How often OEWS is updated

Every wage on an occupation or state page here is labeled with a month and a year, such as May 2025. This page explains what that label means and when it changes.

One reference month a year

OEWS estimates are published once a year and refer to May of the previous year. The May 2025 estimates, for example, were released in the spring of 2026. BLS announces the exact date on its release schedule, usually late March or April. There is no quarterly or monthly update to the occupation wages; whatever is on the page is the latest until the next spring.

The May label is a convention. The data behind one release were collected over six semiannual panels across three years, with earlier panels’ wages adjusted forward to the reference month using the Employment Cost Index. The estimate is best read as “around May of that year,” built from a three-year window.

What changes between releases

A new release changes everything at once: employment, every percentile, every area and occupation. It can also change the structure. Occupational classifications are revised about once a decade, most recently with the 2018 SOC, which was fully phased into OEWS in the May 2021 release. Metropolitan area definitions change after each census and in between, when the Office of Management and Budget issues new delineations. Industry classifications are revised every five years. A table that looks the same from one year to the next can be describing a differently drawn area or a differently defined occupation.

Why BLS discourages year-over-year comparison

BLS states plainly that OEWS is not designed as a time series. Three features make differences between releases hard to interpret. First, the pooled sample means two consecutive releases share two-thirds of their underlying panels, so a change in the estimate reflects one new panel replacing one old one, not a year of change in the market. Second, the classification and area changes above can move workers between rows. Third, the estimation method itself has changed; a model-based approach adopted in recent releases differs from the earlier method.

None of this means wages do not change over time. It means OEWS is the wrong instrument for measuring the change, and the Employment Cost Index is the right one; see wage growth and the ECI.

How this site handles the cycle

The site holds one release at a time and regenerates every data page from it. When a new May release appears, a scheduled check notices the file on bls.gov, the ingest is re-run against pinned hashes and hand-read known answers, and every page, table and generated sentence is rebuilt. The reference month printed on each page changes with it. Previous releases are not shown as a trend line in Wave 1, because the comparable span has to be decided and documented first, and the site would rather show one honest year than a misleading series.

What to do if you need change over time

Use a series built for it. The Employment Cost Index measures quarterly change in wages and salaries holding the mix of jobs fixed. The Current Population Survey publishes median usual weekly earnings quarterly. For a single occupation’s trajectory across several years, BLS’s own guidance is to be cautious, to check that the occupation’s definition did not change, and to treat small movements as noise.

The practical rule for this site: read the month on the page, treat it as the data’s date rather than today’s, and expect the next number in the spring.