US Dollar: Assessing NFP release reliability – Standard Chartered

Standard Chartered economist Dan Pan analyses which US Nonfarm Payrolls (NFP) months provide the most reliable initial signals of labour-market conditions. Using mean absolute error comparisons between preliminary releases and final benchmarked data, he finds April and November stand out as most accurate, while January, September, March and May tend to see larger revisions, even when using three‑month moving averages.

Reading NFP reliability

"High volatility argues against taking preliminary nonfarm payroll (NFP) releases literally. Market attention has shifted towards moving averages of at least three months to smooth monthly estimation errors from seasonal noise, but these near-term trends are also subject to monthly and annual benchmark revisions. This raises the question of which preliminary NFP releases most reliably capture labour-market conditions in a given month."

"Comparing preliminary releases with final benchmark revisions, we note that the first release in April has the lowest mean absolute error (MAE) on a single-month (1m) basis and the second-lowest MAE on a three-month moving average (3mma) basis. November’s 1m MAE is also low, although accuracy has worsened post-pandemic. In other words, when assessing which months’ initial releases are closest to the final estimates, April and November stand out."

"By contrast, January, September, March and May have the largest revisions. The differences are not small: MAEs in the least reliable months are almost twice those in the most reliable – so trade those surprises at your peril."

"MAEs are lower on a 3mma basis than on a 1m basis, as errors in one month often appear to cancel out in the next. On a 3mma basis, April, July and October appear the most reliable, and January, May and September the least. Again, the gap between the most and least reliable is large."

"January and May 3mmas tend to overstate the trend amid upward biases accumulated from prior months NFPs were often understated pre-pandemic, but overstatement is now common post-pandemic."

(This article was created with the help of an Artificial Intelligence tool and reviewed by an editor. Know more.)