Germany reopened two Bunds this morning, a 2047 and a 2056, one billion euros each. The thirty-year attracted 1,241 million euros of bids against 3,049 million seven days ago — a 59 percent collapse — and cleared ten basis points cheaper for the buyer anyway. The published bid-to-cover fell from 2.4 to 1.6, which sounds like it captured the deterioration. It did not capture all of it, because the Finanzagentur divides by the volume allotted, and the volume allotted is something the auction decides after the bids are in. The denominator moves with the answer.
The identity nobody prints next to the ratio
The Bundesbank retains part of every German auction for secondary-market operations. The issuer publishes that explicitly, in a column headed “Retention Quote”, and defines its bid-to-cover as the relation between total bidding volume and total allotted volume — not the volume offered.
That has an algebraic consequence which is worth writing out, because it is exact rather than approximate. If you compute the ratio against the offering instead, the two figures differ by a factor of offered-over-allotted. Work the difference through as a share of the published number and it collapses to the retention rate itself.
Concretely: today’s thirty-year retained 216.40 million euros of a 1,000 million offering — 21.6 percent. Its published cover ratio is 1.6. Bids per euro actually offered were 1.24. The gap between those, expressed as a share of the published figure, is 21.6 percent. The same number, necessarily.
So the portion of a German cover ratio that is contributed purely by retention is not an estimate. It is the retention rate, and the issuer prints it in the next column along. A weaker auction retains more, which shrinks the denominator, which raises the ratio. The measure is flattered by exactly the weakness it is supposed to report.
Today against last Wednesday, with the denominator held still
The thirty-year, ISIN DE000BU2D012, 2.90 percent coupon, maturing August 2056:
- 16 September — 1,500 million offered, 3,049 million bid, 1,257.41 million allotted, 242.59 million retained (16.2 percent), average yield 3.90 percent, published cover 2.4.
- 23 September — 1,000 million offered, 1,241 million bid, 783.60 million allotted, 216.40 million retained (21.6 percent), average yield 3.80 percent, published cover 1.6.
The published ratio fell 33 percent. Bids per euro offered fell from 2.03 to 1.24 — 39 percent. The published series understated the deterioration by six percentage points, and it did so because retention rose from 16.2 to 21.6 percent between the two auctions.
The twenty-year, DE000BU2T000, tells the opposite half of the story and is the control. Bids fell 15 percent, from 2,110 to 1,794 million. Allotment was essentially unchanged, 863.82 to 869.74 million. Retention barely moved, 13.6 to 13.0 percent. Published cover 2.4 to 2.1, bids per euro offered 2.11 to 1.79 — both down about 15 percent, in agreement, because the denominator sat still.
That is the useful shape. The two measures agree whenever retention is stable and diverge whenever it is not, which means the published ratio is least trustworthy in precisely the auctions you most want to read.
Demand halved and the price went up
Both reopenings cleared ten basis points below where the same two bonds cleared a week earlier — 3.78 percent against 3.88 at twenty years, 3.80 against 3.90 at thirty. The thirty-year did that with 59 percent fewer bids.
We are not going to dress this up as a demand story, because it is not one. Bid volume at a German auction measures how many primary dealers showed up to compete for a specific reopening on a specific morning; the clearing yield is set by the whole European curve, which fell over the week. Those are two different flows, and this morning they pointed in opposite directions.
The operational point for anyone running exposure through a European supply window: bid volume is a market-structure signal about that auction, not a rates signal. If your event filter widens a window around a Bund auction because you expect a yield reaction to demand metrics, today gives you a clean counterexample — the worst bid volume in the visible series and a ten-basis-point rally in the same instrument.
This closes a gap our own Macro desk declared this morning
At 06:14 UTC our Macro desk published the US par curve yielding more at twenty years than at thirty — 5.33 against 5.29 on 22 September, four basis points the wrong way up, on all three business days since the Bank of Japan moved. In its humility section it said plainly that it had found no German twenty-year level and would not interpolate one.
The auction has now supplied it, from the issuer. Germany at twenty years cleared 3.78 percent this morning against 3.80 at thirty. Two basis points, the right way up. The same two basis points, the right way up, a week ago at 3.88 and 3.90.
So the two sovereigns’ twenty-to-thirty segments currently have opposite signs, and both numbers come from issuing agencies rather than aggregators. That is a more interesting fact than an interpolated German twenty-year would have been, and it is only available because Macro declined to invent one. We are crediting the article that left the hole rather than the desk that filled it.
One caveat we are not going to paper over: an auction clearing level and a par curve point are not the same object, which is the argument Macro itself made this morning about the US two-year. Comparing 3.78 to 3.80 is clean because both are clearing levels from the same auction session. Comparing either to a US par yield is a looser comparison and we are presenting it as a direction, not a spread.
Spot is between two strikes with two hours to the cut
EUR/USD is trading around 1.1412 to 1.1422 depending on which reader you take — 1.1412 at 08:04 UTC on one publisher, 1.14219 stamped today on another, both down between 0.23 and 0.3 percent. Today’s 10:00 New York cut, at 14:00 UTC, carries strikes at 1.1400 and 1.1450 with spot between them, roughly 12 to 22 pips above the lower one.
The euro fell on a broad beat. Services printed 53.0 against a 51.4 or 51.7 consensus and the composite hit 53.1, a 41-month high on one bank’s reading, and the currency went down. The conventional explanation is broad dollar strength, with the dollar index around 100.75 and up 0.2 percent on the day, and that is probably most of it.
But for the next two hours there is a simpler mechanical reading available: with spot sitting a dozen pips above a strike and the cut approaching, the 1.1400 expiry is a pull, not a level. The publisher that carries the board says as much in its own copy. We cannot size that view, because the notional amounts sit inside an image that does not extract — the fifth consecutive session on which that has been true — so we have strikes without weights and a direction without a magnitude.
What that is worth operationally: it is a reason to be smaller into 14:00 UTC, not a reason to be short. The US flash PMIs at 13:45 land fifteen minutes before the cut, which means the release and the expiry are inside the same window and you cannot attribute the move to either afterwards.
What this does not tell you
The cover-ratio identity is arithmetic, not an accusation. The Finanzagentur documents its own definition and prints the retention figure in the adjacent column; nothing is hidden, and anyone reading the primary table has everything needed. The failure mode is in the secondary reporting that quotes the ratio alone, ours included — this desk published a German cover ratio without its denominator nine times before this morning.
We have no when-issued levels for either reopening, so nothing here is a tail and we are not calling it one. We also have no German secondary-market twenty-year quote to check the 3.78 against, only the clearing level.
The bid collapse is measured against one prior auction, not a long series. The visible results table runs back to early September; two observations of a thirty-year are not a trend, and the 16 September auction was a larger offering, which mechanically attracts more bids. We have used bids-per-euro-offered precisely so that size difference does not do the work, but it does not eliminate it.
And the expiry section is a mechanism, not a forecast. Spot has been below its 100-hour moving average at 1.1468 all session and is in a broader downtrend; the strike may simply be somewhere price was going anyway.
