Source
Every figure comes from the annual securities report (yuho) that each listed company files with the Financial Services Agency through EDINET, Japan's electronic disclosure system. We take each company's most recent report and read the figures from its machine-readable XBRL data rather than its PDF.
The dataset covers 3,789 domestic companies listed in Japan whose latest report was filed between 10 October 2025 and 30 September 2026. Most companies have a March year-end and file in June. A company's page shows which report it uses, with its EDINET document number, so any figure can be checked against the filing.
The unlisted-employer pages and the paid dataset use a second source: MHLW's women's-advancement database (女性の活躍推進企業データベース), where employers post their own figures. The Women's Advancement Act requires every employer with 101 or more permanent employees (常時雇用する労働者) to publish them; the database is the government's recommended place to do so, not the only one, so not every such employer appears in it. It covers unlisted employers, including the Japan subsidiaries of foreign groups, and carries measures the securities report does not: average monthly overtime, paid-leave uptake, women among officers, supervisors and new hires, tenure and ten-year retention by gender, and leave uptake for women. Those figures are self-reported; we apply the same plausibility rules, show each company's reference date, and where a listed company appears in both sources we show both figures. On the free pages we leave out values that cannot be right for an employer of 101 or more: average overtime of zero or above the 80-hour legal ceiling, paid-leave uptake of zero, and average tenure of zero or above 45 years. Where an employer has five or fewer officers, we do not show the share of women among them on the free pages, as it could identify a person. Each figure is shown with the median for employers of the same industry and size band (or, where fewer than ten report it, the same industry). Tenure, women's share of new hires and men's leave uptake are reported for an employment category each employer chooses; we compare them only where that category is the whole workforce or all regular staff, and otherwise show the figure with a note. Used under MHLW's open-data terms (CC BY-compatible); edited by The Hardest Market.
Government certifications and awards (Eruboshi, Kurumin, Youth Yell, the Nadeshiko and Health and Productivity selections, disability-employment and other ministerial awards) come from MHLW's workplace information site, Shokuba Lab (しょくばらぼ, shokuba.mhlw.go.jp), which publishes its full dataset under the government standard terms; matched by corporate number and edited by The Hardest Market. A company page lists a certification only where the site records it as held.
What each figure means
- Employees: the reporting company's own headcount at its year-end, excluding temporary staff, as it states it. For a holding company this is often small, because the staff sit in its subsidiaries.
- Average annual salary, age and tenure: for the reporting company's own employees. Salary usually includes bonuses and overtime pay; the definition is the company's own.
- Women in management: the share of managers who are women, under the Act on the Promotion of Women's Participation and Advancement in the Workplace. Companies define "manager" themselves, usually section manager (kacho) and above.
- Men taking childcare leave: men who took leave in the year as a share of men whose partner gave birth. Companies may count childcare leave only (2,228 do) or childcare leave plus other leave taken for childcare (375); 79 report it under the Women's Advancement Act instead, which counts childcare leave only. The bases are not directly comparable, so each page states the basis. The figure can exceed 100% (179 companies) because leave taken this year can be for a child born the year before.
- Women's pay as a percentage of men's: women's average annual pay divided by men's, for all employees, regular (full-time, indefinite) employees and part-time and fixed-term employees. 100% means parity. These are averages across the workforce, not pay for the same job, so they mostly reflect who holds senior and full-time roles. Read the all-employee figure with the other two: when women are concentrated in part-time work, it can be far below both.
Industry follows the Tokyo Stock Exchange's 33-sector classification. Market segment (Prime, Standard, Growth) comes from the Japan Exchange Group's list of listed companies.
Coverage
Workforce figures are reported by almost every company. The gender measures depend on size. The pay figure has been compulsory for employers with 301 or more employees since July 2022, and men's leave uptake for those with more than 300 since April 2025 (more than 1,000 before). Women in management was optional until April 2026, when it and the pay figure became compulsory from 101 employees; for most companies those first disclosures arrive in mid-2027. Until then, many smaller companies leave the gender measures out.
| Figure | Companies | Share | Of which, a subsidiary's figure |
|---|---|---|---|
| Employees (parent company) | 3,753 | 99% | — |
| Average annual salary | 3,742 | 99% | — |
| Average age | 3,744 | 99% | — |
| Average tenure | 3,743 | 99% | — |
| Women in management | 3,035 | 80% | 250 |
| Men taking childcare leave | 2,726 | 72% | 327 |
| Women's pay as % of men's — all employees | 2,922 | 77% | 251 |
| Women's pay as % of men's — regular employees | 2,908 | 77% | 252 |
| Women's pay as % of men's — non-regular employees | 2,766 | 73% | 316 |
2,539 companies (67%) report all three gender measures.
