The Inequality Within

Economists have a simple tool for measuring inequality within a group: the P90/P10 ratio. Take the wage at the 90th percentile — the threshold where only 10% earn more — and divide it by the wage at the 10th percentile, where only 10% earn less. If everyone in an occupation earned roughly the same, the ratio would be close to 1.0. If the top earners make double what the bottom earners make, the ratio is 2.0. At 3.0 or above, you're looking at an occupation where being at the top means earning three times what being at the bottom means, even though both workers carry the same job title.

Across all 769 detailed occupations in the 2024 OEWS data with valid percentile data and at least 1,000 workers, the median P90/P10 ratio is 2.29. That's the typical amount of internal inequality in an American occupation. The average is higher — 2.43 — because a cluster of high-ratio outliers pulls the mean upward. The minimum ratio belongs to postmasters and mail superintendents at 1.34, and the maximum to legislators at 6.76. Between those poles lies an enormous range of labor market structures, each telling a different story about what drives pay within a job.

Think about what a 2.29x ratio actually means in human terms. If a job has a 10th percentile wage of $30,000, the 90th percentile worker earns about $69,000. That's a $39,000 spread. If the 10th percentile is $50,000, the 90th earns $114,500 — a $64,500 spread. The ratio is a multiplier, and it amplifies as the base gets larger. For high-paying jobs, even modest ratios produce enormous dollar spreads. Computer and Information Research Scientists have a relatively unremarkable ratio of 2.88, but because their 10th percentile starts at $80,670, the 90th percentile reaches $232,120 — a spread of $151,450. The ratio doesn't change, but the human consequence does.

The distribution of ratios follows a predictable pattern. Of 769 occupations, 230 (30%) have ratios below 2.0 — relatively compressed pay structures where the top earner makes less than twice the bottom earner. Another 411 (53%) fall between 2.0 and 3.0, the zone that accounts for the majority of the American workforce. Only 109 occupations (14%) have ratios between 3.0 and 4.0, and a mere 19 (2.5%) exceed 4.0. Extreme within-job inequality is rare, but it concentrates in specific corners of the economy — and employs a disproportionate number of workers in a few very large occupations.

The chart above shows something counterintuitive: the most common experience for American workers is moderate inequality. Some 50.2 million workers (36.3% of the workforce) work in occupations with “standard” P90/P10 ratios between 2.0 and 2.5. Add the 23.5 million in moderately compressed jobs (1.75–2.0x) and the 25.8 million in tightly compressed jobs (below 1.75x), and you find that nearly three-quarters of the workforce — 99.4 million workers — work in occupations where the top earner makes less than 2.5 times the bottom. The dramatic ratios above 3.0 affect only 17.1 million workers, about 12.4% of the workforce.

But employment-weighted averages can mislead. The occupations with extreme inequality may employ fewer total workers, but they include some of the economy's most visible and aspirational roles. Securities and financial sales agents (472,000 workers, ratio 4.57), producers and directors (145,000, ratio 4.61), and the massive 1.2-million-strong army of service sales representatives (ratio 3.85) are all jobs where the gap between the floor and the ceiling is enormous. In these occupations, getting the same job title is the easy part. What you earn after that depends on factors the BLS doesn't capture in its classification: who your clients are, which market you're in, how long you've been building a book of business, and whether luck delivered you a whale account early in your career.

To understand what drives within-occupation inequality, start at the extremes. The table below shows the 20 occupations with the highest P90/P10 ratios in the country — the jobs where what you earn depends most dramatically on where you sit in the distribution.

The pattern is unmistakable. The occupations with the highest internal inequality fall into three archetypes. The first is performance-driven pay — jobs where compensation is tied to individual output, commission, or client revenue rather than a fixed salary schedule. Securities and financial sales agents are the poster child: a junior broker cold-calling prospects might earn $47,000 while a veteran managing institutional portfolios earns $215,000, all under the same occupational code. Real estate agents show the same pattern. At the 10th percentile, an agent earns $31,940 — barely above minimum wage — while the 90th percentile agent earns $125,140, nearly four times as much. The same license, the same MLS access, the same yard signs. But the agent who locked in listing contracts in a hot zip code builds a pipeline that compounds over decades, while the agent working rural foreclosures may never break even after desk fees.

