Locale: en
As of September 30, 2026. Whenever social media debates whether ¥6 million a year is “high” for someone in their twenties, two universes collide: “Everyone around me makes that much” versus “That is rare in your twenties.” The funny part is that both can be true inside each person’s local bubble. The error begins when a local bubble is promoted to a national statistic.
A classic salary argument goes like this.
“¥6 million in your twenties is pretty high.”
Someone replies:
“What? If I gathered 30 university friends, almost none would be below ¥6 million.”
Welcome to the denominator lottery.
Thirty friends are not thirty random Japanese workers.
People cluster by university, education, recruiting channels, industry, city, and company size. Social networks are aggressively non-random.
So ¥6 million can be perfectly ordinary inside one cluster without being ordinary nationwide.
And someone in a low-paying sector can sincerely feel that ¥6 million is mythical without representing the whole country either.
¥6 million is Schrödinger’s “normal”: the answer changes when you open the box labeled “population.”
1. Before asking whether ¥6 million is high, ask: among whom?
Japan’s National Tax Agency reported that among 51.63 million private-sector salary earners who worked throughout 2025, average annual pay was ¥4.87 million.
The average was ¥5.97 million for men and ¥3.42 million for women.
Adding the published salary bands shows that 25.9% earned more than ¥6 million. So for all full-year private-sector salary earners, across ages and sexes, more than ¥6 million is roughly the top quarter.
Calling it “top 20%” is a little too strong. More than ¥7 million was 18.1%, which is much closer to the top fifth.
The official bands are “over ¥5 million to ¥6 million” and “over ¥6 million to ¥7 million,” so 25.9% strictly means more than ¥6 million, not “¥6 million or more.” Nobody needs a civil war over the worker earning exactly ¥6,000,000, but statistical boundaries should still match the table.
The same salary also looks different by sex: about 37.3% of men and about 10.5% of women earned more than ¥6 million.
The first button in a salary debate is not the amount.
It is the denominator.
2. Do not paste an all-age percentile onto a debate about people in their twenties
If the question is “¥6 million in your twenties,” the all-age 25.9% figure cannot answer it.
In the NTA’s 2025 data, average pay was ¥2.80 million at ages 20–24 and ¥4.25 million at ages 25–29. At 25–29, it was ¥4.60 million for men and ¥3.83 million for women.
So ¥6 million is clearly above the average for people in their late twenties.
But an average does not tell you an exact percentile.
To calculate the precise share of people in their twenties above a threshold, you need an age-by-income cross-tabulation. Japan’s 2022 Employment Status Survey table 10200 provides five-year age groups such as 20–24 and 25–29 together with income bands such as ¥6.0–6.99 million and ¥7.0–7.99 million.
The discipline here is simple: do not manufacture a percentile you have not actually calculated.
We know that more than ¥6 million is 25.9% across all ages.
We know the average at 25–29 is ¥4.25 million.
Those two facts do not magically generate “top 7% in your twenties.”
Statistics is not a number-growing simulator.
3. The company lottery is real, but not every difference is caused by the company
The intuition that “similar people can end up in very different pay environments depending on the employer” has real empirical support.
In Japan’s 2025 Basic Survey on Wage Structure, monthly scheduled wages for ages 25–29 were ¥301,700 at large firms, ¥270,600 at medium firms, and ¥258,200 at small firms. The survey defines large firms as 1,000 or more regular workers, medium as 100–999, and small as 10–99.
Even at this age, large-versus-small differs by roughly 17%.
The NTA’s company-capital breakdown shows another large descriptive gap: average annual pay was ¥4.16 million at corporations with capital below ¥20 million and ¥6.92 million at corporations with capital of ¥1 billion or more.
Industry differences are huge too: average 2025 pay was ¥8.54 million in electricity/gas/heat/water, ¥7.26 million in finance/insurance, and ¥2.76 million in accommodation/food services.
Regular employees averaged ¥5.59 million, versus ¥2.10 million for employees outside the regular category.
But these are not experiments where identical workers are randomly swapped between employers. Age, occupation, education, hours, location, selection, and many other factors differ.
So both statements can be true:
“Employer and sector matter a lot.”
“Not every observed gap is caused by the employer alone.”
A RIETI study of male full-time workers from 1995–2013 found that rising between-firm wage variance played a major role in the rise of overall wage inequality, while within-firm variance was broadly flat.
That evidence is historical and limited to a specific worker population. It should not be pasted onto all workers in 2026. But it does support the idea that which firm you land in can matter.
The company lottery is not imaginary.
The mistake is winning an SSR employer and concluding everyone lives in the SSR village.
4. “Almost all 30 of my friends make ¥6 million” can be true and still tell us little about Japan
This is the best part of the argument.
“Almost all of my 30 friends make at least ¥6 million.”
Maybe they do.
Truth and representativeness are different questions.
University friends are not a random sample.
A selective university clusters academic background and family background.
Shared recruiting information clusters employers.
Staying in large cities clusters regional pay.
Entering finance, technology, consulting, or major manufacturers clusters pay levels.
Friends also introduce friends.
Soon, “¥6 million is normal within 50 meters of me” becomes perfectly plausible.
That observation can be accurate.
