Treasury bought back $6B in bonds, but dealers say bigger isn’t enough

The US Treasury bought back $6 billion of older government bonds on Nov. 4, hitting the full limit of its latest operation. On paper, that looks like a success for a program designed to make the Treasury market easier to trade.

But there's a catch. One bond accounted for $4.9 billion, or 82% of everything Treasury agreed to buy. Add another $950 million from a second bond, and just two securities made up almost 98% of the entire purchase. Of the 40 bonds eligible for the operation, only seven were bought.

This is the kind of concentration Wall Street dealers have been worried about for months. In feedback published by Treasury on Nov. 4, dealers said the larger buybacks were helping market liquidity.

However, they also explained why simply increasing the program size might not be enough to achieve that.

Their feedback came before Wednesday's operation, so they weren't responding to this particular result. We now know the Treasury can buy billions of dollars of bonds and still leave dealers holding the securities they hoped to sell. The amount purchased is important, but so is how those purchases are spread around.

The Treasury market is enormous, but not every government bond trades equally easily. The newest issues attract the most attention and typically have the deepest markets. Older bonds, known as off-the-run Treasuries, can be harder to sell quickly without accepting a lower price.

That's where dealers come in. They buy bonds from customers, keep them on their books, and look for someone else willing to take them.

But holding those bonds costs money, sometimes a lot of money. That's why dealers often have to borrow to finance their positions, exposing them to price changes while they wait for another buyer. The longer they hold a bond, the more those costs and risks compound.

Treasury's liquidity-support buybacks give them another potential buyer: the government itself.

In September, Treasury began larger buybacks of longer-dated bonds, at least doubling the previous $2 billion maximum for those operations. The idea was to provide a more reliable outlet for older securities.

According to the Nov. 4 dealer feedback, most dealers reported at least modest improvements. They said the purchases had made it easier to find an exit for older bonds, reduced the risk of sudden price distortions, and helped keep pricing relationships between similar securities more stable.

While that's meaningful progress, the Treasury got mixed answers when it asked whether the changes had improved dealers' ability to handle customer trades.

A lack of unity among dealers is what makes these buybacks so controversial. We now have dealers that welcome a large government buyer but aren't necessarily becoming more willing to buy bonds from customers.

If the Treasury keeps concentrating its purchases in a handful of issues, dealers holding other, older issues still have to find other buyers or keep financing the position. In other words, the buyback program can work without working equally well across the entire market.

Dealers wanted better, earlier communication about operations, smaller purchases more frequently, and limits that would help distribute Treasury buying across a wider range of individual securities.

More frequent operations would give dealers more opportunities to sell, as they'd get another chance sooner.

Limits on how much of one security the Treasury can buy could also prevent a single bond from absorbing most of the available capacity. Each bond has its own identifying number, called a CUSIP, so dealers suggested using per-CUSIP limits to spread purchases more broadly.

Some dealers specifically warned about the risk of missing a concentrated buyback and being left to manage the interest-rate exposure from bonds that remained on their books.

This is exactly what we saw on Wednesday. According to the official data, Treasury received $39.6 billion in offers for an operation capped at $6 billion. That means dealers and other eligible participants offered more than six times what Treasury ultimately accepted.

And despite that demand, one bond maturing in November 2041 absorbed more than 80% of the purchases.

It's worth being careful about what those numbers tell us. Treasury doesn't publish the full breakdown of rejected offers, so we don't know whether every one of the 33 eligible bonds that received no purchases was actually offered. We also can't assume Treasury made a bad decision by concentrating its buying.