surrounding market-moving revisions of non-farm payrolls (NFP) - and a powerful use case for CEIC's Point-in-Time (PiT) data to track how these revisions moved the market.

The US Bureau for Labor Statistics releases the initial NFP figure three weeks after the end of the month in question. The first revision comes a month later, and the second a month after that. Finally, an annual revision is made each year.

We've highlighted three trading days for the S&P 500 over the past year: Sept. 6, Nov. 1, and Jan. 10, when the BLS released flash estimates for payrolls for the months just ended - as well as revisions to previous monthly figures.

Arguably, on all three dates, the revisions were more important than the new NFP figures.

CEIC's Point-in-Time dataset lets users easily compare revisions to equity trading at the time. (PiT also helps manage the potential complexity stemming from revisions, which can pose challenges to modeling and backtesting; even slight revisions can result in substantially different forecasting outcomes.)

Interestingly, different eras tend to show relatively consistent overshooting and undershooting by the BLS. Our second chart breaks down the effects of NFP revisions and nets out the final figures. We can see that much of 2023-24 saw initial reports that proved too optimistic; this trend reversed itself in the fourth quarter of last year.

The latest NFP report showed initial March job gains dramatically exceeded estimates, to the pleasure of President Trump. Will the BLS eventually revise this downward? So far, early 2025 results suggest a shift back to 2023-24 overshooting.

We've added three more charts for last year's June, August and October NFP - showing how significant revisions for these months were.