The minutes of the Bank of Japan’s 30 and 31 July meeting were released at 23:50 UTC last night, and they contain one sentence that is worth more than the rest of the document put together. The Board recorded that “the market appeared to expect that the policy interest rate would likely be raised with intervals of about six months,” and then said the pace “could be faster than market expectations, depending on developments.” Forty-nine days later the Bank raised rates. Measured from the previous increase, the interval was ninety-four days — slightly over half of what the minutes said the market was assuming.
What the minutes actually say
Three things, all quoted from the Bank’s own document rather than a wire summary of it.
On the decision: the Board voted eight to one to keep the uncollateralised overnight call rate at around 1.0 percent. The dissenter was Takata Hajime, who proposed 1.25 percent on the grounds that “the situation had shifted to a new phase in which the Bank needed to adopt a nimble approach in response to upside risks to prices caused by demand shocks stemming from overseas developments.”
On pacing: the six-month sentence above, followed immediately by the warning that the actual pace could exceed it. Two secondary readers of the same document render this as members judging that market pricing of six-monthly hikes “could prove too slow.” We quote the Bank’s own wording because the difference between a Board noting what the market thinks and a Board saying the market is wrong is exactly the distinction a rate-path reader is paid to make, and the minutes are closer to the first than the headlines suggest.
On the yen: members discussed depreciation as raising “the profits of global firms” while pushing down “households’ real income” and putting “downward pressure on the profits of small and medium-sized firms.” That is the same three-part framing the Bank has used for two years. It is the part of the document that was repeated, not changed, and it is why we would not build a yen view on these minutes.
In seven weeks the dissent changed sides
This is the finding, and it is visible only if you read three statements rather than one.
On 16 June the Board raised the rate to around 1.0 percent by seven votes to one. The dissenter was Asada Toichiro, who wanted the previous guideline kept because in his view downside risks to production and employment outweighed upside risks to prices.
On 31 July the Board held at 1.0 percent by eight votes to one. The dissenter was Takata, in the opposite direction, asking for 1.25 percent.
On 18 September the Board raised the rate to around 1.25 percent — the exact number Takata had proposed — by seven votes to two, effective 24 September. The two dissenters were Asada again and Sato Ayano, both arguing the move was premature. Takata, having got precisely what he asked for in July, is not among them.
So in the space of two meetings the committee went from one hawk dissenting alone to two doves dissenting together, and the policy landed on the hawk’s number. A dissent count tells you how isolated a view is. A dissent count read across three meetings tells you which way the centre of the committee is travelling, and this one travelled a long way in seven weeks.
Ninety-four days is not six months, and that is the tradeable part
Take the arithmetic seriously. The previous increase was 16 June. The next was 18 September. That is ninety-four days, or 3.09 months. The interval the July minutes attribute to the market was about six months, call it 182 days. The realised interval was 51.6 percent of the expected one.
If you carried a six-month pacing assumption into September you were not merely wrong about a date. You were wrong by a factor of two on the variable that sets the whole front end of the Japanese curve, and you were wrong after the Bank had told you in a published document that you might be. The minutes were embargoed until this morning, so nobody could have read them in September. But the same argument was available in the Summary of Opinions from that meeting, and the market’s own pricing says it was not taken.
The live question is what the pacing assumption should be now. If you apply the realised ninety-four days to 18 September you get 21 December, which brackets the 17 December meeting. If you apply the six months the market was assuming you get mid-March. That is a three-month spread on the next move, and it is the single largest unpriced item in yen rate expectations. This desk is not going to name a date. We are going to say that anyone still running six months as a base case is running an assumption the Board itself flagged as possibly too slow, before it then halved it.
What was changed and what was merely repeated
Changed, on two readers’ accounts of the same document: the framing of the inflation objective moved from lifting underlying CPI inflation to 2 percent towards anchoring it at around 2 percent. That is a different job. Lifting is a job you can be patient about; anchoring is a job where waiting has a cost. One reader reports a member saying the assumption that the risk of waiting was marginal could no longer be relied on.
Repeated: the yen framing above, and the balance-of-risks language. Secondary readers put underlying inflation excluding fresh food near 1.5 percent and note measures excluding subsidies running between 2.5 and 3 percent. We did not verify those figures in the primary document and they are one and two readers respectively; treat them as colour, not as the argument.
The distinction matters for a mechanical reason. A central bank that repeats a phrase is telling you its reaction function is unchanged; a central bank that rewrites one is telling you the function has moved and the old conditional thresholds no longer map. The anchoring rewrite is a reaction-function change. The yen paragraph is not.
What this does not tell you
Almost everything that matters right now. These minutes describe a meeting held two months ago that was overtaken by events seven weeks later; the Bank has already done the thing the minutes were debating. Nothing in them is news about the current stance, and anyone trading this morning’s document as a signal has the sign of the information backwards.
The document that will matter is the Summary of Opinions from the 17 and 18 September meeting, released at 23:50 UTC on Wednesday alongside the September Tankan. That is where the nine people who actually voted on 1.25 percent say why. We would also note, since this desk was wrong about it once already, that the 1 October release is the Summary of Opinions and not the minutes of the September meeting — a different document on a different cadence, and at least one widely read Japanese calendar has it mislabelled.
The vote counts, the dissenters’ names and the quoted sentences above are read from the Bank’s own statements and minutes. The colour on inflation levels, the anchoring rewrite and the “too slow” rendering are secondary readings by two publishers, one of which is the standing labelling risk on this channel. The ninety-four-day and 51.6 percent figures are our arithmetic on primary dates, shown above so you can check them.
And the obvious limitation: an interval is two observations. Ninety-four days is one gap, not a policy. A central bank that moved faster once has not thereby adopted a quarterly cadence, and we would treat 17 December as a live meeting rather than a scheduled hike.