The second archetype is institutional hierarchy masquerading as one occupation. Judges and magistrates (ratio 4.65) illustrate this perfectly. A small-town justice of the peace earning $46,520 and a federal appellate judge earning $216,540 carry the same BLS code despite operating at entirely different levels of the legal system. Similarly, postsecondary teachers (ratio 4.35) include both an adjunct scraping by on $39,720 at a community college and a tenured professor at a research university pulling $172,850. The BLS classifies them identically, but their labor markets are as different as two separate occupations.

The third archetype is geography and market tier. Radio and TV announcers (ratio 5.07) are the sharpest example. A morning DJ at a small-market AM station in Wichita might earn $26,000. An anchor at a top-five network affiliate in New York earns $131,780 or more. The skill set overlaps — both read news, interact with audiences, manage live broadcasts — but the advertising revenue behind their microphone differs by orders of magnitude. Fashion designers show the same dynamic: a designer at a regional department store brand earns $35,970, while their counterpart at a luxury fashion house in Manhattan earns $169,620. The occupation code is identical; the market is not.

Now consider the other extreme. The occupations with the lowest inequality — the most compressed wage structures — share a different set of characteristics.

The most compressed occupations in America share one overriding feature: there is no way to be dramatically better at them, or no market willing to pay dramatically more for marginally better performance. A postmaster in Iowa and a postmaster in Oregon both run a facility under the same federal pay scale; the ratio of 1.34 reflects nothing more than grade and step differences in the General Schedule. Farmworkers (1.44) are paid near a de facto floor set by minimum wage and local labor market competition; the ceiling is low because no farm can extract much more marginal product from one field laborer over another. Bank tellers (1.54) operate on a narrow pay band set by the institution — the spread between a new hire and a ten-year veteran might be $17,000, not $170,000.

Notice the employment counts on this list. Nursing assistants (1.39 million), stockers (2.78 million), and cashiers (3.15 million) are among the largest occupations in the country, and all have ratios below 1.70. These are the true “flat” occupations — jobs where almost everyone earns roughly the same modest wage, regardless of talent, tenure, or geography. A nursing assistant at the 10th percentile earns $31,390; at the 90th percentile, $50,140. That $18,750 gap is the entire range. Contrast that with software developers, where the 10th percentile starts at $79,850 and the 90th percentile reaches $211,450 — a spread of $131,600, or seven times the CNA range, within a single occupation.

The implications run deeper than statistics. In flat occupations, the career income trajectory is shallow. A cashier who stays in the profession for thirty years will never double their starting pay through skill alone. The path to higher earnings runs through leaving the occupation entirely — getting a different job title, not becoming a better cashier. In high-ratio occupations, by contrast, the same job title can be a career ladder by itself. A financial advisor can go from $47,080 at entry to $215,210 at the 90th percentile without ever changing their BLS code. Whether the ladder inside the occupation is real or illusory determines whether workers should invest in deepening their current skills or pivoting to something new.

When you aggregate the ratios by major occupation group, a clear hierarchy emerges. At the top: Arts, Design, Entertainment & Media, with an average P90/P10 ratio of 3.48. This makes intuitive sense. The arts economy is the original winner-take-all market — a handful of performers, directors, and designers command enormous fees while the median practitioner earns modest wages. A music director at a community theater and a music director for a Broadway production share a BLS code but inhabit different economic planets. The same pattern plays out across fashion designers, film editors, and graphic artists. In the arts, the distribution of pay mirrors the distribution of fame: radically skewed, with long tails.

Legal ranks second (3.28), largely on the strength of judges (4.65) and lawyers (whose ratio runs about 3.42 when including all practice types). The legal profession has always been bifurcated — BigLaw associates in New York start at $235,000, while solo practitioners in rural counties may earn under $60,000 — and the BLS data captures this in stark numerical terms. Management follows at 3.02, driven by the catch-all categories like “Managers, All Other” (3.31) and the wide range of compensation across health services managers (3.14), IT managers (2.80), and marketing managers (2.98).

Sales (3.00) rounds out the four groups above the 3.0 threshold. Commission-based pay structures virtually guarantee high ratios, because the top seller can generate revenue orders of magnitude above the median while the bottom seller scrapes by. Insurance agents (3.73), service sales reps (3.85), and securities sales (4.57) all feature in the most unequal occupations list. Sales is the only major group where compensation is explicitly designed to be unequal — the gap is the incentive.

At the bottom of the inequality ladder sit Healthcare Support (1.91), Production (1.94), Building & Grounds Cleaning (1.98), and Office & Administrative Support (1.99). These are the economy's flat zones — jobs governed by institutional pay scales, union contracts, minimum wage proximity, or all three. In production occupations, the wage structure reflects a manufacturing economy where pay is set by collective bargaining, skill classification, and shift differentials rather than individual performance. An assembler at the 10th percentile and an assembler at the 90th percentile work on the same line; the difference is seniority and overtime, not talent leverage.

