“Half of Japan earns ¥2.5 million or less!” The social-media post had a frightening headline and a neat table. Even worse, the 51.3% figure in the table was real.
The problem wasn't a fake number. It was a real number wearing the wrong name tag.
Real table, different question
The figure came from Japan's Ministry of Health, Labour and Welfare and its 2023 Income Redistribution Survey, released in December 2025. The income reference year was 2022, and the study tabulated 3,003 households.
Adding the table's income brackets below ¥2.5 million gives 51.3%. But those aren't individual salaries. The table classifies households by initial income, a technical measure used before accounting for taxes and public benefits.
A one-person household and a four-person household each count as one household. Replace “households” with “Japanese people,” and you have silently changed what is being counted.
Replace “initial income” with “salary,” and you've changed it again.
The pension income that vanished from the caption
Japan's initial-income measure includes earnings and other specified private income. Public pensions are treated separately as social-security benefits.
Imagine a retired household living mainly on a public pension. It can receive money for daily life while appearing in a low initial-income bracket. That's useful when studying what public benefits change. It's not useful as a shortcut for declaring the household's working salary.
The ministry also publishes income after redistribution. For households below ¥500,000, the shares were 30.4% before and 1.2% after redistribution.
That second number isn't take-home salary either. Redistributed income includes the estimated value of services such as healthcare and long-term care, not just cash deposited into a bank account.
The two measures answer different questions. Picking one and calling it “everyone's pay” answers neither.
You don't have to fake a chart to mislead someone
The viral-sounding claim can be built from legitimate ingredients:
Start with household data. Call them people. Take initial income. Call it salary. Leave the public-pension explanation outside the frame.
Click. A genuine statistical table now looks like a very different story.
We cannot know from a screenshot whether the person sharing it intended to mislead anyone or simply misunderstood the source. The reader can still spot the mismatch.
Before sharing, read the name tag
Ask three questions: Who is being counted? Which kind of income? What's included and excluded?
That's enough to catch the central error here. You don't need to call the 51.3% fake. You need to say what it actually is.
The number made it into the photograph. Its instruction manual didn't.
References (2)
- MHLW survey announcement (December 23, 2025): https://www.mhlw.go.jp/stf/houdou/96-1_r05kekka.html
- MHLW full report: https://www.mhlw.go.jp/content/12605000/R05hou.pdf
