The Q3 2024 QCEW Data

For hardcore data nerds only

This post is mostly for people who follow the labor market data closely, and may not be of interest to more casual readers.

TL;DR: The QCEW data suggest, with weak confidence, an overcount in nonfarm payroll employment.

On Wednesday February 19th, we got the first snippet of Quarterly Census Employment & Wages (QCEW) data for Q3 2024. As a reminder, the QCEW is the primary source of data for the benchmark revision to nonfarm payroll employment (NFP, from the Current Employment Statistics, CES, or “establishment survey”, in the monthly jobs report). Every year in early February, the BLS revises the previous-year-March level of NFP based on the previous-year-March level of QCEW employment.

The QCEW is noisy, which is why the BLS engages in this benchmarking exercise once a year (with 4 quarters in hand) rather than every quarter. So far we have 2 of those 4 quarters for the early 2026 benchmarking exercise (we’ll get a preliminary benchmark revision estimate in August 2025). In my experience, 2 quarters are weakly predictive - they often but not always get the direction of the revision right.

The Q3 2024 QCEW shows an ongoing overcount of NFP relative to the QCEW through September 2024 of about 663K, though you should probably subtract the March delta (that data will not get revised again) which suggests an overcount of 471K. I view this as a low-confidence signal of future negative benchmark revisions - i.e. revisions are more likely to be negative than positive, but the magnitude is extremely uncertain.

Another twist is that historical QCEW data get iteratively revised, and we’ll get the next round of revisions in early March. Q2’s data, currently showing a 578K overcount (770K - 192K), is likely to get revised up by over 100K. And looking further ahead, Q3’s data is likely to get revised up by a smaller amount.