Ratios measure relative inequality, but workers cash paychecks in dollars. A 2.0x ratio means one thing when the base is $25,000 (a $25,000 spread) and something entirely different when the base is $80,000 (an $80,000 spread). To see where within-occupation inequality produces the most dramatic real-world differences, we need to look at absolute dollar spreads — the gap between P90 and P10 in actual wages.

The dollar spread table reshuffles the ranking in important ways. Computer and Information Research Scientists appear here despite a ratio of only 2.88 — their high base wage (P10 at $80,670) amplifies a moderate ratio into a $151,450 gap. Petroleum engineers (ratio 2.90, spread $149,950) show the same effect. These are occupations where even the lowest-paid workers earn solid middle-class incomes, but where the ceiling is $150,000 higher. The ratio looks reasonable; the dollar reality is not.

Conversely, some of the highest-ratio occupations produce smaller absolute spreads because their base wages are low. Radio and TV announcers have a dramatic 5.07x ratio, but because their P10 is only $26,000, the spread is $105,780 — large, but not in the top 15. Legislators have the highest ratio in the data (6.76x), but their spread of $117,440 ranks outside the top 10 because their floor is so low. Ratios and dollars tell complementary stories: ratios measure the structure of pay; dollars measure the stakes.

The most consequential entries on this list are the large-employment occupations. “Managers, All Other” employs 631,000 workers with a $158,730 spread. Medical and health services managers employ 566,000 with a $149,400 spread. Securities sales agents employ 472,000 with a $168,130 spread. These aren't niche roles — they're major pools of American employment where your position in the distribution translates to a six-figure difference in lifestyle. A health services manager at the 10th percentile earns $69,680 — a respectable salary that buys a comfortable suburban life in most markets. The same job title at the 90th percentile earns $219,080 — enough for a home in an expensive metro, private school tuition, and aggressive retirement savings. Same title, same LinkedIn summary, different universe.

The occupations that affect the most people aren't the ones with the most extreme ratios — they're the giant ones. The table below focuses on occupations employing more than 500,000 workers, sorted by their P90/P10 ratio. These are the jobs that define American daily life — the cashiers, nurses, truck drivers, teachers, and sales reps that keep the economy running. Their internal inequality structures determine the economic trajectory of tens of millions of families.

The chart tells a striking story about where America's biggest workforces fall on the inequality spectrum. At the unequal end, service sales representatives (1.19 million workers, ratio 3.85) and wholesale sales reps (1.27 million, ratio 3.55) stand out — these enormous workforces have internal spreads exceeding $96,000. In these occupations, the gap isn't driven by geography or credentials alone; it's the commission engine. The rep who lands a Fortune 500 account earns three or four times more than the rep who sells paper to small businesses, all while sharing the same desk layout and the same CRM system.

In the middle tier, software developers (1.65 million, ratio 2.65) exemplify how a high-skill occupation with a high floor can still have meaningful inequality. A developer at the 10th percentile earns $79,850 — already well above the national median — while the 90th percentile reaches $211,450. The $131,600 spread reflects the premium for specific skills (machine learning, infrastructure at scale), employer tier (FAANG vs. local agency), and geography (San Francisco vs. Des Moines). Accountants (1.45 million, ratio 2.68) show a similar structure: the gap between a bookkeeper at a small firm and a partner-track accountant at a Big Four firm spans $88,640.

At the compressed end, the story changes entirely. Cashiers (3.15 million, ratio 1.66) have an entire P10-to-P90 range of just $15,150. Fast food workers (3.78 million, ratio 1.72) span only $16,180. Nursing assistants (1.39 million, ratio 1.60) fit their entire wage distribution into $18,750. These are occupations where a thirty-year career produces almost no pay differentiation from a three-year career — the structure is flat by design, governed by minimum wage floors, institutional pay bands, and the absence of any mechanism to reward individual output.

The largest occupation in the country — retail salespersons, at 3.8 million workers — has a ratio of 1.87 and a spread of $22,330. That means the 90th-percentile retail worker earns just $47,930. Even at the very top of the retail wage distribution, earnings barely cross the national median. This is the reality for nearly 4 million Americans: the ceiling within their occupation is most workers' starting line in other fields.