Holding companies
A holding company with few employees of its own often reports the gender measures for its operating subsidiaries instead. Where the parent gives no figure, we show the first subsidiary it lists, usually its largest, and the company page names that subsidiary. We do not combine subsidiaries into a group figure.
What we change, and how we mark it
We publish figures as filed, except where the filing is plainly wrong. Every change is marked on the company page, and the figure as filed is kept in the data.
- Salaries in the wrong unit. Some companies enter average salary in thousands of yen in a field that expects yen. Where the corrected figure is plausible, we correct it (10 companies).
- Implausible values. A value that cannot be right (a negative pay ratio, a share of managers above 100%, a salary that no unit correction makes plausible) is left out rather than guessed at (28 companies).
- Pay gap filed as the difference. The law asks for women's pay as a percentage of men's. A few companies file the difference instead: 30 where the ratio is 70. We flag every unusually low pay figure and check it against the filing's own notes and the company's other disclosures. Where it is clearly the difference, we convert it: Emimen Co., Ltd., Mercari, Inc. Where the figure is low but real, we keep it: Global Link Management Inc., GLOME Holdings, Inc., Nittoh Corporation, Poppins Corporation, Premium Water Holdings, Inc., Strike Group Co., Ltd., T&D Holdings, Inc.
- Men's leave filed as a ratio. The uptake rate is a percentage, but some companies file it as a ratio (2.167 for 216.7%). Where the figure can only be a ratio, or the company's own entry on the MHLW women's-advancement database confirms it, we convert it (10 companies). Where neither reading is clearly right, the figure is withheld (13).
- Under review. Where we cannot tell from the filing, we withhold the figure until it can be settled: BayCurrent, Inc., Soliton Systems K.K.
Company names
Listed companies appear under the English name in the Japan Exchange Group's list, or the English name in their EDINET filing. Subsidiaries appear under their own English name where they publish one; otherwise we romanise the Japanese name in modified Hepburn. We change only capitalisation and punctuation.
Unlisted employers appear under the English name they have registered with the National Tax Agency where there is one. Otherwise their page shows an English rendering of the Japanese name, labelled as such: katakana loanwords spelled as the English word, Japanese words romanised, and the legal form (株式会社, 医療法人 and so on) given in English. A rendering is not the company's own English name, and the Japanese name stays on every page.
Boards, targets and executive pay
The same annual securities reports list each company's officers, with their titles, dates of birth and career summaries, and set out the human-capital metrics and targets the company has chosen. From these we show the share of women among officers, the board's age profile, whether an officer's title or current role names HR, executive pay as a multiple of average salary, and the company's own women-in-management target against what it reports. Free pages show counts and shares only; names appear only in the paid data.
MHLW market statistics
Several further MHLW sources sit beside the company data, among them the Survey on Employment Trends, foreign-worker notifications, the General Survey on Working Conditions, the Survey on Wage Increases and regional minimum wages. Pay by occupation comes from the Basic Survey on Wage Structure, MHLW's annual wage census: full-time employees at private establishments of companies with 10 or more staff, with estimated annual pay as monthly contractual pay times twelve plus the previous year's bonus. The job market page uses Employment Referrals for General Workers, MHLW's monthly count of openings and applicants at Hello Work, seasonally adjusted for the national and prefecture series. The paid dataset also carries MHLW's current list of labour-law cases referred to prosecutors; a referral is not a conviction. All three are refreshed when MHLW publishes.
Updates
We refresh the data as each year's reports are filed. The date at the top of this page is the latest refresh. A company's page always reflects one report, so its figures are internally consistent but may be up to a year old.
Licence and credit
EDINET content is published under the Public Data License 1.0. This dataset is an edited version: the figures were extracted, standardised, checked and translated by The Hardest Market. It is not endorsed by the Financial Services Agency, and the original filings are authoritative.
Corrections
If a figure is wrong, or you are a company that would like a figure reviewed, write to editor@hardestmarket.com with the company and the EDINET document number shown on its page (every company and employer page has a link that does this). We check each request against the original filing or MHLW entry and reply within ten business days; a confirmed error is corrected at the next build and noted in the dataset's corrections log. Foreign-ownership labels are reviewed the same way: we confirm ownership from the company's own disclosures, its parent's filings and the commercial registry, and a company can ask us to review its label at any time.