What an Occupational Wage Actually Measures
This chapter focuses on the wage estimate before we ask why it rose. The practical question is how to read a mean, a median, a percentile, or a geographic and industry comparison without treating any one of them as a direct score of an occupation or the people in it. That measurement discipline matters because a later claim about occupational wage growth is only as clear as the comparison underneath it.
The Occupational Employment and Wage Statistics program of the U.S. Bureau of Labor Statistics provides the kind of data most readers encounter first. Its occupational profiles describe jobs, report employment and wage estimates, and show how pay is distributed across workers. Those measures are useful. They can also be overinterpreted if the occupation, population, place, industry, and wage measure are left implicit.
The distribution behind the number
An occupational wage estimate is not one number in the way a thermometer reading might be. It is usually a summary of a distribution. The BLS profiles provide hourly and annual estimates for about 830 occupations, alongside the occupation’s definition, national employment information, and wage data by state, area, and industry. The first task is therefore to identify the object being summarized: which occupation, in which release, using which wage measure and population.
The mean is the arithmetic average across the observed wages. It is useful for describing the overall level, but high wages at the upper end can pull it away from what a typical worker earns. The median—the 50th percentile—is the midpoint of the distribution: half of the observed wages are at or below it and half are at or above it. The two measures can be close, or they can diverge when the distribution is uneven.
The other percentiles describe the spread. At the 10th percentile, roughly 10 percent of workers earn at or below that wage; at the 90th percentile, roughly 90 percent earn at or below it. The BLS page also highlights the 25th and 75th percentiles. Taken together, the 10th, 25th, 50th, 75th, and 90th percentiles show whether an occupation has a narrow range or a large gap between lower- and higher-paid workers. The middle 80 percent lies between the 10th and 90th percentiles, but that interval is not a promise about where any particular person should fall.
This is the first distinction between a wage distribution and a value judgment. A wide spread may reflect differences in experience, education, location, industry, responsibilities, working conditions, or other factors. A high median may describe a high-paying occupation, but it does not reveal whether the occupation’s underlying premium is high because the work is in strong demand, because the workers are selected in a particular way, or because the data combine very different settings.
The BLS measures are especially valuable when read together rather than as a leaderboard. A mean can describe the average level; a median can describe the center; percentiles can show dispersion; employment information can indicate whether a high-wage setting contains many jobs or only a small number. None of these measures is redundant with the others, and none alone explains the mechanism behind the distribution.
Consider two occupations with the same median but different 10th and 90th percentiles. In one, most workers may be clustered near the center; in the other, wages may spread widely across experience levels, employers, or specialties. The median alone would make them look alike while the distribution tells a different story. The reverse can also happen: two occupations may have similar means but different medians if one has a long upper tail. A careful comparison therefore asks whether the decision depends on the center, the range, or the opportunity to reach a particular part of the distribution.
That question is especially important when a reader uses a national statistic to interpret an individual offer. A person is not assigned a percentile because an occupation has a percentile table; their wage is observed within a particular place, industry, role, and stage of experience. The profile can supply a reference point, but it cannot guarantee that a person with a given degree or number of years will land at the 75th percentile. The BLS page itself presents experience and education as factors that may affect position in the distribution, not as a formula that determines it.
The occupation definition is part of the measurement, too. If the definition changes, or if a broad occupational category contains materially different kinds of work, a movement in its average can reflect a change in the mix of tasks or employers as well as a change in pay. The profile’s definition helps establish what is being grouped together. Without reading it, a comparison can appear more precise than the underlying category.
A second discipline is to keep a snapshot separate from a change. The BLS page describes May 2025 OEWS estimates. It is a current-estimate guide to where wages sit and how they are distributed in that release. By itself, it does not show whether an occupation’s wage premium rose from an earlier year, whether the people in the occupation changed, or whether a demand shock caused the movement. A cross-sectional estimate can be precise about a level while saying little about the path that produced it.
That distinction is easy to lose because readers naturally turn a ranking into a story. If one occupation has a higher median than another, it is tempting to infer that its workers are more capable or that its tasks are more socially valuable. The data do not make either inference automatically. They tell us how the wages in the defined comparison are distributed. The explanation requires more information.
Context decides what a comparison means
The same occupational label can contain several labor markets. The BLS page says that experience and education affect where an individual’s wage may fall within the national distribution. Someone new to a field may be near the lower percentiles, while someone with more experience and education may be near the upper percentiles. This is a useful orientation, not a rule for predicting an individual salary. People with similar experience can face different employers, regions, responsibilities, bargaining conditions, or opportunities.
