Reading wage data

Reading a wage percentile table

What the 10th, 25th, median, 75th and 90th percentile wages mean, how BLS estimates them, and how to use the spread between them to read an occupation's pay.

Last reviewed . Explains; does not advise.

The unit of content on this site is a five-number table: the 10th, 25th, 50th, 75th and 90th percentile wages for an occupation in a place. One figure cannot describe how an occupation pays; five can come close. Here is how to read them.

What a percentile is

A percentile wage is the wage below which a given share of workers in the occupation earn. The 10th percentile is the wage that 10 percent of workers earn less than, and 90 percent earn more than. The 50th percentile is the median: half earn less, half earn more. The 90th percentile is the wage that only 10 percent of workers exceed.

Percentiles describe the workers, not the jobs advertised. An occupation’s 10th percentile is close to what the lowest-paid tenth of people already doing the job earn, which is often what an entry-level position pays, but the two are not the same thing. BLS does not publish starting wages.

How BLS estimates them

Employers report headcounts in twelve wage intervals rather than exact wages, so each percentile is an estimate. BLS orders the intervals, finds the one in which the cumulative employment crosses the target share, and interpolates within that interval assuming wages are spread evenly inside it. The Handbook of Methods describes the procedure. The consequence for readers is that two occupations with the same published median can have slightly different true medians, and that a change of a few cents between years is noise.

Reading the spread

The distance between the 10th and 90th percentiles is the most useful single thing a table shows. A narrow band, where the 90th percentile is less than twice the 10th, means pay is compressed: experience, employer and location change little. Cashiers, fast-food workers and many jobs near the minimum wage look like this, because the legal floor holds up the bottom and the market offers little above it.

A wide band means the occupation covers very different situations. For lawyers, the 90th percentile exceeds the top code BLS reports while the 10th percentile sits where many salaried professionals start. For such occupations the median is a weaker summary and the whole row should be read.

Compare the 25th-to-75th range, the middle half of workers, when you want to ignore the extremes at both ends. That range is often the best answer to “what does this job usually pay.”

Comparing tables

Compare like percentiles across places. The state tables let you set one state’s median against another’s, or one state’s 10th percentile against the national 10th. Comparing a state’s 75th percentile with the national median tells you nothing, because the two statistics answer different questions.

Within a single table the figures always rise from left to right. If one of them is missing, BLS withheld it; see suppressed cells. If the 75th or 90th percentile reads “at or above” a value, the true figure is above BLS’s top code; see top-coded wages. Neither gap is filled in on this site.

Hourly and annual rows

Every table on the occupation pages shows percentiles twice, hourly and annual. For most occupations the annual row is the hourly row multiplied by 2,080 hours. For occupations that do not work a standard year, such as teachers, pilots and athletes, BLS publishes only one row and the other is blank; see annual versus hourly figures.

What percentiles do not tell you

They say nothing about why a worker is at the 90th percentile rather than the 10th: years of experience, employer size, specialty, hours and location all play a part, and OEWS collects none of them. They are also computed from wage and salary workers only, so the self-employed are absent. Used with those limits in mind, the five-number table is a far better description of an occupation’s pay than any average.

Sources

  1. BLS, OEWS Frequently Asked Questions (percentile wages)
  2. BLS Handbook of Methods: Occupational Employment and Wage Statistics, Calculation

More in reading wage data