Data Revision

When a statistical agency first publishes a major indicator — say, GDP growth or the monthly jobs count — it is working with incomplete information. Survey responses come in late. Administrative records take time to compile. So the agency publishes a preliminary "first print," then revises it as better data arrives. These revisions are normal and expected, not signs of error or manipulation.

Revisions follow a standard schedule for most series. A GDP release, for example, typically goes through an "advance" estimate, a "second estimate," and a "third estimate" over the following two months. Annual benchmark revisions then revisit years of data at once. The revision from advance to final can sometimes be substantial — large enough to flip a reading from positive to negative, or to change how economists characterized that period.

Suppose the advance jobs report shows 150,000 positions added (hypothetical figure). A month later the revision shows it was actually 210,000. Markets often respond more to the new data at the time of release than to the revision, even though the revision is more accurate. That gap between first print and revision is part of why economic surprises recur in the same data series.

Traders and analysts typically track revisions alongside the headline number on any economic calendar release. Seasonal adjustment factors are a common source of benchmark revisions, since the adjustment model itself is periodically re-estimated. Understanding revisions is foundational to reading any economic indicator series with clear eyes.