The Bank of Japan announces a decision in about three and a half hours that will take its policy rate to 1.25%, a level the country has not seen since 1995. Two readers describe that as a 31-year high and we have no reason to doubt them. What we cannot tell you is how priced it is, because the three answers in front of us are 80%, roughly 97% and 100% — and they are not three measurements of one thing. They are three different questions. Only one of them has a price attached.
Eighty per cent, ninety-seven, and a hundred are not a disagreement
Here is what this desk read this morning. OANDA Japan’s preview of today’s inflation release says the overnight index swap market prices “roughly 80%” for a hike at today’s meeting. Gaitame.com’s 07:06 JST morning note carries the same figure independently. FXStreet, writing at 20:57 UTC last night, says a quarter-point increase to 1.25% is “priced at 100%”. And the Reuters poll this desk has been carrying since 10 September, conducted 1–8 September, had 66 of 68 economists expecting exactly this move, which is 97% of the sample.
On 15 September this desk published a rule after finding three feeds giving three probabilities for a Fed hike — 85%, above 90%, and 93% — and said to read at least two and name the feed whenever the probability is load-bearing. Today the spread is twenty points wide, which is far worse, and the rule does not go far enough. The three numbers are not competing estimates. An OIS price is what somebody will transact at. An economist survey is a count of opinions collected ten days ago by people who are not obliged to trade them. And “100%” is a sentence, not a quote from a market.
If you are sizing a position, the only one of the three that constrains you is the swap. Eighty per cent priced means one hike in five, on that market’s own pricing, does not happen — and a no-hike at 80% priced is a far larger yen move than a hike at 100% priced. The 97% and the 100% are useful for understanding the consensus narrative and useless for sizing. If you have been reading “fully priced” all week and concluding the decision itself carries no risk, the swap market disagrees with you by twenty points.
The Bank’s own neutral rate is a band 140 basis points wide
Gaitame Research Institute’s analyst set out the arithmetic in last night’s outlook, and this is one reader, so treat it as one desk’s reading of the Bank’s published work rather than as the Bank’s own statement. The Bank of Japan’s estimates of the natural rate of interest span −0.9% to +0.5%. Add the 2% inflation target and the nominal neutral policy rate falls somewhere between 1.1% and 2.5%.
That band is 140 basis points wide. After today’s move the policy rate is 1.25%, which is fifteen basis points above the bottom of it and 125 basis points below the top. So the question “is the Bank of Japan still accommodative after this hike” has the answer “the Bank cannot tell you, and neither can we”. This is the same discipline we applied to a four-dollar range in WTI and to a five-pip disagreement about a moving average: when the sources cannot agree on a number, publish the band and let it be the error bar. The difference is that this band belongs to the institution setting the rate.
It matters for one reason. Every hawkish interpretation of today’s guidance rests on some view about how far below neutral the Bank still is. If the honest answer is “between fifteen basis points and a hundred and twenty-five”, then the guidance is not going to resolve it either, and a position sized on the assumption that it will is sized on a number nobody has.
Four hikes buy you thirty basis points of carry
This is the part that explains why the yen is trading at 156 into a hike to a 31-year high rather than rallying at it.
The Federal Reserve’s target midpoint after Wednesday’s increase is 3.875%. The Bank of Japan’s rate going into today is 1.00%. The gap is 287.5 basis points. FXStreet publishes twelve-month market-implied rates for both: 4.60% for the Fed and 2.03% for the Bank of Japan. That is one reader and we label it as one. Take those at face value and the gap in a year’s time is 257 basis points.
So: the market expects the Bank of Japan to raise four times, taking the policy rate to a level it has not held in three decades, while the Federal Reserve adds roughly three quarters of a point on top of an already-hiking cycle — and the carry differential compresses by about thirty basis points out of two hundred and eighty-seven. Roughly one basis point of compression for every ten you started with.
If you are short the yen on carry, that is the number that should govern your stop placement rather than the decision headline. A twelve-month view that is entirely correct about the Bank of Japan’s path buys you about a tenth of the position’s reason for existing. That is not an argument for being short. It is an argument that the decision at 03:00 is the wrong event to be sizing around, and that the guidance about 2027 is the right one.
The market prices a higher terminal rate than the economists do
This is the inverse of the usual complaint and we have not seen it stated anywhere this week, so we will state it.
The Reuters poll, carried by two readers, has economists at 1.5% by the end of March 2027 and 1.75% in the following quarter, with most putting the terminal rate at at least 1.75%. Gaitame’s analyst expects four hikes taking the rate to about 2% by the July 2027 meeting. The swap market, on FXStreet’s reading, has 2.03% in twelve months — which is to say by September 2027.
The survey and the market are not far apart on the destination, and where they differ, the market is the more hawkish of the two. The usual shape of a Bank of Japan story is a market that refuses to believe the normalisation; this one has the market pricing normalisation slightly faster than the profession forecasts it. Whatever you think of either number, the trade that assumes the market is underpricing Japanese rates is fighting the survey, not confirming it.
One dissenter, two months, and a name we will not print
Two readers put a dissent on the record and they do not place it in the same month. One says the Policy Board voted 8–1 in July to hold near 1%, with the dissenter proposing 1.25% — the exact level the Bank is expected to reach today. The other says a named board member dissented in June and may dissent again today.
Both may be true; a member can dissent twice. But we cannot second-source the name against either month, and this desk has a standing rule about that: when a proper noun will not corroborate cleanly, cut the name and keep the event. So the event is this. There is at least one member who has already voted for 1.25% before the committee got there, and a hike today with that member dissenting for something larger is a different signal from a unanimous hike, even though the headline rate is identical. If your news filter keys on the rate and not on the vote, those two outcomes look the same to it, and they are not.
What this does not tell you
We do not have a confirmed announcement time. One reader gives 02:45 UTC; the Bank does not publish a minute in advance and past decisions have landed around 03:00 UTC. Governor Ueda’s press conference at 06:30 UTC is on three readers and we are confident in that one.
The neutral-rate band and the twelve-month implied rates each rest on a single reader, and we have said so in the sections that use them. The October and December meeting odds we saw — 25% and 65% respectively — are also one reader, which is why they are in this paragraph rather than in an argument.
We have not read the Bank’s outlook report, because it does not exist yet. We have not read the Statement on Monetary Policy. Everything above is about the shape of the decision rather than its content, and the content is the part that will move the yen. If the guidance surprises, none of this arithmetic protects you, and we would rather say that plainly than dress up a pre-decision note as a view.
