Wage Dispersion by Job: 369 Occupations, Ranked

Most salary-negotiation advice assumes the room exists. Counter 10–20% above the offer, anchor high, hold your number. One useful piece of context is how much published wages vary inside your occupation.

In some jobs, two people with the same title in the same country are paid very differently. In others, the middle of the distribution is much tighter. We measured that dispersion; it is a signal about the market, not a direct measure of what one employer will negotiate.

What we measured

The federal Occupational Employment and Wage Statistics survey publishes wage percentiles — the 10th, 25th, 50th, 75th and 90th — for around 830 occupations, built from a sample of about 1.1 million establishments. The gap between the 25th and 75th percentile is the middle half of everyone doing that job.

Divide that gap by the occupation’s median and you get a number that is comparable across jobs regardless of pay level. A value of 0.25 means the middle half of the profession spans a quarter of the median salary. A value of 1.0 means it spans the whole of it.

We started with 398 occupations with a national median above $45,000 and enough employment to benchmark. We then removed 29 residual SOC buckets such as “All Other,” leaving 369 actual occupation groups for the headline ranking. Figures are as published for the May 2025 reference period (released 15 May 2026), without adjustment.

The result: the spread ranges from 0.22 to 1.26 — a 5.7-fold difference between the tightest occupation and the loosest. The median occupation sits at 0.47.

Where published wages are widest

Many jobs near the wide end have individually attributable output or long seniority ladders. OEWS also includes commissions and production bonuses in its wage measure, which can widen the published distribution for roles such as financial advisors and sales agents. A personal financial advisor at the 25th percentile records $72,440; one at the 75th records $176,790.

Where there is almost none

The narrow end is not simply the low-paid end. It clusters around unionised transport, federal schedules and licensed clinical roles where credentials and formal structures often influence wages. That is an interpretation of the pattern; the survey itself cannot identify the cause.

The finding that surprised us: pay level barely predicts room

The intuitive assumption is that better-paid jobs have wider distributions. Across the 369 occupations, the correlation between median wage and relative spread is 0.43 — positive, but nowhere near decisive. Plenty of well-paid work has a narrow middle.

Every one of these is a licensed clinical profession. A nurse anesthetist earns a median of $236,590 — well above the 25th percentile for chief executives, which is $129,540 — and still sits in the tightest quarter of the sample for variation. If you are in one of these roles, the published pattern is a reason to investigate shift differentials, sign-on, schedule and employer-specific ranges rather than assuming the median reveals what your base can do.

Before deciding how much to counter, it is worth knowing whether your profession has a wide published distribution or a narrow one. That context cannot tell you what to ask by itself, but it can tell you which employer-specific questions to investigate.

Geography changes the answer more than you would think

Take a single occupation — software developers — and look at the twelve largest markets for it:

The two highest-paying markets have the narrowest relative OEWS wage bands. San Jose records a median of $213,110 — 59% above Austin — and its middle half spans half as much, relatively speaking, as Austin’s does.

The federal technical note defines wages as straight-time gross pay: base rate, cost-of-living allowances, commissions and production bonuses are in; nonproduction bonuses, equity and benefits are excluded. One plausible reading for equity-heavy tech markets is that base salary is more standardised while variation sits in stock and level, but OEWS cannot test that explanation directly.

So the reading is not “don’t negotiate in San Jose.” It is a prompt to inspect the employer’s base band and then ask separately about equity, refresh, sign-on and level.

What this does not tell you

Three limits worth stating plainly, because the temptation with a dataset this size is to over-read it.

It is neither total compensation nor pure base pay. OEWS includes base rate, cost-of-living allowances, commissions, production bonuses and tips. It excludes overtime, shift differentials, nonproduction bonuses, equity and employer benefit costs. That definition matters especially for advisors and sales roles, where commissions can contribute to the wide observed distribution.

Occupations are broad. “Software Developers” is one federal occupation code covering an enormous range of seniority and specialisation. Part of every spread above is the gap between a junior and a principal, not just between two comparable people.

A single estimate blends three years. Each OEWS figure combines six semiannual survey panels collected over three years, with older panels adjusted forward to the reference date. It is a smoothed model-based estimate, not a snapshot of one month.

We also excluded 3,840 local estimates — 5.7% of the total — where two adjacent published percentiles collapsed onto a single value. In 2,903 of them the collapse is in a tail: firefighters in Bakersfield-Delano are published with a 75th and 90th percentile both at $470,940 against a median of $72,410. In the remaining 937 it is in the middle: airline pilots in Atlanta are published with a median and 75th percentile both at $618,090 against a 25th percentile of $211,400. Both are estimation artefacts rather than real distributions, and leaving them in would have put wildly wrong occupations at the top of the list. No national figure was affected by either.

What to do with this

If your occupation sits in the wide half, do not assume the whole spread is bargaining room: it can also reflect seniority, employer, industry, worker mix and commissions. If it sits in the narrow half, do not assume nothing moves. Use the distribution to form better questions about the employer’s actual band and the other terms of the offer.

Either way the number is only half of it. Knowing your range and saying it out loud under pressure are different skills, which is why the next step after picking a figure is usually a salary negotiation script you have actually rehearsed.

To see where your own offer sits in this distribution — your occupation, your metro, adjusted to the current quarter — use the counter-offer calculator. It runs on the same federal data as this analysis.

Method

Source: BLS Occupational Employment and Wage Statistics, May 2025 reference period, released 15 May 2026 — the most recent release available. Cross-industry estimates only (industry_code 000000), detailed occupations only (SOC display level 3).

Spread is the interquartile range divided by the median: (p75 − p25) / p50. Occupations were included where the national median exceeded $45,000 and national employment exceeded 25,000, leaving 398. Local cells required employment of at least 150 and a strictly increasing set of percentiles, leaving 64,113 occupation-by-area cells across 528 metropolitan and nonmetropolitan areas.

Figures are as published, without escalation. Anyone wanting to reproduce this can download the same oe.* time-series files from the Bureau of Labor Statistics and compute the ratio directly — or take the whole occupation-by-metro table, with the derived spread measures already computed, from our pay band width data release, published under CC BY 4.0. That companion piece asks the geographic question the other way round: instead of one occupation across metros, it holds occupation mix constant and compares the areas themselves.

The full occupation-by-metro table behind this article is published as an open dataset under CC BY 4.0, together with the price-adjusted version of the same wages.