Why Pay Differs: Skills, Productivity, and the Gaps Economists Study

What this means

An employer deciding what to pay is answering a narrow question: how much value does this worker add to the business, and what does it cost to hire someone who can add it? Economists call the first half marginal productivityThe additional output or value a business gains from employing one more worker.. A worker who produces more value, produces it faster, produces it with fewer errors, or produces something the employer cannot easily get elsewhere is worth more to that employer, and competition among employers for such workers pushes pay upward. This is the core of the benchmark: more educated, more skilled, and more productive workers generally command higher pay.

Two qualifications belong immediately alongside that. First, productivity is measured in dollars, so what a business sells matters. The same effort applied in an industry with high revenue per worker supports higher pay than identical effort in a low-margin one, which is why a highly skilled worker in one sector can earn less than a moderately skilled worker in another. Second, pay is set in a market, so it depends on labor supplyHow many workers are available and willing to do a job at a given wage. as well as on demand. An occupation requiring rare skills, long training, or a license that limits entry has a smaller pool of qualified workers, and small supply relative to demand raises the wage. An occupation many people can do immediately has a large pool, and wages sit lower even when the work is genuinely difficult or important. Difficulty and social value are not the same thing as scarcity, and the labor market prices scarcity.

You can see productivity-linked pay most directly in jobs where output is measured. Commissioned sales, contract and freelance work paid per project, piece-rate manufacturing, tipped service work, performance-bonus roles in finance and technology, and self-employment all tie earnings closely to what the individual produces. In other jobs the link exists but runs through slower channels: raises, promotions, and the ability to move to a better-paying employer. A teacher, a nurse, or a public employee on a published pay schedule is not paid per unit of output, yet skill and experience still move their pay through steps, credentials, specialization, and advancement.

Now the second outcome: why pay varies across different jobs and within the same job. Across occupations, the drivers are skill and education requirements, length and cost of training, licensing, working conditions including danger and unpleasant hours, and the supply of people able and willing to do the work. Economists use the term compensating differentialExtra pay that compensates a worker for undesirable features of a job, such as danger, night shifts, or isolation. for the extra pay attached to jobs with genuinely worse conditions.

Within a single occupation, pay still varies substantially, and this surprises students. Experience is usually the largest factor, since productivity in most work rises with years on the job. Beyond that: employer size and profitability, industry, geographic location and its cost of living, specialization within the field, union or contract coverage, individual performance, and negotiation. Two registered nurses with identical licenses can earn meaningfully different amounts depending on whether they work in a rural clinic or an urban trauma center, in which state, in which specialty, and on which shift. The Bureau of Labor Statistics publishes wage data by occupation, industry, and metropolitan area precisely because a single national number for an occupation hides all of this.

Finally, the third outcome, which requires care. When researchers measure average earnings by race and by gender, they find persistent differences, and the differences do not disappear entirely after adjusting for the observable factors above. Economists study several explanations at once, and the honest description of the field is that all of the following are documented as contributing, while their relative weight remains an active research question:

Occupational and industry sorting. Workers are distributed unevenly across occupations and industries that pay differently. A substantial share of measured gaps is accounted for by which jobs people hold rather than by unequal pay within the same job. This immediately raises a further question researchers pursue: why the sorting occurs, which involves everything from field-of-study choices and early exposure to hiring patterns and expectations about who belongs in a given occupation.

Hours and work schedules. Some higher-paying roles pay disproportionately more for long, inflexible, or unpredictable hours. Where caregiving responsibilities fall unevenly, workers who need flexibility sort toward roles that pay less per hour, which shows up in earnings comparisons.

Experience interruptions. Time out of the labor force, most commonly for caregiving, reduces accumulated experience and tenure, and earnings in most fields rise with both. Research finds that earnings differences by gender widen notably after the arrival of children, a pattern large enough that economists study it as a distinct phenomenon.

Discrimination. Differential treatment in hiring, pay setting, assignment, or promotion is measured directly in audit and correspondence studies, where researchers submit otherwise identical applications differing only in signals of race or gender and compare callback rates. These studies find differences in treatment, which is evidence that discrimination operates alongside the other factors rather than being merely a residual.

Two methodological points make you a better reader of any statistic in this area. First, an "unadjusted" or "raw" gap compares all workers of one group to all of another and makes no attempt to hold anything constant; an "adjusted" gap controls for factors like occupation, hours, and experience. They answer different questions, they produce different numbers, and quoting one while arguing about the other is the most common error in public discussion of this topic. Second, controlling for a factor does not explain it away. If occupation accounts for much of a gap, the gap has been relocated rather than dissolved, and the question becomes why occupational distribution differs. Look up current figures from the Bureau of Labor Statistics or the Census Bureau rather than trusting a number you remember, and check whether the figure you found is adjusted or unadjusted before you use it.

Why it matters

Two practical things follow. The first is that you have real influence over the skill and productivity side of your own pay, through credentials, specialization, experience, choice of industry and location, and willingness to negotiate. Knowing that pay varies within an occupation is what tells you to research the range for your field in your area before an employer names a number, rather than accepting the first figure as fixed.

The second is that you will encounter claims about pay gaps constantly, usually in a form designed to end an argument rather than inform one. Being able to ask whether a quoted number is adjusted or unadjusted, what it controls for, and what the control itself might be hiding is a genuinely useful skill, and it is the skill this benchmark is asking you to build.

Try it

1. List five occupations where earnings are tied directly and visibly to individual output. For each, name the specific mechanism: commission, per-project contract, piece rate, tips, performance bonus, or self-employment profit. Then list five where the link between individual productivity and pay runs through raises, promotion, or credentials instead, and describe that channel for each.
2. Choose four occupations that differ widely in pay. Using the BLS Occupational Outlook Handbook and Occupational Employment and Wage Statistics, record for each: typical entry education, required licenses, median wage, and the 10th and 90th percentile wages.
3. For each of the four, write one sentence explaining the pay level in terms of skill requirements, training time, working conditions, and supply of qualified workers. Identify which of the four has the strongest compensating differential for bad conditions, and defend the choice.
4. Now go inside one occupation. Pull that occupation's wage data for three different metropolitan areas and, if available, for three different industries. Calculate the spread between the highest and lowest median you found.
5. Write a paragraph listing every factor you can identify that could produce that spread for workers holding the same job title. Rank your factors by how much of the spread you think each explains, and mark which ones you could actually verify with published data and which are speculation.
6. Find current published earnings-by-race and earnings-by-gender figures from the Bureau of Labor Statistics or the Census Bureau. For each figure, record precisely: what population it covers, whether it compares median annual earnings or weekly earnings, whether it is restricted to full-time workers, and whether it is adjusted for any characteristics.
7. Locate one figure that is unadjusted and one that adjusts for factors such as occupation, hours, and experience. Write a short explanation of why the two numbers differ and what question each one actually answers.
8. For each of the four explanations discussed in this lesson, write two to three sentences describing the mechanism and naming what evidence would help assess how much it contributes. Then write one paragraph on why the statement "occupation accounts for much of the gap" relocates the question rather than closing it.
9. Write a one-page synthesis. State what the data you gathered does show, what it does not settle, and where you found researchers disagreeing. Do not conclude with a single-cause explanation, and flag any place where you had to rely on a source that was arguing a position rather than reporting measurement.

Check yourself