This is Kazakhstan, not China. I don't know what it says about the sovereign credit. Photographer: Leon Neal - Pool/Getty Images
This is Kazakhstan, not China. I don't know what it says about the sovereign credit. Photographer: Leon Neal - Pool/Getty Images

Here is an International Financing Review article about credit default swaps on China. There are a lot of those swaps, is the main point, and they have a pleasing, or irritating, lack of sense:

The net notional outstanding in China CDS now stands at US$14.1bn according to the DTCC - a leap of 109% compared to a year ago. The country is now the third largest single-name CDS contract in the world by net notional, despite having only a handful of small outstanding bond issues.

In fact, there's only a little over $2 billion worth of debt that could be delivered into CDS.1 Pimco, the major seller of China CDS, has written around $3 billion notional of protection, meaning that Pimco is long more Chinese government debt on swap than there is Chinese government debt in the world.2

This is weird, but I guess it is not the weirdest thing. Kazakhstan, for instance, has no external government debt,3 but it has half a billion dollars of net notional ($7.6 billion gross) of CDS outstanding. You can buy credit default swaps on Kazakh debt that does not exist and so cannot default, or be delivered into a CDS contract if it did somehow default. As one guy says about China:

"The default protection angle has always been very weak in my opinion. There's just about nothing to deliver and nothing that would trigger it, so it's unlikely to help you at all," said the hedge fund manager.

So why do it? Well here is the classic answer:

"People are not realistically betting on China defaulting. It's more of a mark to market hedge - that's why it trades at around 70bp outright," said [Citi trader Salih] Unsal. "If there really was a China default, people would have bigger worries than a lack of CDS deliverable obligations."

You hear this a lot -- about U.S. government CDS too -- and, while it is correct in its way, it is also deeply unsatisfying.4 Imagine if I wrote you a contract with the following features:

  • You give me money up front.
  • I never give you money back.
  • But, like, it's a mark to market hedge?

That would clearly be nuts. The argument that CDS that can never pay off is somehow "more of a mark to market hedge" is almost equally silly. If there's some structural feature that makes it impossible for CDS to pay off, then it can never be a mark to market hedge. CDS is only a mark to market hedge because its price movements reflect some perceived probability that it will pay off. You can make money trading CDS if that perceived probability goes from 1 percent when you buy to 2 percent when you sell -- it doesn't need to actually pay off -- but if the probability starts at zero and remains at zero then the whole thing is nonsense and you can't address the nonsense by mumbling about mark to market.

Here I am talking like a derivatives purist, not like a trader. Obviously, if you are a trader and you buy a thing and its price goes up and you sell it, you have made money, and if it was nonsense at all relevant times, that is not your concern. The money is not nonsense. But as a purist, I want a derivative's value to have some meaningful relationship with its contractual cash flows in possible future states of the world. If those cash flows are always zero then, I mean, there might be money, but there is no meaning.

Also, financial markets have some glancing acquaintance with efficiency,5 so it's worth looking for something other than the mass-delusion explanation for China (or Kazakhstan) CDS trading. The most reasonable explanation that I've heard is: Well, look, there's no deliverable debt now. But there are various tail-risk-y things that could happen in these markets; various borrowers in those countries -- particularly the banking system, or maybe the local governments -- could run into trouble that they can't get out of. And then maybe the national government would step in to rescue them. And maybe it would borrow money in international markets to fund that rescue. If it did that, then there would be debt for the CDS to reference. And once there's debt for the CDS to reference, then that debt could default, and be delivered into the CDS. The CDS might be getting ahead of the underlying debt, but maybe the debt will catch up eventually.

And in fact, Chinese, and Kazakh, CDS trades wider when those countries' banking systems run into trouble.6

In other words, a five-year China credit default swap isn't so much a bet that China will default on its debt in the next five years, as it is a bet that China will issue some debt, and then default on it, all in the next five years.7 It's a pretty unlikely outcome -- that's why the CDS spreads are so low -- and a doubly contingent one. But double contingency is okay; derivatives can handle lots of contingencies all in one place. Contingency is what derivatives are all about. Not, you would hope, irrationality.

1 According to the article:

Thomson Reuters data reveal only four outstanding Chinese bonds that would be deliverable into a CDS auction. One of these issues is due to mature later this year. After that, there is a 2015 local currency bond of RMB10bn in size, a US$100m dollar bond due 2027 and a US$400m private placement maturing in 2096.

