Two Averages and One Line

The line is one number: women's average annual pay divided by men's, times a hundred. Toyota's note defines it that way — 女性労働者の平均年間賃金÷男性労働者の平均年間賃金×100％ — and Sony Group's note says the same in prose, the average wage of women shown as a proportion of the average wage of men. It sits in the employee section of the annual securities report, 従業員の状況, next to the headcount and the average salary read in the first piece of this series, under a heading that names three indicators: the share of women in management, the take-up of parental leave by men, and 労働者の男女の賃金の差異, the difference in wages between men and women. It is printed three times, for all workers, for regular workers and for part-time and fixed-term workers.

The line arrived with the January 2023 amendment to the Cabinet Office Ordinance on Disclosure of Corporate Affairs, applied to fiscal years ending on or after 31 March 2023. The amended item (29), 従業員の状況, asks in sub-items d, e and f for the three figures of the filing company and of each of its consolidated subsidiaries — 提出会社及びその連結子会社それぞれ. Sub-item f, the pay line, is defined by reference to the figure an employer publishes under the Act on Promotion of Women's Participation and Advancement in the Workplace, 女性活躍推進法, and where the filer or a subsidiary does not publish that figure under the Act, the entry may be omitted; the other two sub-items carry provisos of their own. Sub-item g lets the figures for subsidiaries other than the principal ones be moved to Part 7 of the report, 提出会社の参考情報, with a cross-reference.

So the answer to the question this piece set out with — is the figure the parent's or the group's — is neither, as the ordinance is written. It is organised by entity — the filing company on one row and each named subsidiary on its own, with no consolidated line required — and the population behind a row is whatever the report's notes say it is: Honda's note says its filing-company row includes four other companies covered by its labour agreement, and NTT's note says its Docomo row includes NTT Docomo Business and its Data Group row includes NTT Data and NTT DATA, Inc. Five of the thirty-five reports, as read here, add a group aggregate of their own beside the entity rows; they are named below. The thirty-five are the same June 2026 annual securities reports read in the two earlier pieces of this series, and each report's table for the filing company was read as a table, the value taken from under the printed column heading.

Thirty-two of the thirty-five print the line for the filing company. Three print a sentence in its place. Mitsubishi UFJ Financial Group writes 提出会社は、各指標を公表しておりません — the filing company does not publish the indicators — and then lists three subsidiaries, beginning with MUFG Bank at 53.2. Sumitomo Mitsui Financial Group writes that it is not required to publish the indicators under the Act and omits them, then lists Sumitomo Mitsui Banking Corporation at 46.9. FUJIFILM Holdings writes that the filing company is not subject to the publication duty and omits the entry, then lists FUJIFILM Corporation at 74.6. In each case the sentence is the company's own.

Across the thirty-two filers that print the all-worker line, the lowest figure is FANUC's 40.0, then HOYA's 40.1, and the highest is East Japan Railway's 88.9. Seven are below 60, nineteen are in the sixties and seventies, and six are at 80 or above. Toyota prints 67.0, Sony Group 82.8, Keyence 43.2 and Recruit Holdings 85.8, each for the filing company alone.

Keyence's table has one row for the filing company. The heading over it names two indicators, 男性労働者の育児休業取得率及び労働者の男女の賃金の差異, and the row under it has four cells as of 20 March 2026: 84.7 for men's parental leave, then 43.2, 43.7 and 92.0 for the pay line across all workers, regular workers and part-time and fixed-term workers. The same page prints the filing company's headcount, 3,306, under 提出会社の状況. The two notes under the table name the two laws under which the figures are calculated.

HOYA prints its line on a row headed 当社, the company: 40.1 for all employees, 58.6 for regular employees, 84.5 for non-regular employees. Its table carries a column the ordinance does not ask for, the share of women among regular employees, 33.2, beside the share of women in management, 18.1. A second row, 当社及び国内連結会社 — the company and its domestic consolidated companies — reads 41.1, 59.8 and 80.1, and the one subsidiary listed between them, HOYA Technosurgical, prints 69.2.

