Making a market in bonds

On 19 August the US Treasury doubled the size of its long-end bond buybacks. Long yields fell about ten basis points. By the close on 20 August this had reversed.

The intervention was initially rejected. Markets have a habit of testing the resolve of policy makers. That is the event. The mechanism works like this.

The Treasury raised the maximum size of its liquidity-support buybacks for long bonds from $2bn per round to $4bn. The following day the Secretary spoke.

“We are going to make a market in these,” Bessent said, and the size “could be more than $4 billion per issue.” He said part of it was signalling.

And what he was trying to get the markets to notice is that the yields on long bonds do not reflect the underlying fundamentals. He would like the market to reconsider. And then reprice for lower yields. The jawboning approach, tell the market what you want.

The week before had gone badly at the long end. The $31bn thirty-year auction cleared at 5.216%, the highest since 2001.† On 18 August the thirty-year touched 5.33%, a nineteen-year high. Federal debt passed $40 trillion the same day. The last trillion of it took 154 days.

The twenty-year auction was on the afternoon of the nineteenth, hours after the buyback was doubled. The twenty-year cleared at 5.204%. Indirect bidders — the usual proxy for foreign demand — took 62.9%, about three points below the recent average. The intervention was half a day old and the first auction to test it went badly.

This week the front end repriced for rate rises. The long end is pricing something else. It is not only an American problem. The UK thirty-year is near 5.8%, the highest since 1998. Japan’s super-long yields are at records too.

A little indigestion from the size of the feast? Everyone issues long debt into the same finite pool. The game’s afoot.

When central banks and the Treasury seek lower rates does the long end implement their desires, or misbehave? Because keeping control of the whole interest rate curve is the name of the game. Much rests upon it.

Since the Fed began lowering rates in September 2024 it has taken 175 basis points off the overnight rate. Over the same period the thirty-year has gone from about 3.9% to 5.21%, and the ten-year from about 3.6% to 4.72%. In the seven previous cutting cycles since the 1980s the ten-year fell within a hundred days of the first cut. Every one of them. This cycle so far is the exception.

The Treasury does not create money like the Fed. Every dollar it spends buying a long bond it must first borrow. So it borrows short and lends long. And accepts the rollover risk incurred.

And the Treasury is not involved with debt retirement but in swapping duration, long for short, knowing you will have to renew them earlier than you thought. Who knows what the market will be offering on that day. But a problem for another day — so it can be deferred.

The Treasury knows that borrowing short and lending long comes with interest rate risk. It increases your sensitivity to movements.

A central bank’s promise is their bid has no ceiling. It can print. The Treasury’s bid has a ceiling: it is funded by issuance. Bessent said part of his message was signalling.

And he is part right. The Treasury can’t really make the market in T-bonds. It can only nudge. So in reality it’s all signalling.

But his inference is clear, the price of debt is wrong. The fundamentals are better than the market believes. Please Mr Bond Market, take a look and reconsider.

If the long end will not answer the Treasury, the pressure moves to the central bank — which is where they keep the bazooka. Yield curve control.

The Treasury’s intervention has not failed but it does have a half-life. Intervention has a habit of becoming the condition it was meant to cure. Each round makes the next one larger. Just ask any central banker.

Two things worth watching. The buyback cap went from $2bn to $4bn in August. If it goes to $8bn, the intervention is rolling up.

And there’s central bank verbiage. Warsh describing the long end as a policy concern rather than a market outcome. That’s the bazooka being loaded. They will tell you what they are doing before they do it. Which might work in itself.

Warsh spoke at Jackson Hole on the day this note closed and did not mention the long end at all. Unsurprisingly, far too early for that. But nevertheless the long bond needs to come down. Or with a bit of deft market control, perhaps they can flatten the curve a bit. The tougher the Fed sounds on inflation, the less premium the long bond needs. Still hurting at the longer rates but less noticeable.

The long bond is not misbehaving. It is being asked to fund two trillion dollars a year at below the market. It has a legitimate complaint.

† Though the auction series has a gap in it. The thirty-year was suspended after 2001 and not sold again until 2005, so that comparison steps over four years with no auctions in them.