Felix Salmon

The Treasury held one of its regular official meetings with Wall Street on Tuesday, to talk about the market in government debt. The minutes are here, and reveal a presentation on the possibility of issuing extremely long-dated debt:

The presenting member first discussed “ultra-long” bond issuance, which were defined as securities issued with a tenor of 40-, 50- and/or 100-years. The member noted significant demand exists for high-quality, long-duration bonds from entities with longer-dated liabilities. It was noted that duration tapers off rapidly with maturity and is dependent on the underlying coupon on the bond. As a result, liability-driven investors would likely use the STRIPS market to capture additional duration exposure.

As the WSJ notes, referring to but not linking to the minutes, a recent Mexican century bond met with very high demand, and a US version would probably be similarly successful. Not that it’s going to happen:

A senior Treasury official, speaking after the meeting, shot down the idea of issuing “century” bonds, which would mature in 100 years, but was open to discussing expiration dates going beyond 30 years, the longest U.S. government bond available.

Why is the government dismissing the idea out of hand like this? The WSJ declares that “issuing such long-dated bonds would have an obvious benefit for the U.S. government,”, and against that says only that “the prospect of deferring debt payments for generations carries political risk,” without explaining what kind of political risk they mean.

The fact is that the decision not to issue century bonds makes perfect sense even absent vague premonitions about possible political pushback. There are three main reasons why:

Firstly, as the minutes themselves note, there’s already a healthy market in STRIPS. When the government issues a 30-year bond, dealers can strip the coupons from the principal and trade them separately — which means that if you want an ultra long duration Treasury bond, you can already buy a pretty liquid 30-year zero-coupon instrument very easily. A century bond would not improve on that: a zero-coupon 30-year bond has a duration of 30 years, while a century bond with a 4% coupon trading at par actually has a significantly shorter duration of just 24.5 years. (Check out the Investopedia duration calculator here.)

Secondly, Treasury does not go in for opportunistic issuance. If it wants to add another arrow to its quiver of liability-management instruments, it’s going to add something which will be issued repeatedly and predictably over time. A regular 50-year bond might be a possibility, but a 100-year bond isn’t: no one issues such things regularly. A century bond would be a weird off-the-run oddity in a market designed to be as liquid and orderly as possible.

Finally, a century bond maximizes the interest-to-principal ratio of bond repayments: that 4% century bond ends up paying out $4 in interest for every $1 borrowed. The US government has no need to commit to paying back many multiples of its budget deficit over the course of generations to come: the deficit is big enough as it is already.

Investors love century bonds, largely because they’re very convex: when interest rates fall they rise in value a lot, but when interest rates rise they don’t fall so much. And to a certain extent it makes sense for borrowers to give investors what they want. But ultimately a win for investors is likely a loss for the borrower. And so absent any compelling reason to take the bait, Treasury’s quite right to rebuff Wall Street requests for century issuance.