Recruit Holdings prints the line on a row headed 提出会社 and says in its first note that it is under no publication duty under either law and discloses the figures voluntarily, as reference information. The row reads 55.0 for women in management, 200.0 for men's parental leave, and 85.8, 86.5 and 134.6 for the pay line — the only filer of the thirty-two whose part-time and fixed-term figure is above 100. Its third note says that most of the filing company's workers are seconded from its subsidiary 株式会社リクルート, and that the figures include staff seconded in from other companies and exclude staff seconded out, so that a figure for the company's own registered staff alone would differ. That subsidiary prints 74.9 on the same page; Recruit Staffing prints 74.6, Staff Service Holdings 71.0 and Indeed Recruit Partners 82.6, and twenty-three other subsidiaries are referred to Part 7.

Toyota prints 67.0 for all workers, 66.8 for 正社員 — its word for regular employees — and 59.7 for part-time and fixed-term employees, and then, in its seventh note, breaks the regular figure down by job type for employees aged 30: 93.5 for engineering and clerical staff, 82.0 for skilled staff, 87.1 for medical staff, and no figure for business staff because there are no men in that category. The note says the company's pay system makes no distinction by sex and attributes the regular-worker difference to average age and to the headcount of each job type; that is the company's account, printed under its own table. Nine principal subsidiaries follow on the next page — Hino Motors at 78.7, Daihatsu at 78.2, Woven by Toyota at 78.4 — and the rest run from page 219 to page 225 of a 234-page report.

Sony Group's row is headed ソニーグループ㈱ and reads 82.8, 83.4 and 66.9, with 20.5 for women in management and 84 for men's parental leave, all printed with a percent sign. The entity on that row is the one whose headcount the same report gives as 2,166 against a consolidated 94,900 — the holding-company row the first piece of this series was about. Nine principal domestic subsidiaries share the page: Sony Interactive Entertainment 80.6, Sony Music Entertainment 68.1, Sony Corporation 81.6, Sony Semiconductor Solutions 73.8. The rest are on page 277, where Aniplex prints 63.6 and Polyphony Digital prints 61.4 for all workers and 該当なし, not applicable, for regular workers.

Five reports print a group figure of their own next to the entity figures. Hitachi's first table is headed 当社及び当グループ, the company and the group, and reads 82.4 for all employees, split into 72.9 for the consolidated group in Japan and 91.2 for the consolidated group outside Japan; the filing company, 株式会社日立製作所, follows on the next page at 71.6. KDDI prints 83.3 for the filing company and 76.2 on a row headed KDDIグループ連結. Mizuho Financial Group prints 62.2 for the filing company and 53.0 for the filing company and its four principal subsidiaries combined. NTT prints 79.1 for the filing company and, under a heading marked （参考）任意の連結グループ — for reference, a consolidated group of its own choosing, which its note lists as the company, NTT Docomo, NTT Data Group, NTT East and NTT West, with NTT Docomo Business counted under Docomo and NTT Data and NTT DATA, Inc. under Data Group — 80.3. HOYA's 41.1 is the fifth. A group figure can sit on either side of the filer's own.

The bank holding companies show what the entity rule does to a group whose people are in the subsidiaries. Mitsubishi UFJ Financial Group's three rows are MUFG Bank 53.2, Mitsubishi UFJ Trust and Banking 69.6 and Mitsubishi UFJ Morgan Stanley Securities 62.6, each followed in brackets by the change from the prior year — +1.5 for the bank — and a second table gives the bank's regular workers by tier: 88.4 for executive grades, 81.2 for staff with a title, 83.0 for staff without one. Sumitomo Mitsui Financial Group's first row is the bank at 46.9 for all workers and 53.5 for regular workers, with a dash and a note where the part-time figure would be. Mizuho's figures carry a second number in brackets, 62.2 (60.2), which its note explains as the prior year recalculated on this year's revised basis; the basis, it says, was changed this time in the wages counted, the workers counted and the method.

A number of reports add a table the ordinance does not ask for: the same ratio inside a grade. ITOCHU prints 56.9 for all employees and then, by rank, 98.5 for non-managerial staff, 97.1 at manager level, 99.4 at deputy section-chief level and 97.3 at section chief and above, with the share of women at each rank beside it — 35.6, 18.1, 11.6 and 2.3. Mitsui & Co. prints 64.9 for all workers and 91.0 for managers and 92.5 for non-managers, with women at 12.0 and 61.9 of those groups. Tokyo Electron prints 60.6 for the filing company and says on the page before its table, in the same employee section, that, for the domestic group, the figure is 89.8 for managers and 90.1 for non-managers in the career track and 114.9 for the general-clerical track. FUJIFILM Corporation's regular workers split into 97.3 for those in management positions and 82.2 for the rest; Hitachi's group figure for indefinite full-time employees, 83.5, splits into 90.8 for managers and 86.7 for non-managers; Mitsubishi Corporation's 65.2 for regular employees becomes 73.9 for the career track alone. East Japan Railway prints 88.9 for all workers and 86.0 for regular workers, and then, under a heading that names regular workers, the figure by age: 103.3 in the twenties and below, 94.2 in the thirties, 93.5 in the forties, 98.7 in the fifties.

