Treasury announced yesterday that it was going to start issuing floating-rate notes, probably at some point next year, although the details are still extremely vague:
Treasury projects that it must now sell an estimated $667 billion of additional debt to the public over the next four years. The floating-rate notes will help give the government flexibility in its increased debt offerings…
Officials confirmed that the notes will not be indexed to the Libor overnight lending rate, which has come under scrutiny in the wake of a scandal that it was manipulated during the financial crisis.
The Borrowing Advisory panel said it was interested in referencing the securities to the DTCC GCF Repo index.
None of this makes a huge amount of sense to me. For one thing, if you're the US Treasury, with $16 trillion of debt outstanding, an increase of half a trillion over four years is not a really huge deal - especially in a world where demand for Treasury securities remains incredibly robust.
What's more, a floating-rate note isn't really a new product. At heart, it's incredibly similar to an investment in T-bills; the only real difference is that you don't need as many roll-overs with a floating-rate note.
And then there's the question of the reference rate. Don't be embarrassed if you haven't heard of the GCF Repo index: it's less than two years old, and it has been tradable for only a couple of weeks.
The Repo index gives an overnight interest rate from the repo markets, and as Stephen Stanley of Pierpoint Securities points out (via Pedro da Costa), there are serious risks there.
Continuing in the mode of offering self-serving advice that would be bad for Treasury, the TBAC recommended that Treasury use the new GCF index as the reference rate for FRNs. This would put Treasury in a position of taking on private credit risk, since, if we had a 2008-style meltdown and general collateral (rates) widened out due to counterparty risk, Treasury's FRN borrowing costs would soar.
Good luck to any Treasury Secretary who would have to explain that to the House Oversight Committee. That sounds like a one-way ticket to a forced resignation.
Essentially, repo rates always include a certain amount of counterparty risk, and they can spike alarmingly during a credit crisis when no one trusts anybody any more, and there's a huge flight-to-safety trade going on. It would be utterly bonkers, in that situation, for the yield on floating-rate Treasury securities to go up, rather than down.
But even if the reference rate was Fed funds, or some other interest rate which doesn't include counterparty risk, there are still problems here: I don't think it should be an overnight rate at all. The way that floating-rate notes work, they pay an interest payment every coupon period, which is calculated by looking at some reference rate. The interest payments can come every quarter, or twice a year, and the reference rate, similarly, is usually a three-month rate, or maybe a one-month or six-month or one-year rate. The period between coupon payments is generally pretty close, on the yield curve, to the duration of the reference rate.
But the Treasury floaters, by contrast, seem to anticipate using an overnight rate to calculate the periodic interest payment. And that's very dangerous, because overnight rates are by their nature more volatile and unpredictable than rates for 30 or 60 or 90 days. During crises, it's quite common to see yield curves which are very steeply inverted between overnight and 30 days: the overnight rate could be in triple digits while the three-month rate could be in the teens. One of the tools in every central bank's arsenal is the ability to ratchet up overnight interest rates in order to crack down on speculative activity; again, there's absolutely no reason why that kind of action should result in higher interest payments on government debt.
Finally, it seems a little bit weird for Treasury rates to be linked to themselves, which is the other obvious alternative: using the three-month Treasury yield, say, as the reference rate for a longer-dated Treasury FRN would set up all manner of odd arbitrage trades with no real underlying value.
If a country is borrowing in someone else's currency, then I can absolutely see why floating-rate notes can make a lot of sense for both issuers and investors. But the U.S. borrowing in its own currency? I really don't see it. It might well provide some nice profit opportunities to the dealers who sit on Sifma's Treasury Borrowing Advisory Committee. But I'm far from convinced that Treasury should listen to them.