Why negative swap spreads are maybe not so surprising

Sure, the inverted 10-year swap spread is gone, and Goldman Sachs thinks it is not coming back.

But the inversion remains in the 30-year spread, which is around -20bps as I write. And that has persisted for most of the past couple of years, so there is still a mystery to explain.

Last time, I described a “financial perpetual motion machine” built from such a negative spread. Some people dismiss such constructs out of hand, since everyone knows free money — like perpetual motion — is impossible. But I believe you can learn something by examining one and figuring out exactly where it fails.

Recall the negative swap spread perfect arbitrage: Borrow short at LIBOR; use that to buy Treasuries; swap the fixed Treasury coupons for (floating) LIBOR; use those payments to pay interest on the short-term loan; and roll the short-term loan every three months.

What could possibly go wrong?

First, nothing really requires zero capital. But I do not think this is the answer. I suspect a perfectly-hedged structure like this would not tie up much capital — even regulatory capital — but I admit my knowledge of accounting rules is insufficient to be certain. Regardless of such regulatory constraints, I also suspect large banks would find a way around them if it meant truly risk-free money. Besides, the inversion in the 30-year is so large and has lasted for so long… Perhaps capital constraints were sufficient explanation in early 2009, but surely by now somebody would have come up with the capital to fund such a pure arbitrage?

Second, the U.S. Treasury might decide not to pay your coupon some day. This is the “US default-risk meme”, and like Bond Girl, I am skeptical. (Although anything is possible. Anybody know a free source for monitoring U.S. Treasury CDS?)

Third, the counterparty on your plain-vanilla swap might decide not to make his or her payment some day. But I consider this even less likely. Never mind that swaps are collateralized… If defaults started happening in the swaps market — the multi-hundred-trillion-dollar global swaps market — that would be even more of an Armageddon than U.S. sovereign default. Governments and central banks around the world would almost surely step in before the dominoes even started trembling.

Fourth, and finally, you might not be able to roll your loan every three months. Maybe you think this is obvious; “of course” you cannot be certain that you will be able to roll a short-term loan. But recall our hypothesis: “You” are a large bank borrowing in the London interbank market, and you have a swap paying LIBOR. If you are having trouble rolling your loan, then LIBOR is skyrocketing. Indeed, the definition of LIBOR is precisely the rate at which a large bank can, in fact, roll that loan. Since the swap payments track LIBOR, they should always be sufficient to fund the loan, no matter how expensive it becomes.

The punchline: LIBOR is a fiction.

Ms. Tett, September 2007

“The Libor rates are a bit of a fiction. The number on the screen doesn’t always match what we see now,”
complains the treasurer of one of the largest City banks.

“The screen will say one thing but people are actually quoting a different level, if they are quoting at all,”
says one senior banker.

WSJ MarketBeat Blog, September 2008

The London Interbank Offered Rate, or Libor, has been driven sharply higher as banks have adopted the fetal position, reluctant to lend to each other because of need for their own cash. But those rates may not tell the entire story.

What they’ve found Thursday morning is an even greater reluctance to lend to each other than is expressed in Libor rates. Lou Crandall, chief economist at Wrightson ICAP, a subsidiary of ICAP, says the Thursday New York three-month rate was 3.71%, or 0.51 percentage point above Libor.

You can find similar stories throughout the crisis. What we learned in the last 2½ years is that the British Bankers’ Association will lie about LIBOR during times of financial distress. So if you enter this hypothetical arbitrage on (say) the 30-year swap spread, and sometime in the next 30 years another financial crisis develops, you may find that the variable side of your swap — which tracks the BBA’s reported LIBOR — is insufficient to cover your true cost of borrowing.

The BBA falsified LIBOR because they were trying to avoid making matters worse. I believe inverted swap spreads are an unintended consequence. Theoretically, swap spreads represent “systemic risk”, both current and anticipated. But if that were ever true in practice, it certainly is no longer, because we now know that LIBOR becomes understated during systemic events.

In fact, I would argue that a negative swap spread represents the market’s discounting the risk of another meltdown. Which is the exact opposite of what it is supposed to mean. How ironic.

7 comments to Why negative swap spreads are maybe not so surprising

  • Bond Girl

    Markit publishes the last quotes on the most liquid CDS here:


  • bob

    I think you hit on something with the LIBOR “lie”. Not so much that it’s a lie, but it isn’t a direct feed of actual transactions. There’s no hi/lo/open/close. It’s a daily ‘survey’ of sorts. So the actual rate on any real position is probably not a direct match to the published LIBOR.

