It must be a challenge to quantify how the US Treasury’s policy of “regular and predictable” issuance has affected its borrowing costs.
But you’d think a study published by the New York Fed — the guys who actually run the auctions for Treasuries — would approach the topic in a fairly comprehensive way.
An Economic Policy Review piece by Paul Glasserman, Amit Sirohi and Allen Zhang aims to quantify “the short-term cost of forgoing opportunistic issuance of Treasury bills.” The research, done while Zhang was deputy director at the Treasury’s Office of Debt Management and Glassman was consulting for that office, reaches the following conclusion: “the Treasury is not forgoing significant short-term gains by electing to follow a program of regular and predictable—rather than opportunistic—issuance of Treasury bills.”
It also raises some questions.
First of all, limiting the research to the market for bills, which mature between four weeks (trading at a rate of 0.51 per cent, as of Thursday) and 26 weeks (trading at a 0.58 per cent rate) is a confusing choice.
That’s because the US’s commitment to regular and predictable Treasury issuance is sometimes used as a cautionary note in the debate over introducing ultra-long bonds, which would mature in 50 to 100 years.
Borrowing long-term when rates are low should reduce future interest expense — and the potential savings are even higher when investors pay a premium to buy long-term Treasuries, like they did last year, according to the New York Fed’s models. Yet in recent years, Treasury officials have chosen to issue more long-term debt without introducing new longer-term bonds, citing the risk of upending the auction process or losing credibility. (Or perhaps alienating primary dealers.)
So, if the analysts wanted to make a compelling case against tactically choosing its timing and maturity of Treasury issuance, using bills was a strange way to do it. Even if it’s tricky to model ultra-long issuance, why not look at potential savings from tactically issuing debt along the entire yield curve, and factor in possible changes in the premium investors would pay (or demand) to buy long-term debt?
In any event, this study found that regular and predictable bill issuance raises costs by $11m to $49m in a period of about half a year. That’s pretty small when compared to $216bn, which is half the Treasury’s reported interest expense for fiscal 2016.
The analysts also modeled $8m to $20m of savings from the issuance of cash-management bills, which “do not follow a fixed issuance calendar [our emphasis] like other Treasury securities.”
So the cost of regular and predictable issuance is offset by… securities that are not regularly issued.
The study also left out another extremely important factor:
The analysis ignores the ancillary benefits of reducing market uncertainty and facilitating investor planning, both of which could be expected to promote auction participation; the analysis may, therefore, understate the overall benefit of regular and predictable issuance.
If it’s possible to model the effects of information-sharing between bidders, it must be possible to model the effects of prior knowledge of auction timing and size, right?
While there’s a benefit to remaining focussed in analysis, there are more convincing ways to make the case that the Treasury shouldn’t rock the bond-issuance boat. One of those might be a paper from Kenneth Garbade, also with the New York Fed, exploring pre-1970 examples of ultra-long Treasury bond issuance. We’ll get to that next.