How did a sentence about spring wage negotiations become the most-watched monetary policy signal in the G10? Governor Kazuo Ueda's framing — that the shuntō outcome is "key to judging the inflation trend" — has been read across the street as a dovish hedge, a hawkish setup, a non-statement, and a coded commitment, often in the same week. The consensus reading is wrong, and it is wrong in a specific way. The desks writing those notes are reading the sentence as a rate-path signal. The sentence is a diagnostic one. The distinction matters, and the record from 2013 forward explains why.
March 2013: The Kuroda Regime and the Wage Problem It Could Not Solve
Haruhiko Kuroda took the chair on March 20, 2013 and within three weeks announced quantitative and qualitative easing on a scale no advanced-economy central bank had attempted. The target was explicit. Two percent inflation, two years, doubling the monetary base. The number was borrowed from the Fed and the ECB and the Bank of England. The mechanism was not. QQE pushed base money through the banking system at a velocity that only made sense if you believed Japanese prices moved the way American prices moved. They did not.
The problem Kuroda inherited was not that money was scarce. It was that firms were not raising wages. Japan's wage-setting institution — the shuntō, literally the "spring offensive" — had, since the late 1990s, settled year after year inside a one-percent band regardless of what the BOJ did at the short end. QQE could move the yen. It could move Nikkei futures. It could not reach into the Rengo bargaining tables where ten million unionised workers set the base compensation that then diffused outward through the SME labour market.
By 2016 the policy was already running against its own cost curve. Negative rates arrived in January that year — a decision the published minutes record as narrowly contested — and yield curve control followed in September. These were not escalations. They were workarounds. The two-percent target had not been hit, and the shuntō was still printing between 2.0 and 2.3 percent gross wage growth, of which base pay — the figure that actually matters for trend inflation — was closer to 0.4 percent. The gap between the BOJ's intended transmission and the labour market's actual response is the fact that frames everything Ueda would later inherit.
September 2022: The Intervention That Exposed the Execution Layer
On September 22, 2022, the Ministry of Finance instructed the BOJ to sell dollars for yen for the first time since 1998. The pair had moved from the 115 handle in March to the 145 handle in September, a 26 percent decline driven almost entirely by the US–Japan rate differential as the Fed hiked and the BOJ did not. The intervention was large, the move was large, and the recovery lasted less than a week. By late October the MOF intervened again and by November the pair had made new lows.
The street covered the policy angle exhaustively. The execution angle received almost no sustained reporting. What happened inside the plumbing during those sessions matters more than the price chart, because the plumbing is what every retail and institutional account is actually plugged into. The MOF intervention windows coincided with liquidity gaps in which top-of-book depth on EBS and Reuters Matching compressed to a fraction of its usual quote size. STP broker flows that had been clearing against prime-broker aggregated feeds found their last-look rejection rates climbing. The published post-mortems from the venue operators are terse. The desk-level account, reconstructable from the market structure papers the BIS CGFS has issued on FX liquidity since 2019, is less so.
Operators who lived through the 2015 Swiss franc episode will recognise the pattern. FXCM's negative-balance crisis that January was not caused by the SNB decision in isolation. It was caused by a reconciliation gap between the price the retail client was shown, the price the broker's liquidity feed cleared at, and the price the prime broker ultimately booked. That gap existed in September 2022 as well. The yen move did not break anyone the way the franc move broke FXCM, but it tested the same plumbing. The plumbing held — mostly. The fact that it held should not be mistaken for the fact that it was unchanged.
April 2023: Ueda Inherits a Framework Built Around a Number That Does Not Exist Yet
Kazuo Ueda took the chair on April 9, 2023. He was the first academic economist to lead the BOJ in the post-war period. His confirmation hearings in the Diet, which are in the public record, included an exchange in which he was asked how he would judge whether inflation had become sustainable. He answered — and this is the sentence the street has been misreading ever since — that the spring wage negotiations would be central to that judgement.
A rate-path reading of that sentence treats shuntō as a trigger. If the shuntō prints above X percent, Ueda raises. If below X, he holds. This is how most sell-side notes modelled the 2024 and 2025 decisions, and it is why the yen reaction function to the preliminary Rengo tallies became so mechanical. The problem is that this is not what Ueda said. He said the shuntō outcome would be key to judging the inflation trend — a diagnostic function, not a decision function. The sentence treats wage data as evidence about whether the price process is self-sustaining, not as a threshold that determines policy.
