How to read human capital disclosure in Japanese annual reports

Japanese listed companies publish, every year and for free, the numbers I most want about a competitor's people: how many there are, how long they stay, what they earn, whether women get promoted, whether men take parental leave. It's all in the securities report, which anyone can download. When I draw a market map or a talent map, this is the input that is hardest to get anywhere else. Few people read it.

What became mandatory

For decades the employee section of the securities report has carried four numbers: headcount, average age, average tenure and average annual pay. From the fiscal year ending March 2023, a revised Financial Services Agency ordinance added two more blocks.

One block is text. Companies describe a talent development policy and a workplace environment policy, with indicators and targets. The other block is three numbers: the share of women in management, men's parental leave uptake, and the gender pay gap. Strictly, the three numbers are mandatory only for companies already required to publish them under laws such as the Women's Advancement Act. In practice that means nearly every large company.

Six numbers, one portrait

Put the four old numbers next to the three new ones, deal with one scope problem (below), and you can read far more than the company website's "we value our people" will ever tell you.

Start with average tenure, and read it with average age. Tenure of 18 years with an average age of 44 is a company people join and never leave. Promotion goes by seniority, and mid-career hires stay outsiders. Tenure of 5 years with an average age of 35 is a fluid company where joining mid-career is normal. For a recruiter those two are different worlds. Whether people can be moved, how much room there is to negotiate pay, what to say in interviews, all of it changes. How long Japanese stay at one company gives the national yardstick.

Average pay divided by average age is the crudest possible "seniority slope." Take two companies that both pay 8 million yen on average. One has an average age of 46, the other 38. The second one pays its young people far more.

I treat men's parental leave uptake as a proxy for psychological safety. The national figure for 2024 is 40.5%. At a company reporting 80%, men dare to ask and their bosses don't block them. A company at 15% usually has the policy and nobody who dares to use it. That is harder evidence than anything a company says about its atmosphere.

Read the share of women in management together with the pay gap. A high share with a small gap means real promotion. A high share with a large gap usually means women are promoted mostly into low-paid grades. A low share with a small gap may mean few women, treated equally. Nationally, women hold 13.1% of section-chief-and-above roles.

Scope: four numbers for one company

The biggest trap is scope. The same indicator can be reported for the parent company alone, domestic consolidated, overseas consolidated, or fully consolidated. That gives four quite different figures. In Hitachi's 2025 report, the share of women in management runs from 6.8% to 22.7%, and every one of those figures is correct. One company, four percentages for women in management takes that example apart.

So align the scope before you compare two companies. Most listed companies report only the parent-company figure. Put one company's parent figure next to another's consolidated figure and you get the wrong answer.

Five minutes on EDINET, then a table

First, search the company on EDINET, download the latest securities report and open the employee section. That takes five minutes.

Then put the six numbers in a table next to two or three peers in the same industry. That table is the first draft of the function-cluster-by-company layer of the talent map. It is more reliable than any recruiter's verbal judgment.

Then look at the trend. Reports come once a year. Line up three years and you can see whether tenure is getting longer or shorter, and whether men's leave is rising or flat. One year's number describes a company. Three years of change let you judge it.

What it can't tell you

These numbers describe structure. They say little about culture. A company at 80% men's leave can still have a department people want to quit every day. To fill that gap you need other sources, like interviews with people who left and the neutral answers in employee surveys. As a starting point, though, six free numbers are more honest than most paid reports.

Think of the company you researched most recently. What is its average tenure? If you don't know, finding out takes five minutes today.