    What would be interesting to see is how much of a spread between published and actual LIBOR makes this risk-free trade profitable on the surface, and how big a position some bank would have to take to make the trade worthwhile. Unfortunately for me, I don’t have the accounting skills to do that.

    Even if it is risk-free income to the position holder, the choice of using money for that purpose has to be weighed against the other alternative means of making money. Is this risk-free arbitrage really yielding more than a straight direct purchase of Treasuries? I would be surprised if one could make this arbitrage play without tying up capital in some way, otherwise one could apply the strategy to an infinite sized position. And if I was some shrewd person who went down that road (infinite position), how would all my transactions affect the rates in the marketplace for all the legs in the strategy? There’s a counter-party to every transaction, so one can’t just start accumulating positions and not create counter-party demands throughout the system. I would venture to guess the negative spread represents the sum-total cost of commissions and fees throughout the system of taking on that arbitrage.

    And here’s another angle. You said, “If defaults started happening in the swaps market … that would be even more of an Armageddon than U.S. sovereign default” and “swap spreads represent “systemic risk””. If those two statements are correct, then the negative spread is perfectly consistent. The negative spread is the market’s way of saying “the likelihood of default is as absurd as the moon crashing into the earth in our lifetime, so if you want to buy some insurance against that event we’ll be willing to sell you as much as you want to buy.”

  • negahban

    there is no such thing as dollar denominated US Treasuries CDS, BUT there is one denominated in the euro.


  • Ted K

    You guys are at your best when you post specifically on bonds and interest rates. Plus there is a void of good Bond info in the blog world so that’s a good way to differentiate yourself. I ordered Kwak and Johnson’s book today. I’m like a 10 year old, I still get excited about getting books through the mail. I can’t freaking wait.

  • acemaxx-analytics

    Yes, negative swap spreads are not surprising.

    The whole thing has to do with the hedging demand related to many derivates. Options traders, who have structured options with discontinuous payoffs, have high exposures in the segment of 10 – and 30-year bonds. They strive fixed payments particularly in the 30-year segment of swaps. Their demand is driving the spreads of the corresponding swap to bottom.

    That’s why we have negative swap spreads particularly in 30Y segment. Because everybody wants to pay floating and to receive fixed payments. Why ? Because longer duration fixed rate yields more than short term. That’s why we have been watching massive demand for fixed interest payments like in Japan during the time the country was in liquidity trap. They have got negative spreads in 2001 too. Cause the world economy stil stuck in a liquidity trap and macroeconomics at zero bound doesnt apply.

    Just because swap spreads became negative doesnt mean that it is to be seen as an imminent sign of increasing risk for US bankruptcy. There was a very „nice“ chart from Japan brought by Tracy Alloway via FT Alphaville: 2Y, 5Y and 10Y swap spreads turned negative around 2001 in Japan.

    Japan’s negative swap spreads

    In my view dipping below zero for the swap spread in the US is a result of massiv derivative hedging-related demand. Its not a matter of „credit risk“.

  • British Bankers’ Association here. Sorry – been taking an Easter break so came to this posting late.

    Nemo, how could (or would) we “lie about LIBOR during times of financial distress”? The LIBOR fixing mechanism is completely transparent.

    The individual rates (from which we calculate the banchmark) are all available through the main financial data providers. You can look them up yourself.

    The calculation (actually carried out by ThomsonReuters, after assiduous checking of the data) is simple and transparent: they throw out the top and bottom outliers and take the average of the middle eight rates. Again, you can calculate it yourself from the submitted data, since it is all available.

    Bob is right to note that it’s a benchmark rather than a feed, but the banks do have to prove that they submit authentic rates or risk expulsion from the panel. And the governing body for LIBOR – the Foreign Exchange and Money Markets Committee – is independent of the BBA, though we do provide it with administrative support. Chapter and verse on the calculation and governance of LIBOR is available online at our website http://www.bbalibor.com.

    So the LIBOR-setting mechanism and its governance are as transparent as we can possibly make them. But we have stated – again, publicly – that we are always open to ideas for their improvement.

  • JoeLeTaxi


    “the banks do have to prove that they submit authentic rates or risk expulsion from the panel”

    what sort of proofs would be considered valid here ?

Leave a Reply