Two primary documents from the BOJ's own publication record illustrate the distinction. The Outlook for Economic Activity and Prices reports — the quarterly document in which the Policy Board publishes its forecast — have, since 2023, included dedicated sections on the wage–price feedback loop that reference the shuntō by name. The published statements from the Monetary Policy Meetings, by contrast, continue to describe policy decisions in terms of the two-percent target and the sustainability condition. One document treats shuntō as an input to the model. The other treats the model's output as the input to policy. Both are operative, and the distance between them is where the consensus reading goes wrong.
March 2024: The First Shuntō That Mattered and the Policy Pivot It Triggered
March 15, 2024 was the day Rengo announced the preliminary first-round tally of the 2024 shuntō. The number — a weighted average that eventually settled above five percent — was the largest headline wage gain in Japan in more than three decades. The BOJ moved four days later, on March 19, ending the negative rate policy that had been in place since January 2016 and exiting yield curve control. The decision was decided by a vote the published minutes record as 7–2, with the dissents focused on the pace rather than the direction.
The street declared this the vindication of the shuntō-as-trigger thesis. The number came in. The hike came out. The two were causally linked. Case closed. This reading is seductive and it is wrong. The specific five-percent figure did not determine the decision. The decision was determined by the Policy Board's judgement, formed over the preceding six months of regional branch surveys and the BOJ Tankan readings on labour market tightness, that the wage round would print strongly enough across the SME tier — not just the large-cap exporters — to produce a diffuse rather than concentrated wage push. The headline Rengo number is a leading indicator of the smaller-firm settlements that follow. It is not the datum. The datum is the diffusion.
This is the detail every rate-path model missed. The 2024 pivot was not conditional on the Rengo headline. It was conditional on the Policy Board's view that the Rengo headline would propagate. If the 2025 shuntō had printed above five percent in the first tally but been followed by SME settlements below two, the policy response would not have been another hike. It would have been a hold followed by a downgrade in the Outlook report. The sell-side notes modelling 2025 as a mechanical follow-through missed this because they were still reading the sentence from the 2023 confirmation hearing as a trigger, not a diagnostic.
October 2026: Where the Consensus Reading of Ueda's Latest Shuntō Line Goes Wrong
By October 2026 the pattern has repeated often enough that it should be visible. Each time Ueda uses the shuntō framing in a press conference or Diet appearance, the sell-side circulates notes within hours predicting the next move. The predictions diverge, sometimes by 25 basis points on the policy rate and by three big figures on USD/JPY. The reason the predictions diverge is that they are all trying to extract a decision rule from a sentence that was never a decision rule.
Let me concede something the critics of this view will raise. The shuntō outcome does correlate with BOJ decisions. The 2024 pivot followed a strong round. A hypothetical weak round in 2027 would, all else equal, correlate with a hold. The correlation is real. The error is treating the correlation as the mechanism. The mechanism is the Policy Board's assessment of whether the wage process has become self-sustaining in the SME tier, where more than seventy percent of Japanese employment sits. The shuntō headline is a signal about that assessment. It is not the assessment.
Everyone — the sell-side desks, the macro funds reading those desks' notes, the retail aggregators parroting the funds — recommends the wrong framework for Ueda's language. The right one is to treat each shuntō comment as a diagnostic update, read the Outlook report two weeks later to see which direction the Policy Board's wage-propagation estimate moved, and treat that move as the actual policy signal. The press conference sentence is the preview. The forecast revision is the film. Trading the preview as if it were the film is why the yen reaction function has become so noisy around these dates, and why the execution layer — the same layer that compressed in September 2022 — now sees liquidity holes around every shuntō calendar event. The plumbing problem is downstream of the misreading. Fix the reading, and the hole gets smaller.
What It All Means
The sentence Ueda keeps repeating is a diagnostic one, not a rate-path one, and the thirteen-year arc from Kuroda's 2013 QQE announcement through the 2022 intervention and the 2024 pivot is the evidence. The BOJ's problem has never been that it could not move the short rate. It has been that the Japanese price process is pinned by a wage-setting institution the central bank does not control and can only observe. The shuntō is the observation point. Ueda's framing acknowledges this explicitly in a way Kuroda's framing never did. That acknowledgement is the policy innovation, and it is almost universally missed.
