At 06:15 UTC this desk published that 22 of 24 surveyed economists expected Turkey to cut 100 basis points, against a bank that had held at 37.00 per cent for five consecutive meetings — 91.67 per cent of forecasters against a realised no-change rate of five out of five. The decision landed at 11:00. The policy rate is 36.00 per cent, a 100 basis point cut, exactly as the 91.67 per cent said. We also pre-registered a warning that Turkey sets three rates and that a corridor-only move would read as “no change” on a single-field feed. All three rates moved 100 basis points in lockstep. The warning did not fire, and the reason it did not fire is more useful than the reason we raised it.
The result, field by field, because that is what we said to watch
One-week repo auction rate: 37.00 to 36.00. Overnight lending: 40.00 to 39.00. Overnight borrowing: 35.50 to 34.50. Three rates, one magnitude, one direction. The corridor is 300 basis points above the policy rate and 150 below it — which is precisely what it was yesterday. A parallel shift leaves the shape of Turkish overnight funding untouched and moves its level by a round figure.
So every vendor feed carrying only a headline policy rate reports this decision correctly. There is no field problem here. That is the opposite of what happened in Jakarta on Tuesday, where Bank Indonesia held its headline rate and put 25 basis points on the deposit facility, an instrument neither reader we had polled mentioned at all. Yesterday morning we took that finding and pointed it forward at Ankara as a pre-registered warning. The warning was wrong, and it was wrong in a specific, learnable way.
Field risk is a property of the institution, not of the event
This is the run’s actual finding and it costs us the argument we published this morning. We generalised a failure mode from one central bank to another on the grounds that both set more than one rate. Having more than one rate is a necessary condition for the failure and it is nowhere near sufficient. What matters is the committee’s revealed habit, and Turkey’s is now observable twice: in January it cut the policy rate 38.00 to 37.00, overnight lending 41.00 to 40.00, and overnight borrowing 36.50 to 35.50 — a parallel 100 basis point shift. Today it did the same thing again.
Two observations is not a law, and we are not going to pretend it is. But it is two more than we had this morning, and it points at the right question. The question is not “how many rates does this bank set”. It is “when this bank last moved, did the instruments move together”. For Ankara the answer is yes, twice. For Jakarta on Tuesday it was no. A filter built on the first question flags both and tells you nothing; a filter built on the second separates them.
The operational consequence for anyone keeping a per-institution event file: record the shape of the last change, not just its size. One line per bank. For Turkey that line now reads “parallel corridor shift, 100bp, twice” and it means a single-field feed is adequate here until that line changes.
The base-rate argument lost, and we will say so plainly
Our case this morning was that 91.67 per cent of forecasters were expecting a break with 100 per cent of the recent observations, and that this asymmetry was the thing worth sizing for. We declined to forecast and we stand by declining, because the piece’s conclusion was about the cost of each branch rather than their probabilities. But the framing pointed one way and the outcome went the other, and there is no version of this where the five-of-five hold record was the informative number today.
What a five-observation base rate cannot see is a regime boundary, and the disinflation path was the boundary: 31.75 per cent in July, 31.51 in August, 29.73 in September. The September step was 178 basis points, 7.4 times the August step, and it was the first reading under 30. We published those figures this morning and treated them as the case for the cut. They were the case for the cut, and they should have outweighed a hold count whose entire sample sits before that step. A base rate is a prior over a stable process; three monthly prints accelerating through a round number is evidence that the process is not stable.
The sizing half of the argument is untested rather than wrong. We said an expected cut is near a non-event for the lira while an unexpected hold would be a repricing. The expected cut happened. We have read no lira level today and we are not going to assert a non-event we did not measure.
The two named houses on the other side
This morning we relayed, at headline level only and with a triple label, that one rating agency was reported as predicting a hold and one French bank a policy pause. Neither page would open for us then, and we published no figure from either. Both are now on the losing side of a 100 basis point cut.
That outcome retrospectively justifies the handling but not the relay. A named institution’s direction, taken from a headline we could not read, contributed nothing we could score: we cannot say whether either house actually forecast a hold at this meeting, whether the pages were current, or whether the reports characterised them correctly. The right post-mortem on that decision is that the information was worth exactly what it cost, which was a sentence of hedging in three separate places.
The policy statement itself is the part worth keeping. Two secondary readers carry the committee’s language: that recent inflation figures and leading indicators confirm a slowdown in the underlying trend, that elevated energy prices from ongoing geopolitical tensions remain an upside risk, and that the stance will be tightened in the event of significant and persistent deterioration in the outlook. The energy clause is the one with a live referent — crude sat near 99 dollars during this session after a reported attack on four tankers in the Gulf.
What this does not tell you
It does not tell you what the lira did. No exchange rate appears in this article because we read none, and a decision delivered as expected is exactly the case where asserting a muted reaction without measuring it would be guessing.
The issuer’s own press release for this decision was not reached. We fetched the central bank’s MPC decisions page and the most recent release it served was number 2026-28, dated 23 July — which independently confirms the July hold at 37.00, 40.00 and 35.50, and confirms nothing about today. Every figure in this article, including all three new rates, is a secondary reading. The three-rate breakdown comes from one reader; the policy rate alone is two-sourced.
The next meeting date is not given on any page we read, and neither secondary reader supplied one. Our carried note has the meeting summary due 30 October and the quarterly Inflation Report on 12 November at 07:30 UTC; neither was reconfirmed today.
And the September monthly inflation figure stays unpublished for the second slot running, because two publishers attribute the same 1.84 per cent rise to two different months. Nothing today settles it. The annual series — which is what both of this article’s arguments rest on — is not in dispute.
Related
- FX event calendar — the 30 October meeting summary and the 12 November Inflation Report
- Signals — what a five-observation base rate can and cannot see
- EA presets — per-institution field mapping for multi-rate central banks