Looking at Bloomberg (CHINA <Govt> <Go>), I see:

  • that bond maturing this year (in October) is 1 billion euros,
  • that $100mm bond due in 2027 and that 2096 USD private placement, though I see it at $100 million, and
  • 20 billion yen ($200 million-ish) of bonds due in 2015.

So I don't know, $1 billion maturing in three months, a few hundred million of other stuff, and I guess RMB 10 billion (around $1.6 billion) of local currency bonds.

2 A couple of caveats to that:

  • I count about $2.8 billion of CDS in the Total Return Fund, from Pimco's filings; IFR says "about US$3bn notional" so close enough.
  • When I say "Chinese government debt in the world," I mean "external debt of the Chinese national government that is deliverable into CDS," which is not quite the same thing.

3 "The last time the sovereign sold debt internationally was in April 2000 with a $350 million seven-year bond"; since then it's pondered, but not yet pulled the trigger on, a new international bond. It seems to have around $22 billion of local debt.

4 Here's another, similarly unsatisfying answer:

"A lot of EM investors are looking for a so-called cheap tail hedge and, given the low spreads, China CDS currently has much less negative carry than other shorts," said one Singapore-based macro hedge fund manager.

The argument "well, it's nothing, but it's a cheap nothing" is a little silly. Buying actual nothing is even cheaper than buying this nothing. Actual nothing costs nothing.

5 That link to my post about Cynk was meant as an efficient-markets joke, but there's some naked short selling relevance here too. You can think of buying CDS on Chinese debt as being "naked short" that debt. In some loose sense there is $16 billion of external Chinese government debt in the world: $2 billion of it issued by China, and $14 billion of it "issued" by CDS purchasers. If all of those credit instruments were fungible -- if you were indifferent between "get long China credit by buying a bond" and "get long China credit by writing CDS" -- then if you just went into the market to get long Chinese government credit, there'd be a 6 in 7 chance that your position would come from CDS. (As Pimco's does.)

Of course they're not fungible.

But when I write about naked short selling of stock, people ask, "well if you allow naked short selling, what's to stop people from shorting more than 100 percent of the outstanding shares?" And the answer is:

  • nothing,
  • nothing stops people from shorting more than 100 percent of the outstanding shares now, and
  • why would you care if people shorted more than 100 percent of the outstanding shares?

After all, if someone is short 300 percent of the outstanding shares, that just means that someone else is long 400 percent of the outstanding shares (the 300 percent that's short, plus the actual outstanding shares). Every seller gets a buyer, etc. Here someone is short 700 percent of the outstanding stock of Chinese government debt, but that just means that someone else -- mostly Pimco -- is long 800 percent.

The non-fungibility of debt and CDS makes this less intuitively troubling. Like, if you're buying Chinese credit, you know if you're buying bonds or writing CDS. Those are very different transactions. If you're buying Cynk stock, there's nothing to tell you whether you're getting shares from the company or from a short seller. In a world where naked short selling is banned, the answer doesn't matter: Whatever shares you get are "real" shares. In a world where naked short selling is allowed, this does matter -- only "real" shares should vote, for instance -- so you need some way of managing it, e.g. by making the naked short shares non-fungible with the real shares.

6 So here is Kazakhstan five-year CDS:

That spike corresponds to a banking crisis in Kazakhstan (and, I mean, elsewhere too), in which the government nationalized several banks. At the time, remember, there was no Kazakhstan government debt outstanding, so it still couldn't default. But the likelihood of (1) government debt springing into existence, and (2) it then defaulting, had gone up.

And here's China:

That late-2011 peak corresponds to a trough for the MSCI China Financials Index, as well as to this article saying, "Most global investors predict China will face a banking crisis within the next five years," so. Generally, China CDS trades sort of inversely to China bank stocks:

7 And a Kazakhstan credit default swap is exactly literally that, mutatis mutandis.

Also of course the China one has the further contingency of: "... and the world wouldn't end." (That's Unsal's "people would have bigger worries" line.) You hear that about U.S. credit default swaps too -- "Why would you buy that, if the U.S. defaults you'd have bigger worries than your CDS payments" -- but there's actually a reasonably satisfying answer there. Maybe in the next five years China will issue lots of external debt, adopt a debt ceiling, fail to raise the debt ceiling due to bitter partisan politics, default briefly, and then get back on track after a CDS auction. I mean. Probably not.

Presumably if Kazakhstan defaulted on its debt your big worry would be your Kazakhstan deliverable obligations.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.