Did average pay in Japan really rise?

3,608 listed companies, two years of filings

Average annual pay is the headline number on this site. It is also total payroll divided by headcount, so it rises when low-paid junior staff leave, even if nobody gets a raise. Japan has seen widely reported wage increases since 2024. We checked whether the numbers in the filings are real raises or a change in who is counted.

Result: mostly real. Composition does not explain it

82% of companies raised average pay, median +5.7% over two years. The correlation with the change in average age is weak (ρ=+0.129), and age, tenure and headcount together explain only 5% of the variation. Whatever else is going on, "the young ones left" does not account for it.

1. Split by change in average age

Quintile Δ average age (yrs)Δ average pay Share that rosen
Q1-1.1+4.0%71%721
Q2-0.2+5.6%85%722
Q3+0.3+6.0%87%721
Q4+0.8+6.3%87%722
Q5+1.8+6.5%83%722

The gradient is real but small: a 2.9-year spread in ageing maps to 2.4 percentage points of pay. In the regression, one extra year of average age is worth +1.2% — not enough to produce the +5.7% observed.

2. Split by change in headcount

Hiring should drag the average down, since new joiners are paid less.

Quintile Δ headcountΔ average pay Share that rosen
Q1-9.3%+5.2%79%600
Q2-1.4%+6.2%86%601
Q3+2.7%+5.7%87%601
Q4+7.8%+6.2%83%601
Q5+23.1%+5.6%79%601

Essentially flat (ρ=+0.049). Companies that grew headcount by 23% still raised average pay +5.6%. The raises outweighed the dilution from hiring.

Limitations

Source: annual securities reports (EDINET, Financial Services Agency of Japan). Method: pipeline/analyze_raise.py. Japanese edition with the full tables: 平均年収が上がった会社では、何が起きていたのか.