Where a report explains its figure, the explanation is printed with the table — under it, beside it in a column headed 補足説明, or on the page before — and this piece repeats only what is printed. Nintendo's note attributes its 72.7 to differences in length of service and average age, and says its pay and evaluation systems treat men and women alike. FANUC prints 40.0 for all workers, 42.8 for regular workers and 47.5 for non-regular workers, and beside the aggregates prints the regular figure by category — 81.3 for senior staff, 84.9 for general staff, 100.7 for indefinite-contract staff — and 100.2 for fixed-term and part-time staff; its note says that about six in ten of its women are factory contract workers, which it names as the main reason for the all-worker and regular-worker figures. Daikin's note gives women's average length of service as 12.3 years against men's 17.6. Mizuho lists two reasons: the low share of women among managers, whose pay is relatively high, and a difference in hours worked. SoftBank Group's note attributes the difference in total remuneration to the number of men in senior positions. SMC's note lists two: women's shorter average service, which it links to increased hiring of women in recent years, and the low share of women in management.

The three columns are not labelled the same way. Twenty-seven of the thirty-two filers head the first column 全労働者, all workers, and five head it 全従業員, all employees. Twenty-eight head the second with a word containing 正規, regular; Toyota writes 正社員, Oriental Land うち社員, Recruit うち無期雇用従業員 — employees on indefinite contracts — and Hitachi 無期雇用・フルタイム従業員, indefinite and full-time. The third is headed with パート in twenty-one reports and 非正規 in eight; Mitsui & Co. writes 有期雇用労働者, fixed-term workers, ITOCHU うち有期雇用従業員, and Oriental Land うち嘱託社員、出演者、準社員等 — contract staff, performers, associate staff and others. Mitsubishi Corporation and Kyocera print the all-worker column last; Kyocera's row reads 72.9, 34.6 and 55.9 in the printed order, and 55.9 is the all-worker figure. In seven reports the regular-worker figure is below the all-worker figure — Toyota, NTT, KDDI, Daiichi Sankyo, Murata Manufacturing, East Japan Railway and Central Japan Railway — because the all-worker line is a ratio of two averages over a different mix, not an average of the two columns beside it.

The line also comes in more than one layout. Mitsui & Co. prints it as a table of two fiscal years, 59.9% for the year to March 2025 and 64.9% for the year to March 2026, with regular workers moving from 60.4% to 65.7% and fixed-term workers from 54.1% to 56.9%. Central Japan Railway prints it as three rows in a list of indicators, 75.9%, 74.9% and 78.9%. Recruit's figures stand on a line of text under a column heading. And in several reports — Toyota, Sony Group, Honda, NTT, Oriental Land, Sumitomo Mitsui Financial Group, Panasonic Holdings, FUJIFILM Holdings and Recruit among them — a note sends the reader to Part 7 for the subsidiaries that are not principal ones: Toyota's occupy pages 219 to 225, Sony Group's page 277, FUJIFILM Holdings' page 214 and Recruit's pages 216 and 217.

What the line is, then: one company's women's average pay over its men's, for one year, for the population the report defines — the filing company alone in most of the thirty-two, though Honda's note says its filing-company row includes four companies covered by its labour agreement, 本田技術研究所, ホンダ・レーシング, ホンダ学園 and ホンダアクセス — calculated under the ordinance of the Act and printed where the ordinance says. Thirty-two of the thirty-five large March-year filers read here print it for themselves; three say in their own words that the filing company does not. What it is not is a comparison between companies. Hitachi counts secondees out and staff on leave in; Recruit converts part-time staff to eight-hour equivalents; Mizuho changed its basis this year and printed both; Toyota counts part-timers by head rather than by hours. Each report says what it counted, and the number on the row is the number for that row.