Mechanism

Auction tail

An auction tail is when a Treasury auction clears at a higher yield than the market was quoting seconds before bidding closed, meaning the government had to pay up to find enough buyers. It is measured in basis points, and it is the closest thing the bond market has to a live demand reading: a tail says the bidders who showed up wanted a discount, and the dealers who are obliged to bid ended up owning the rest. A run of tailing auctions is how a slow repricing announces itself as indigestion.

How it actually works

The US Treasury sells new debt at single-price auctions. Before each one, the not-yet-issued security already trades in a forward market called the when-issued market, so at the moment bidding closes there is a live consensus yield for the bond about to be sold. Bidders submit the yield they will accept and the size they want. The Treasury fills from the lowest yield upward until the whole issue is gone, and everyone pays the price implied by the highest accepted yield — the stop-out.

is the gap between the stop-out yield and the when-issued yield at the deadline. If the when-issued 10-year was quoted at 4.20% and the auction stopped at 4.22%, that is a two-basis-point tail (both figures illustrative). Buyers collectively refused to take the bond at the price the market had been advertising, and the Treasury paid two basis points more than the screen said it should. Clearing below the when-issued yield is the opposite outcome, called stopping through, and it means demand was stronger than the market expected.

Two basis points sounds like nothing. In price terms on a thirty-year bond it is not, and more to the point it measures something otherwise invisible: how much real money is actually willing to fund the government at today's yield, as opposed to how much is willing to quote.

The mechanical consequence sits with the primary dealers. They are required to bid in every auction, which makes them buyer. When an auction tails, dealers take down a larger share than usual and go home with inventory they did not want, financed overnight in . That inventory has to be hedged or sold, which pushes yields up further in the hours afterwards. It is why a bad auction can the whole curve at 1:01pm rather than 1:00pm.

The numbers released with it

A tail is read alongside two other figures published in the same instant. The bid-to-cover ratio is total bids divided by the size sold, a rough measure of how deep the bid was. The allotment breakdown splits the buyers into indirect bidders (mostly foreign central banks and asset managers bidding through dealers), direct bidders, and the dealers themselves. A big tail with a low indirect share is the specific combination that says foreign demand stepped back and the dealers ate it.

Why it matters to this crash

Our bond thesis is that is being asked to absorb more borrowing than it has natural buyers for, from governments and now from AI infrastructure at the same time, and that the price of absorption is a permanently higher discount rate for everything. Auctions are where that thesis gets tested twice a month in public. Yields on a screen can drift for a hundred reasons. An auction is a real transaction of real size at a real price, and the tail is the receipt.

We have already seen it. In our 19 August dispatch on the we noted a $25bn 30-year sale that cleared at 5.216%, the highest auction yield since 2001, and a $42bn 10-year that cleared at 4.683%, the highest since 2007 — and that both tailed. That is the distinction we keep drawing between repricing and indigestion. A market that has calmly decided long bonds should yield more will clear an auction cleanly at the new level. A market that is choking will clear it badly, repeatedly, and force dealers to warehouse the difference each time.

The composition of the marginal buyer makes this sharper. As we wrote on the same day, the marginal holder of US is increasingly a repo-financed relative-value trade rather than a real-money investor, with Federal Reserve research putting the at roughly $830bn as of September 2025. A leveraged buyer will absorb supply at a price, but only while its funding holds. A pension fund that has decided it wants the bond will show up regardless. Tails tell you which kind of bidder was in the room.

What would make this dangerous

A single tail is noise. Auctions are lumpy and one bad afternoon means little. The things that would matter are all directly observable in the Treasury's own auction results.

A sequence of tails across consecutive coupon auctions in the same tenor, rather than one ugly 30-year. Tails widening from low single-digit basis points into the mid-teens or worse. Primary dealer takedown rising well above its recent average while the indirect bid falls, especially at the 20- and 30-year points, which would be the clearest sign foreign official demand is stepping back. Bid-to-cover ratios sliding at the same time, rather than holding up while the tail widens.

The genuinely bad version is a tail large enough that dealers cannot hedge the inventory without moving the market, so the post-auction concession does not fade by the next morning and instead becomes the new level going into the following auction. That is the mechanism by which a funding market stops discovering a price and starts hunting for one. We have not covered evidence that this is happening yet. We have covered two auctions that tailed at the worst yields in two decades, which is where you would start looking.

As seen in

Every dispatch we have filed that touches this. Newest first.

Written 2026-08-20. Crashopedia entries are drafted from sourced evidence, fact-checked against it, and edited by hand. If something here is wrong, it is wrong in git and can be fixed there.