How OEWS wages are measured

Every occupation, state and metro wage on this site comes from one program: the Occupational Employment and Wage Statistics survey, run by the Bureau of Labor Statistics with the state workforce agencies. Knowing how it works tells you what the numbers can and cannot say.

Who is asked

OEWS is an establishment survey, not a household survey. BLS does not ask workers what they earn; it asks employers what they pay. Each year the program contacts a sample of nonfarm establishments drawn from the unemployment insurance records that nearly all employers must file. Over a full three-year cycle the sample covers about 1.1 million establishments, collected in six semiannual panels of roughly 180,000 to 200,000 each. Responses from all six panels are combined to produce one set of estimates, with older panels’ wages adjusted to the current period using the Employment Cost Index.

Combining three years of data is what makes estimates for small occupations in small areas possible at all. It also means a published figure is not a snapshot of a single month, even though it is labeled with a reference month (May).

What employers report

Each sampled establishment reports, for every occupation it employs, the wage of each worker. Since the May 2022 estimates BLS processes most private and government records as exact wage rates; the small share of records without an exact rate is placed in one of twelve wage intervals, as all records were before. Intervals are defined in hourly terms, and salaried workers are converted from annual pay at 2,080 hours (see annual versus hourly figures).

Occupations are classified under the Standard Occupational Classification, and employers are given descriptions to assign their jobs. How well they do so is a known source of error; BLS edits the responses and follows up on inconsistencies.

What counts as a wage

OEWS wages are straight-time, gross pay before deductions. They include base rate, cost-of-living allowances, guaranteed pay, hazardous-duty pay, incentive pay such as commissions and production bonuses, tips, and on-call pay. They exclude overtime premiums, back pay, shift differentials, severance, holiday bonuses, stock bonuses, tuition reimbursement, and employer-paid benefits. A job whose compensation is mostly discretionary bonus or equity will look lower in OEWS than in total compensation.

The survey covers wage and salary workers in nonfarm establishments. It does not cover the self-employed, owners and partners in unincorporated firms, household workers, or unpaid family workers. For occupations where self-employment is common, such as real estate agents or writers, the published figures describe only the employed part of the occupation.

From reported wages to published figures

From the reported wages and the remaining interval counts BLS estimates employment, a mean wage, and the 10th, 25th, 50th (median), 75th and 90th percentile wages for each occupation in each area. Percentiles are found by locating the interval that holds the relevant worker and interpolating within it; the mean uses an average wage per interval taken from the National Compensation Survey. How to read the resulting table is covered in reading wage percentiles and mean versus median.

Every estimate carries sampling error, which BLS publishes as a relative standard error. Estimates that fail quality or confidentiality tests are withheld; see suppressed cells. Very high wages are reported as being at or above a top code; see top-coded wages.

How the figures reach this site

BLS publishes the May estimates each spring as spreadsheets. This site downloads them, checks them against pinned hashes and a set of hand-read known answers, and regenerates every occupation and state page from them. The reference month and release are printed with each figure, and nothing is adjusted, averaged across areas or filled in where BLS withheld a value.

What OEWS is not: it is not designed as a time series, because the sample, the occupational classification and the area definitions change over time, and it is not a measure of what any individual earns. It is the most detailed public picture of what employers pay for a job, in a given place, around a given May.