Earnings by education: what the tables measure

The claim that more education means higher earnings rests on a handful of federal tables. They agree on the direction and differ on the numbers, because each one defines earnings, workers and education slightly differently. The trends section of this site will carry these series; this page explains what each one counts.

Three sources

Current Population Survey, Annual Social and Economic Supplement. Each March the Census Bureau asks a large household sample about the previous calendar year’s income. Table PINC-03 cross-tabulates total money earnings with educational attainment for people 25 and over, with breakdowns by sex, race, Hispanic origin and work experience. The Census Bureau’s annual income report draws on the same supplement.

American Community Survey. The Census Bureau’s continuous survey of about 3.5 million addresses a year produces table B20004, median earnings in the past twelve months by educational attainment for people 25 and over, with one-year estimates for the nation, states and large areas. Its sample size makes it the source for state-level figures.

BLS, from the CPS basic monthly survey. The “Education pays” chart reports median usual weekly earnings of full-time wage and salary workers 25 and over by attainment, from the monthly questions about the current job rather than the annual supplement, alongside unemployment rates.

Definitions that move the numbers

Earnings versus wages. The Census tables measure what a person received over a year: wages and salaries plus self-employment income, across all jobs, affected by hours and weeks worked. The BLS chart measures usual weekly earnings on the main job. Neither is an hourly rate. The OEWS figures on the occupation pages are straight-time rates for a job, with no education dimension; a bachelor’s degree holder working as a cashier appears in the cashier row, not a degree row.

Who is counted. The headline Census figures are usually for full-time, year-round workers, which removes most of the effect of hours; the full table includes everyone with earnings. The BLS chart is restricted to full-time wage and salary workers, excluding the self-employed. Changing the population changes the medians by thousands of dollars.

Education categories. The standard attainment levels are less than high school, high school graduate (including equivalency), some college no degree, associate’s degree, bachelor’s degree, master’s, professional and doctoral. Older series grouped the last three; comparing across years requires checking the grouping. Field of study is not in these tables; the ACS has a separate item for bachelor’s field.

Age. Tables begin at 25 so that most people have completed their schooling. Within the 25-and-over population, degree holders are on average younger than high school graduates in some cohorts and older in others, which affects a median that does not control for experience.

Reading a gap

A difference between the bachelor’s median and the high school median is a description of two groups, not the effect of a degree on a given person. The groups differ in occupation, industry, hours, region and age, and the tables do not hold those constant. Research that attempts to isolate the effect of schooling finds a smaller gap than the raw tables show, but still a substantial one; the tables are the right place to see the pattern and the wrong place to read a causal claim.

Why the occupation tables tell a different story

OEWS asks what a job pays, not who holds it. An occupation with a wide percentile spread, such as registered nurses, contains workers with associate’s, bachelor’s and graduate degrees, and the table cannot separate them. The Occupational Outlook Handbook’s “typical entry-level education” for each occupation, linked from every occupation page, is the bridge between the two kinds of table; see what the OOH is.

Release cycle

The Census income report and PINC tables appear each September for the previous year; the ACS one-year tables in September as well; the BLS chart is updated each year when the annual averages for the CPS are released. The trends pages will state the year and the exact table behind each figure.