Perhaps the most surprising finding in the 2024 data is how dramatically within-occupation inequality has narrowed since 2019. Across 710 occupations with valid comparison data, the average P90/P10 ratio fell from 2.684 in 2019 to 2.412 in 2024 — a 10.1% decline. This is not a minor statistical blip. The compression was overwhelmingly one-directional: 630 occupations (88.7%) saw their ratios narrow, while only 80 (11.3%) saw them widen. Zero occupations remained unchanged.

The mechanism is clear from the wage growth data. Across those 710 occupations, wages at the 10th percentile grew an average of 29.4%. At the 90th percentile, the average growth was 15.8%. The bottom outpaced the top by nearly two to one. This is the opposite of the pre-pandemic trend, where decades of research documented widening inequality both between and within occupations. Something changed after 2020 — and the OEWS data captures it with unusual precision.

What changed? Several forces converged. First, the minimum wage surge. Between 2019 and 2024, more than 30 states raised their minimum wages, many aggressively. California went from $12 to $16; Florida phased from $8.56 to $13. Even states without legislated increases saw de facto minimum wage increases as employers competed for scarce workers. These increases disproportionately raised the P10 (the floor of the wage distribution) while leaving P90 relatively unchanged, mechanically compressing the ratio.

Second, the pandemic labor shortage. The occupations with the most compression are disproportionately those that experienced extreme labor scarcity from 2020 to 2023 — food service, hospitality, transportation, healthcare support. When restaurants couldn't find dishwashers at $11 an hour, they offered $15 or $16. When hospitals couldn't find enough nursing assistants, signing bonuses and wage bumps followed. The bottom of the distribution surged because that's where the shortage was most acute. Meanwhile, workers at the 90th percentile already had enough bargaining power to command premium wages before the pandemic; their growth was more incremental.

Third, wage transparency and tight labor markets made it harder to sustain large within-occupation gaps. When job postings are required to show salary ranges (as in Colorado, New York City, Washington, and California), workers at the bottom can see what the top earns. When unemployment is 3.5%, a nursing assistant can credibly threaten to leave for a competitor offering $2 more an hour. The combination of transparency and low unemployment compressed wage structures from the bottom up, exactly the pattern the data shows.

While the dominant story is compression, a handful of occupations moved sharply in the opposite direction — and they share a distinctive profile. Flight attendants saw the largest widening, from a ratio of 2.77 in 2019 to 4.06 in 2024, a jump of 1.29 points. The mechanism is seniority-linked pay in a unionized profession that spent 2020–2021 furloughing junior staff and then rehired en masse. New hires in 2023 started at low base rates while senior attendants accumulated contractual raises through successive agreements. The 10th percentile grew only 16.3%, while the 90th percentile surged 70.5%. The seniority escalator, paused during furloughs, resumed at different speeds for different cohorts.

Gaming dealers widened by 1.16 points, from 2.13 to 3.29. The Las Vegas-led casino boom of 2021–2023 — when pent-up demand produced record gambling revenue — sent tips surging for dealers at high-stakes tables while entry-level positions at smaller regional casinos saw more modest gains. Dealers at Bellagio and Wynn rode a tip tsunami that their counterparts in Biloxi or Atlantic City never experienced. The same dynamic explains bartenders (widened by 1.01 points): the cocktail bar revival in major cities pushed top-percentile earnings far faster than the base rate at a budget chain restaurant.

The waiters and waitresses case is particularly instructive. The ratio went from 2.47 to 3.38, a widening of 0.91 points. The 10th percentile grew only 6.2% — barely above zero in real terms after inflation — while the 90th percentile surged 45.5%. This isn't a story about base wages; it's a story about tips. As fine dining recovered and tip inflation pushed suggested percentages from 18% to 22% or higher, servers at upscale restaurants captured an enormous windfall. Meanwhile, servers at Denny's and Waffle House saw their tipped income barely budge. The same occupation, but entirely different post-pandemic recoveries.

What unites these exceptions? Every one is either tipped or tip-adjacent (bartenders, waiters, gaming dealers), seniority-structured (flight attendants, legislators), or in a profession where a particular segment experienced a localized boom. The 80 occupations that widened are not random noise; they're occupations where specific post-pandemic forces — a luxury spending boom, a tipping culture shift, a seniority-driven contract structure — disproportionately benefited the top of the distribution.

Tipped occupations deserve special attention because they represent a unique form of within-job inequality. Unlike salaried occupations, where the employer sets the pay scale, tipped workers receive a significant portion of their compensation directly from customers. This creates a pay structure that is inherently more variable — and more sensitive to the economic fortunes of the patrons rather than the employer.