For the execution layer — the venues, the aggregators, the prime brokers, and the retail-facing operators plugged into them — the implication is more concrete. The dates on which Policy Board members comment on wage data have become volatility events in their own right. The 2022 intervention episode showed what happens to top-of-book depth when a surprise lands into a one-sided book. The 2024 pivot showed what happens when the surprise is anticipated by consensus but mispriced in the tails. Both are symptoms of a market reading the signal through the wrong frame. The historical record of broker events — Refco in 2005, MF Global in 2011, FXCM in 2015 — is unanimous on one point. Reconciliation failures do not happen because the price moved. They happen because the price moved in a way the frame did not permit. Ueda's shuntō sentence does not permit the frame most desks are using. Something eventually gives.
This piece does not cover the yen's path from here, which is a function of the Fed as much as the BOJ and is not what the shuntō comment is about. It does not cover the specific SME diffusion metrics the BOJ publishes in its regional Tankan, which are a separate argument about which signals are leading and which are lagging. And it does not address the question of what the Ministry of Finance will do if USD/JPY tests the 2022 intervention levels again, which is a political question rather than a monetary one. Each of those is a separate piece.
FAQ
What exactly did Ueda say about the shuntō, and when?
The framing has appeared in multiple settings since his April 2023 confirmation hearings — in Diet testimony, in post–Policy Board press conferences, and in the BOJ's quarterly Outlook reports. The common thread is that spring wage negotiations are central to judging whether inflation is becoming sustainable. The phrasing varies but the content is consistent: shuntō outcomes are a diagnostic input to the Policy Board's view of the wage–price loop, not a mechanical trigger for rate decisions.
Why does the consensus misread the sentence as a rate-path signal?
Because the sell-side is incentivised to publish decision rules, not diagnostic frameworks. A note that says "if shuntō prints above X, BOJ hikes" is tradable. A note that says "shuntō updates the Policy Board's estimate of wage diffusion in the SME tier, which the Outlook report will then revise" is accurate but harder to sell. The error is structural. It persists because it serves the desks writing the notes more than it serves the clients reading them.
How did the September 2022 intervention actually move through broker plumbing?
The Ministry of Finance operation compressed top-of-book depth on the primary interbank venues and raised last-look rejection rates on aggregated feeds that STP brokers depend on. The gap between retail-visible prices, broker liquidity-feed clears, and prime-broker bookings widened measurably during the active intervention windows. No major broker failed, but the plumbing was tested in a way the published post-mortems from venue operators have not fully addressed. The BIS CGFS papers on FX liquidity since 2019 describe the structural conditions that make these gaps possible.
Does the 2015 FXCM event tell us anything about what could happen in a future yen episode?
It tells us the specific mechanism to watch, which is reconciliation. FXCM's negative-balance crisis in January 2015 was not caused by the Swiss franc move per se. It was caused by a reconciliation gap between the price shown to the retail client, the price the broker's feed cleared at, and the price the prime broker ultimately booked. That same gap category exists on yen venues today. A sufficiently large and sufficiently sudden move — intervention, policy surprise, or both — tests it. The 2022 episode was close enough to that scenario to be instructive.
Why does base pay matter more than the headline Rengo number?
Headline shuntō figures include seniority increments and lump-sum bonuses that do not persist year to year. Base pay — the component Japanese economists call "base-up" — is the compensation floor that actually transmits into trend inflation. Historically the headline has run two to three times base pay, which is why a strong-looking total can mask a weak core. The BOJ Policy Board watches base-up and the SME diffusion of base-up, not the headline. The street reads the headline because it is published first.
Does any of this change the near-term direction of USD/JPY?
Direction is a function of the US–Japan rate differential and the risk environment far more than of any single shuntō comment. What the shuntō frame does change is the variance around the direction. Misreading the diagnostic as a trigger produces mechanical positioning into each wage-data event, which amplifies gaps when the actual Policy Board response diverges from the mechanical model. The pair is not more directional because of this. It is more fragile.
Which primary BOJ documents are worth reading alongside the press conference lines?
The quarterly Outlook for Economic Activity and Prices contains the Policy Board's actual forecast revisions and, since 2023, a dedicated discussion of the wage–price feedback loop. The Monetary Policy Meeting minutes publish with a lag but reveal how individual members weighted the wage data. The regional branch Tankan surveys publish separately and are the clearest window into SME-tier wage settlement expectations. Reading the Outlook and the regional Tankan together is more informative than any single press conference transcript.