Location is one of the clearest sources of context. OEWS provides wage information for states, the District of Columbia, selected U.S. territories, and approximately 530 metropolitan and nonmetropolitan areas. The BLS page notes that workers in large metropolitan areas often earn higher salaries, where cost of living and other factors can push wages upward. Among the areas shown in its example, San Francisco has higher mean hourly wages for the selected occupations than Minneapolis. That comparison is descriptive: it tells us that the labor markets differ, not that location alone caused the difference or that the higher nominal wage produces a higher standard of living.
Industry can change the comparison even when the occupation remains constant. The page illustrates this with computer systems analysts. In May 2025, the listed annual mean wage is $99,520 in colleges, universities, and professional schools, compared with $134,970 in aerospace product and parts manufacturing. The listed medians are $99,920 and $135,410 respectively. The difference is large enough to matter in a salary discussion, but the occupation label has not supplied the explanation. The BLS page points to working conditions, where industry employment is concentrated, union prevalence, and other factors as possible contributors.
This example should not be turned into a ranking of industries. It demonstrates a comparison problem. If an analyst observes that computer systems analysts earn more in aerospace than in higher education, several interpretations remain possible: the work may involve different technologies or constraints; employers may compete for different kinds of expertise; the locations may differ; the workforce may have different experience or education; or institutional conditions may affect pay. The table describes the difference. It does not isolate its cause.
It also shows why “same occupation” is not the same as “same work situation.” An analyst moving between these industries may face a different mix of clients, security requirements, production schedules, team structures, or scarce technical knowledge. Those differences may be part of what employers are paying for, but the table does not tell us which one. Nor does it show whether the higher-paying industry attracts more experienced workers, selectively retains them, or offers fewer positions. A wage comparison is strongest when its interpretation stays no broader than the evidence that produced it.
For practical use, a reader can treat the profile as a sequence of increasingly specific comparisons. Start with the occupational definition and the national distribution. Then ask how the local area differs from the national reference, whether the relevant industry has its own wage pattern, and whether the employment base is large enough to make the comparison useful for the decision at hand. This does not convert the profile into a personal prediction. It makes the reference frame visible and exposes where additional information is needed.
Employment information adds another important check. A high mean wage in a particular area or industry can coexist with relatively few jobs. Conversely, a setting with a lower mean may offer a much larger employment base, more opportunities for entry, or a different path for acquiring experience. The BLS profiles and maps help a reader examine those dimensions rather than treating the highest wage as the only relevant outcome. For a worker, compensation is one consideration alongside access, stability, learning, autonomy, and the likelihood of finding a suitable position; the wage table cannot settle those trade-offs by itself.
The individual position inside a distribution also needs careful language. If a person’s wage falls near the 25th percentile, that does not prove that they are underperforming. It may reflect limited experience, a lower-paying industry, a different region, a narrower role, or a deliberate trade-off. If a person is near the 90th percentile, that does not prove that they are intrinsically more valuable than everyone below them. It shows that their observed wage is high within the specified distribution. The reasons still need to be examined.
This is where the BLS view and the research question about wage growth meet but do not merge. OEWS tells us how to define and compare an observed wage. It can reveal that two workers with the same occupational label are situated in different distributions. It cannot, on its own, separate an occupation-specific wage premium from the composition of workers or from the returns to experience. That is a different empirical task.
A study of wage growth must also decide what remains constant. Are we comparing the same occupation across years? The same location and industry? The same workers, or everyone classified into the occupation at each date? If the group changes, an average can move even when the wage schedule facing a comparable worker changes less—or when the underlying premium changes more than the average suggests. These are not technical footnotes. They determine what the comparison means.
The practical reading rule is therefore straightforward. First identify the occupation and the definition used. Then identify the wage measure and the center or percentile being reported. Next hold location, industry, experience, education, and time period in view. Finally ask whether the evidence is describing a distribution at one point or explaining a change over time. A wage number becomes more useful when its comparison frame becomes explicit.
Once that frame is held steady, a harder puzzle comes into focus: two observations can use the same occupational label and still contain different kinds of workers. The next question is not only where the average sits, but who is inside it—and how sorting into and out of the occupation can hide the movement of the occupation’s underlying wage premium.