Look at the P10 wages for bartenders and waiters: $19,930 and $18,500 respectively. These are among the lowest entry-level wages in the entire economy. Yet the P90 wages — $71,920 for bartenders, $62,510 for waiters — are solidly middle-class, above the national median of $49,500. The gap between these extremes is driven almost entirely by where you work. A bartender at a neighborhood dive in a college town relies on $3 PBR tips from students. A bartender at a rooftop cocktail lounge in Manhattan mixes $22 Negronis for Wall Street bankers, collecting 20% on a $180 round for a table of four. The craft is similar; the context is entirely different.

The 2.3 million waiters and waitresses represent the second-largest occupation in the country, and their P90/P10 ratio of 3.38 means a waiter at the top earns more than three times what a waiter at the bottom makes. But remember: the BLS attempts to include tip income in its wage data, collected through employer payroll records. This means the reported figures likely understate the true inequality, because cash tips — still common in many markets — aren't fully captured. The real P90 for waiters is almost certainly higher than $62,510, and the real ratio is likely above 4.0. Tipping doesn't just create inequality within an occupation; it creates inequality that the official data can only partially measure.

Gaming dealers are the most dramatic case. Their ratio widened from 2.13 to 3.29 between 2019 and 2024, and their P90 wage essentially doubled (from roughly $36,000 to $73,530). This is the tip economy on steroids: a dealer at a $25-minimum blackjack table at Caesars Palace during a post-pandemic Vegas boom can earn tip income that transforms a working-class wage into a middle-class salary. Meanwhile, a dealer at an off-strip casino in Tunica, Mississippi, earns tips that barely supplement their base rate. Same shuffle, same cards, same math — entirely different compensation, driven entirely by the wealth of the people sitting across the felt.

The data points to five structural forces that determine how much inequality lives inside a given occupation. These forces aren't mutually exclusive — many occupations are shaped by several at once — but each one creates a distinctive pay distribution.

1. Compensation structure. The single strongest predictor of within-occupation inequality is whether pay is fixed or variable. Salaried occupations on institutional pay schedules (teachers, postal workers, government employees) have compressed ratios. Occupations with commissions, tips, or performance bonuses (sales, real estate, financial services) have high ratios. This is not accidental; it is designed. Commission structures exist precisely to create unequal outcomes as an incentive mechanism. When an occupation pays $47,000 base plus uncapped commission, the employer has deliberately chosen to outsource wage-setting to the market, and the market produces enormous variation.

2. Geographic variation. Occupations that exist in both high-cost and low-cost markets show wider within-occupation gaps than occupations concentrated in one type of market. Software developers work in San Francisco (P90 territory) and Oklahoma City (closer to P10); the geographic wage premium alone can account for a 2x spread. Postmasters, by contrast, earn federal wages that vary only modestly by locality — hence the tightest ratio in the data (1.34). The more geographically dispersed an occupation, the more its P90/P10 ratio reflects the cost-of-living gradient of America itself.

3. Credentialism vs. skill differentiation. Occupations where a license or credential sets a relatively uniform skill floor (licensed practical nurses, pharmacy technicians, dental hygienists) tend to have compressed ratios. The credential ensures that all practitioners meet a minimum competency, and the pay range reflects seniority and locale rather than talent dispersion. In occupations without credential gates (managers, sales, many tech roles), the talent distribution is wider — and the pay distribution follows.

4. Employer heterogeneity. When an occupation spans many types of employers, inequality widens. Health services managers work at rural clinics ($69,680 at P10) and at university hospital systems ($219,080 at P90). The BLS counts both as the same occupation, but their employers operate at entirely different revenue scales. A restaurant manager at Applebee's and a restaurant manager at a Michelin-starred Manhattan spot share a job title but not an economic universe. The more diverse the employer pool within an occupation, the wider the within-occupation wage range.

5. Winner-take-all dynamics. In some occupations, a small number of practitioners capture outsized rewards because their output generates disproportionate value. Athletes (ratio 3.57, mean-median gap 316%) are the extreme case, but the same logic applies to securities sales, personal financial advisors, producers and directors, and to some extent software engineers at top-tier firms. In these jobs, marginal differences in skill or positioning translate into enormous compensation differences because the economic output at the top is so much higher. A software developer who designs a system that saves a Fortune 500 company $50 million is paid differently from a developer who maintains a WordPress site, even though both write code.