Usual weekly earnings (CPS) versus OEWS

Two BLS medians are often quoted as “the typical wage”: the median usual weekly earnings of full-time workers, released quarterly, and the OEWS median for an occupation, released each spring. They come from different surveys asking different people different questions. The median weekly earnings page carries the first; the occupation pages carry the second.

Where the weekly figure comes from

The usual weekly earnings series is tabulated from the Current Population Survey, the monthly household survey the Census Bureau conducts for BLS with a sample of about 60,000 eligible households across all states and the District of Columbia. Earnings questions are asked of one-fourth of the sample each month, and only of wage and salary workers; the self-employed, incorporated or not, are excluded from the earnings estimates.

Respondents describe their own pay. Since 1994 they have been asked to pick the easiest basis for reporting, hourly, weekly, biweekly, monthly, annually or other, and to give the amount they usually earn on that basis; BLS converts it to a weekly equivalent. “Usual” is whatever the respondent takes it to mean; if asked, interviewers define it as more than half the weeks worked in the past four or five months.

What is in the figure

Usual weekly earnings are earnings before taxes and other deductions and include overtime pay, commissions and tips usually received, at the main job for people holding more than one. Full time means usually working 35 hours or more a week at the sole or principal job. The headline median therefore describes full-time wage and salary workers, all occupations together, with their usual overtime included.

The median itself is estimated by placing each weekly value into a $50-wide interval centered on a multiple of $50 and interpolating within the interval that holds the midpoint. BLS notes two consequences: the overall median can move in a direction different from every subgroup’s median when the subgroups’ shares change, and medians sitting in an interval where many people report a round number, such as $500, move more slowly than others. Constant-dollar versions use the CPI-U.

How OEWS differs

OEWS is an establishment survey. Employers, not workers, supply the figures, and the wage concept is straight-time gross pay: overtime premiums are excluded, while commissions, production bonuses and tips are included. OEWS covers part-time and full-time wage and salary workers together, since employers report jobs rather than hours, and it reports by detailed occupation and area with a May reference month and a three-year sample. It has no age, sex, race or education dimension; the CPS has all four, and no detailed geography below the state.

Put side by side, the two medians differ in population (full-time only versus all wage and salary jobs), in pay concept (usual pay including overtime versus straight-time), in unit (weekly versus hourly and annual), in who answers (worker versus employer), in period (a quarter versus a May built from three years) and in detail (demographic groups and broad occupation groups versus about 830 occupations in about 530 areas). An OEWS annual median divided by 52 is not the CPS weekly median for the same occupation, and neither is wrong.

What each is good for

The CPS series is designed for change over time and for comparisons across demographic groups: the weekly earnings by education, by sex, race and ethnicity and union versus nonunion tables all come from it. OEWS is designed for level: what a specific occupation pays in a specific place around a specific May, with percentiles. The state pages here are OEWS; nothing on them is from the CPS.

Reading them together

A reader wanting to know whether pay in general is rising faster than prices looks at the CPS series or real earnings. A reader wanting to know what a dental hygienist in Ohio earns looks at OEWS. When a news story quotes “median weekly earnings” next to an occupation’s “median annual wage,” the two numbers are each correct on their own terms, and the terms are the ones above. Nominal versus real wages covers the price adjustment that applies to both.