In spring 2026 Rengō reported a 5.26% shuntō wage settlement (first tally, as-of 23 Mar 2026). The government's Monthly Labour Survey put realised cash-earnings growth at +3.4% (February 2026 final). These are two different defined quantities measuring one wage-growth idea — different reference windows, different populations. This brief places the two on one sheet and decomposes which pay rise each number actually describes.
Rengō's 1st-tally shuntō settlement (as-of 23 Mar 2026) — the negotiated raise won by unionised, large-firm-weighted workers — was 5.26%. The MHLW Monthly Labour Survey put realised total cash earnings, across all establishments of 5+ employees, at +3.4% (February 2026 final). Two different reference windows, two different defined quantities measuring one wage-growth idea — a ~1.9-point gap (RAOSCAFF arithmetic). Both numbers are correct under their own definitions.
Rengō — Japan's trade-union confederation — put the 2026 spring-round average wage settlement at 5.26% in its first tally (as-of 23 Mar 2026), the third consecutive year above 5%. The line reads like a fact about the country. It is a fact about one slice of it: the unionised, large-firm-weighted workforce whose negotiated raises Rengō counts.
The Ministry of Health, Labour and Welfare measures a different quantity under a different lens. Its Monthly Labour Survey put realised total cash earnings — across every establishment of 5+ employees, non-union SMEs and part-timers included — at +3.4% YoY (February 2026 final). These are two different reference windows and two different defined quantities measuring one wage-growth idea. That is a gap of roughly 1.9 percentage points (RAOSCAFF arithmetic: 5.26 − 3.4 = 1.86pp): the settlement runs about ~54.7% higher in relative terms (RAOSCAFF arithmetic: 5.26 ÷ 3.4 − 1 = 0.547).
Both numbers are correct. The gap is what you would expect when a large-firm union settlement is set beside an all-establishment earnings average — a coverage-and-method difference, not a discrepancy, and certainly not leakage or any fault of either body. Both are positive: this is a gap of size, not direction.
Set the two numbers side by side. The 5.26% settlement (Rengō, 1st tally as-of 23 Mar 2026) is itself two things — a base-pay floor lift, and a seniority escalator stacked on top. The +3.3% provisional MHLW figure (final: +3.4%, February 2026) is a different instrument entirely — the all-establishment survey lens. The dotted bracket marks the ~1.96pp gap between the settlement top and the provisional survey read (RAOSCAFF arithmetic: 5.26 − 3.3 = 1.96pp).
The 5.26% moves along three axes at once. Population: it is the negotiated raise for unionised, large-firm-tilted members — not realised earnings across all establishments under the MHLW all-establishment survey lens, which puts the figure at +3.4% (February 2026 final); the ~1.9pp gap is RAOSCAFF arithmetic (5.26 − 3.4 = 1.86pp). Component: only 3.85% (¥13,013) is a base-pay floor lift; roughly 1.4pp is the seniority escalator — the automatic annual step up the existing pay grid, which on its own does not raise the firm's overall scale (RAOSCAFF arithmetic: 5.26 − 3.85 = 1.41pp). Reference moment: 5.26% is the early, large-firm-heavy 1st tally (as-of 23 Mar 2026); by the 3rd tally (as-of 1 Apr 2026) the overall figure stood at 5.09%.
| The number | What it measures | Population / scope | Source |
|---|---|---|---|
| 5.26% · Rengō, 1st tally, as-of 23 Mar 2026 | negotiated wage settlement | unionised, large-firm-tilted workforce | FACTS § H-2 |
| +3.4% · MHLW, Feb 2026 final | realised total cash earnings | all establishments, 5+ employees | FACTS § H-1 |
| 3.85% (¥13,013) | base-pay (ベア) floor lift | the structural part of the 5.26% | FACTS § H-2 |
| ~1.4pp | seniority escalator (定昇) | automatic step up the existing grid | RAOSCAFF arithmetic |
| 5.09% / SME 5.00% · 3rd tally, 1 Apr | the same settlement, later | overall & SME, as smaller firms report | FACTS § H-6 |
The number that compares cleanly to a cost-of-living question is the 3.85%, not the 5.26% — the headline carries the seniority escalator inside it. And a one-year step back sharpens the point: the prior year, a 5%-plus settlement headline coexisted with real wages −1.3% (2025 final, the fourth consecutive negative real-wage year), because the settlement and the realised-earnings-after-inflation answer different questions. 2026 turned mildly positive — real wages +2.0% in Feb 2026 final.