The invalid step is:
“It is normal around me” → “Therefore it is normal in Japan.”
The sample just teleported.
The reverse error is identical. Someone surrounded by low-paid work can conclude that ¥6 million is only for a tiny elite, even though their local sample is also selective.
Human intuition is a live broadcast from your cluster.
It is not a census.
5. Biased local perception does not mean real inequality is fake
This distinction matters.
Saying “your 30 friends are a biased sample” does not imply that wage inequality is an illusion.
Sampling bias and genuine inequality can coexist.
If nearly everyone in your social network works as a regular employee at large firms, low-paid workers may barely appear in your daily social field.
Yet people can work in the same building, factory, or city under very different employers, contracts, occupations, and hours.
You generally do not see another person’s employment contract.
You do not see their payslip, bonus, pension, housing subsidy, overtime, stock plan, or severance package.
So a real labor-market gap can exist while remaining visually invisible in everyday life.
“Invisible” is not the same as “nonexistent.”
6. Would publishing everyone’s salary make us happier? Not necessarily
Knowing more is useful.
It can also make you miserable.
A 2012 experiment by Card, Mas, Moretti, and Saez informed a randomized subset of University of California employees about a website where coworkers’ salaries could be searched.
Employees paid below the median for their unit and occupation reported lower pay satisfaction and job satisfaction and showed greater intention to search for another job.
Employees above the median did not show an equally strong positive effect.
In crude form:
“I make less than them” → hurts.
“I make more than them” → does not pay an equal happiness dividend.
The comparison game has terrible rewards.
A person can be satisfied with life at breakfast, learn a friend’s salary at lunch, and feel underpaid by dinner.
Their bank balance did not change by one yen.
Only the reference point moved.
The study was conducted in a particular workplace setting and should not be treated as a universal psychological law. But it is strong evidence that salary transparency can have psychological side effects.
7. Total ignorance is risky too: compare yourself with the market, not with people
Should we therefore avoid all salary information?
No.
Zoë Cullen’s 2024 review of pay transparency emphasizes that not all transparency is the same.
Horizontal transparency—seeing what coworkers make—can affect peer comparisons and internal wage setting.
Cross-firm transparency—knowing what other employers pay—can support job search, negotiation, and wage competition across firms.
The useful question is therefore not:
“Friend A makes ¥7.13 million and coworker B makes ¥6.48 million. Where do I rank?”
It is:
“For my role, experience, region, and working hours, what is the market range?”
Those are different information products.
The first easily turns into a human leaderboard.
The second is decision data.
Compare with the market, not with people.
8. Treat compensation as a contract package, not one annual number
Even ¥6 million is not one lifestyle.
At the same salary, one job may require 40 hours of monthly overtime and another five.
One may require relocation.
One may include housing support.
One may have volatile bonuses.
One may include severance, generous leave, remote work, stability, or autonomy.
One commute may be ten minutes; another ninety.
If salary alone becomes the ranking, a ¥7 million job that consumes your life can automatically “beat” a ¥6 million job that gives you time and freedom.
That is bad measurement.
When comparing the market, look at total compensation, hours, benefits, location, stability, workload, and autonomy.
The purpose of salary information is not to beat another human.
It is to decide whether your employment contract buys the life you actually want.
9. Conclusion: ¥6 million is Schrödinger’s normal. Label the box first
Is ¥6 million a high salary?
Write down the population first.
Across full-year private-sector salary earners of all ages, more than ¥6 million is about 25.9%—roughly the top quarter.
For a debate about workers in their twenties, you cannot reuse that all-age rank. Average pay at 25–29 is ¥4.25 million, so ¥6 million is clearly above average, but an exact percentile requires an age-by-income table.
Inside clusters of large firms, high-paying industries, particular universities, and big cities, ¥6 million can still feel perfectly ordinary.
That local observation does not have to be false.
It is simply not a national estimate.
And you do not need every friend’s payslip to understand the labor market.
A useful information diet is:
Keep person-to-person comparison minimal; keep market comparison informed.
You do not need thirty friends’ salaries to estimate the Japanese economy.
That is not a survey.
That is a drinking party with a spreadsheet.
Sources checked September 30, 2026
- National Tax Agency, 2025 Private Sector Salary Survey https://www.nta.go.jp/publication/statistics/kokuzeicho/minkan/gaiyou/2025.htm
- National Tax Agency, 2025 survey report https://www.nta.go.jp/publication/statistics/kokuzeicho/minkan2025/pdf/2025_001.pdf
- Ministry of Health, Labour and Welfare, 2025 Basic Survey on Wage Structure https://www.mhlw.go.jp/toukei/list/chinginkouzou_a.html
- e-Stat, 2022 Employment Status Survey, Table 10200 https://www.e-stat.go.jp/index.php/stat-search/database?statdisp_id=0004008241
- RIETI, Rise in Wage Inequality between Firms: Evidence from Japan 1995-2013 https://www.rieti.go.jp/jp/publications/summary/24030002.html
- Card, Mas, Moretti & Saez, Inequality at Work https://www.aeaweb.org/articles?id=10.1257/aer.102.6.2981
- Zoë Cullen, Is Pay Transparency Good? https://www.aeaweb.org/articles?id=10.1257/jep.38.1.153
