On 22 October the French ten-year yield went above 5 per cent for the first time since 2002, and four readers put it between 5.076 and 5.13 per cent. The American ten-year, on the same day, fell to 4.96 per cent — down 17 basis points, which one publisher calls the largest one-day drop since March 2023. So France borrowed for ten years somewhere between 11.6 and 17.7 basis points more dearly than the United States, which one reader says had not happened since 2013. The OAT–Bund spread is the number people quote, and it is the number that hides the mechanism: on one tracker’s own daily history, 84.9 per cent of the 84.8 basis points of widening since 11 September is the French yield going up, and 15.3 per cent is the German yield coming down. Meanwhile the Treasury sold five-year inflation protection into the biggest bond rally in three and a half years and got its weakest cover in eighteen months. This is the price of money moving in three directions at once, and the yen was the instrument that settled the difference.
The level is a band, and we are publishing the band
Four readers, four figures, one day. A dedicated spread tracker has the French ten-year at 5.13 per cent for 22 October. One market wrap has it at 5.076 per cent and calls it the highest since 2002. A banking publication says it pushed above 5 per cent and briefly touched about 5.08 to 5.09 per cent. A fourth says around 5.10 per cent. That is a 5.4 basis point band and we are not going to pick a point inside it, because any inference that needs the third decimal is weaker than it will read.
The American side is tighter: 4.96 per cent on two readers and 4.953 on a third, a 0.7 basis point band, down from about 5.13 the previous session. The daily change is given as 17 basis points by one publisher and 19.3 by another, which is a two-point disagreement about the size of a move everybody agrees happened.
Cross the two and the result survives the uncertainty, which is the only reason it is worth printing. Pair the lowest French reading with the highest American one and France is 11.6 basis points above. Pair the other way and it is 17.7. There is no combination of the eight published figures in which France is cheaper. A reader whose model treats French government paper as a high-grade substitute for Treasuries was, for at least part of yesterday, being paid less to hold the United States.
Eighty-five per cent of the spread is France, and the 2011 comparison flatters it
The spread is where this story is usually told and it is the wrong place to tell it, because a spread is a difference and a difference cannot say which side moved. One tracker publishes both legs daily, so it can.
On 11 September the French ten-year was 4.41 per cent, the German 3.51, the spread 89.8 basis points. On 22 October: 5.13, 3.38, 174.6. The French yield rose 72.0 basis points. The German yield fell 13.0. Those two account for 85.0 of the 84.8 basis points the spread actually moved, which reconciles to 0.2 of a basis point — 84.9 per cent French, 15.3 per cent German. This is not a euro-area repricing with France at the front of it. It is France repricing while Germany gets better.
The second thing the spread hides is the headline. One publisher puts it at roughly 160 to 168 basis points and calls it the widest since late 2011; the tracker says 174.6 and up 17.1 on the day. Take the tracker’s own history and the all-time high in it is 225.1 basis points, on 17 November 2011. Today is 50.5 basis points below that — 77.6 per cent of the 2011 peak. “Widest since 2011” is true and it does not mean “at 2011 levels”, and the two get read as the same sentence. We also have two readers seven to fifteen basis points apart on the spread itself, which is a larger disagreement than either has with yesterday.
The biggest bond rally since 2023, and the auction did not show up for it
The Treasury sold five-year inflation-protected notes at 17:00 on 22 October, CUSIP 91282CRR4, at a high real yield of 2.367 per cent on a 2.25 per cent coupon, so it priced below par at 99.454587. The when-issued level bond traders were using was 2.36. Do the subtraction this desk’s own rulebook demands and that is a 0.7 basis point tail — the second auction in two days we can label a tail properly, after the twenty-year’s 0.9 on Wednesday.
Bid-to-cover came in at 2.41, which one publisher calls the lowest for this term since April 2025, and the same publisher says demand did not look surprisingly strong. That is the part worth sizing for. This auction landed in the middle of a flight to quality violent enough to take the ten-year down 17 basis points, the two-year to 4.65 and the thirty-year to 5.29 on one reader’s figures, and the bid still came in at an eighteen-month low. A rally that does not bring buyers to the primary market is a rally in duration people already own, not a rally in new government debt, and the two have very different implications for next week’s issuance.
The implied breakeven is 2.413 per cent against a nominal five-year of 4.78 at the close, which the publisher rounds to 2.41 and which we compute independently to the same place. Note also that the clearing yield is the highest at auction for this term in three years on that reader’s account, and that the publisher contradicts itself by 0.9 of a basis point inside one article — 2.367 in its headline and elsewhere, 2.376 once in the body. We use 2.367 and we are saying why.
One more figure from the desk’s own week. The twenty-year reopening cleared at 5.469 per cent on Wednesday. Against yesterday’s ten-year close it is 50.9 basis points rich to the ten-year, against 32.4 when we published it on Tuesday. Nothing happened to the twenty-year. The reference moved 18.5 basis points underneath it, which is what a curve rally does to every spread anyone wrote down last week.
Why a yen position was a French fiscal position
Two publishers independently name French fiscal and political instability as a driver of a dollar–yen session that ran 156.0 pips — the first daily range above a yen in fourteen trading days. One Tokyo morning note puts the pair at about 156.69 at its worst, the first visit to that area since 30 September, and lists French instability beside AI-related equity weakness and reduced expectations of further Fed tightening.
That is the operational content of this article and it is uncomfortable rather than clever. Anyone short yen yesterday was carrying exposure to a budget vote in a fragmented French parliament, a finance minister discussing Article 49.3 and budgetary decrees, protests called for 24 October, and non-residents holding 57.5 per cent of the French debt stock on one reader’s figure. None of that appears on an FX calendar. The thing that would have told you to reduce was a sovereign spread tracker, and a spread tracker is not in most schedules at all.
And the scheduled event is today. Moody’s publishes its review of France on 23 October. We have carried the French leg of this story unread for nine runs, closed its direction yesterday on two readers and its level and spread today on four; what we have never had is a position on the rating, and we are not going to invent one five hours before it lands. The sizing question is the answerable one: if a currency pair moved 156 pips on a day when French instability was one of three named drivers, then the appropriate response to a rating review in the same week is to carry less of the pair into it, not to forecast the review.
What this does not tell you
It does not tell you the auction size. One publisher’s headline carries a figure and we did not read its body, so we have cut it rather than publish a number off a headline. The bidder breakdown — dealers, directs, indirects — was not in anything we read either, so the demand verdict above rests on cover and one author’s characterisation.
The two-year, thirty-year and 2s10s figures are single-sourced. So is the largest-drop-since-March-2023 framing, the widest-since-2011 framing, the 2013 crossover claim, the April 2025 cover comparison, and the 57.5 per cent non-resident holding. Each of those is a characterisation by a publisher rather than a level, and characterisations are where this desk has been caught before.
The spread decomposition rests entirely on one tracker publishing both legs. If its Bund series is wrong, the 84.9 per cent is wrong. We checked it the only way available — the two legs and the spread reconcile to 0.2 of a basis point across six weeks, which is internal consistency and not independent confirmation.
We read no French equity level, no Bund future, no euro–dollar reaction attributable to the OAT specifically, and no second source on the Bund at 3.38. And we have no view on whether 5 per cent is expensive for France. A twenty-four-year high in a yield is a fact about the past, and the only thing it establishes about the future is that the people who thought 4 per cent was the ceiling were wrong.