If you read one thing: “5.26%” is the union settlement — read whether a wage headline is the settlement or the earnings survey before reading it as “pay.”
Japan's 2026 pay rise is not one number. It is a negotiated union settlement — 5.26% in Rengō's first tally (as-of 23 Mar 2026), of which 3.85% (¥13,013) lifts the structural floor and roughly 1.4 points is the seniority escalator — and a figure from the MHLW all-establishment survey lens, +3.4% (February 2026 final). These are two different defined quantities measuring one wage-growth idea, from different reference windows. The coverage-and-method gap is about 1.9 points (RAOSCAFF arithmetic: 5.26 − 3.4 = 1.86pp). Both are correct under their own definitions; both are positive. Mirror Brief JP-01 makes one claim: read whether a wage headline is the union settlement or the all-establishment earnings survey before reading it as “pay.” Every figure traces to a primary-grade source listed in the companion FACTS.md; the ~1.4pp split and the ~1.9pp gap are RAOSCAFF arithmetic. The brief alleges nothing against either body — each measures a different, correct population — and it takes no view on policy, the Bank of Japan, JPY or rates, or whether any wage level is adequate.
Research approach. Mirror format — RAOSCAFF anchors on the publishers' own figures and decomposes which pay rise each describes. JP-01 places two numbers on one sheet: Rengō's 2026 shuntō settlement tally (the negotiated, union, large-firm-weighted raise; 1st tally as-of 23 Mar 2026) and the MHLW Monthly Labour Survey (realised average cash earnings across all establishments of 5+ employees; February 2026 final). These are two different defined quantities measuring one wage-growth idea, from different reference windows. The hero is a two-ruler panel — the 5.26% settlement split into its 3.85% base-pay (ベア) floor lift and a ~1.4pp seniority escalator (定昇), beside the +3.3% provisional MHLW survey read (final: +3.4%), with a dotted bracket marking the ~1.96pp provisional-basis gap. No primary data collection, no analyst estimate; the only derived figures — the ~1.4pp escalator (RAOSCAFF arithmetic: 5.26 − 3.85 = 1.41pp), the ~1.9pp final-basis gap (RAOSCAFF arithmetic: 5.26 − 3.4 = 1.86pp), the ~1.96pp provisional-basis gap (RAOSCAFF arithmetic: 5.26 − 3.3 = 1.96pp), and the ~54.7% relative comparison (RAOSCAFF arithmetic: 5.26 ÷ 3.4 − 1 = 0.547) — are labelled RAOSCAFF arithmetic.
Source standards & limitations. Every figure carries its tally or month and its provisional-vs-final status, and traces to a primary-grade source in FACTS.md § H: Rengō settlement figures verified via JILPT (the MHLW-affiliated research institute), with nippon.com as corroborating source; MHLW earnings figures verified via JILPT (one step removed, as the raw MHLW release returned garbled encoding — to be confirmed against the e-Stat numeric table for final publication); the 3rd-tally figures via Jiji Press as corroborating source. The brief is anchored on the 1st-tally 5.26% (the number that dominated coverage) with the 3rd-tally figures (5.09% overall / 5.00% SME as-of 1 Apr 2026) noted; the final Rengō tally is due in early July 2026 and is not yet published as of 2026-06-04. Feb 2026 real wages (+2.0%) use CPI-less-imputed-rent; the all-items deflator differs marginally — both are official MHLW reads, and the brief flags the deflator dependence rather than picking one. JP-01 translates the gap into no JPY, JGB, or rates view, attributes it to no policy or central-bank action, and takes no view on whether wages are adequate. The settlement rate and surveyed earnings growth are stated to be two different defined quantities measuring one wage-growth idea, from different reference windows, every time they are compared; neither Rengō nor MHLW is criticised — each measures a different, correct population.
Sources 01–03 and 05 are JILPT publications (the MHLW-affiliated research institute), each a primary statistical source, fetched live 2026-06-04. Source 04 (nippon.com) and Source 06 (Jiji Press) are corroborating sources, not primary statistical publishers; the figures they report are cross-checked against the JILPT primary sources. MHLW earnings figures via JILPT are one step removed from the raw MHLW release — to be confirmed against the e-Stat numeric table for final publication. Full citations and the verified figures behind every number in this brief are listed in the companion FACTS.md, § H.