One of the things I’ve been missing in recent months is a smart and detailed attack on credit default swaps: my slogan when it came to CDS is that "the less you know, the worse they look". So I’m very happy that Arnold Kling, in his testimony to Congress yesterday, devotes five closely-argued pages of prose, totalling over 2,700 words, to explaining why he considers credit default swaps to be primarily useful only for "creating systemic risk".
Of course, I think that Kling is wrong. But at least now there are clear and specific arguments to respond to, rather than vague hand-waving about how "complex financial instruments" caused "trillions of dollars in losses".
All the same, I do wish that Kling had been clearer. For instance, he starts off his CDS section with this:
A number of commentators have pointed out that the loss of market value at financial institutions
appears to substantially exceed the markdown in housing values. I believe that credit default swaps
played a major role in causing this loss multiplier effect.
A couple of numbers here would have been very useful. "Market value" is normally considered to be market capitalization; I don’t know what the fall in market cap of US financial institutions has been, but let’s call it a couple of trillion dollars. Meanwhile, housing values have fallen say 20% from a peak of about $25 trillion — an erosion of wealth of about $5 trillion. And that’s not even counting the markdowns in commercial mortgages, leveraged loans, and other bank assets. Even if he’s only including the value of homes with mortgages, I’m not sure how Kling has come to this conclusion.
Kling then continues:
In my opinion, the problem with credit default swaps is not counterparty risk. The problem is that there
is no natural seller of default swaps.
Kling has clearly never spent much time with big bond investors. If you buy a bond, you get a steady income stream, while running the risk that you might lose substantially all your money. If you sell default protection, you get exactly the same thing: the only difference is that in some (but not all) cases, you don’t have to put all your money up immediately.
Basically, any institutional investor is a natural seller of default swaps. They’re ways of deploying risk capital: if there’s no event of default, you end up with more money at the end of five years than you started with. If you’re selling a CDS, you’re an investor: you’re taking on a certain amount of risk, in exchange for an expectation that you’re going to come out ahead. What makes Kling believe that such investors don’t exist?
Kling never stops to think that there might be such a simple motivation for writing default protection. Instead, he says that all CDS sellers are taking their side of the deal for one of two reasons: either they’re betting on diversification, or else they’re engaging in "dynamic hedging".
I don’t get this at all. Kling’s arguments against diversification — the idea, basically, that correlations are low — make perfect sense, but I don’t see the connection to CDS. He writes:
One way to think about credit default swaps
is that some quantitative financial engineers believe that a diversified portfolio of B-rated bonds can
have lower risk and a higher reward than a lone AA bond.
I think Kling’s confusing credit default swaps, here, with collateralized debt obligations, or CDOs. To be sure, at the height of the credit bubble, there were some CDOs which were funded solely with CDS, rather than cash bonds — but the diversification idea underlying the CDO market has nothing to do with CDS, and applies equally to plain-vanilla cash CDOs.
What’s more, only part of the ratings transformation in CDOs came from diversification: most of it came from tranching and overcollateralization.
As for dynamic hedging, Kling has constructed a truly elaborate fiction. Whence it sprang I have no idea, but do check it out:
Another strategy for selling credit default swaps is dynamic hedging.
Suppose that the seller of a default swap on a bond issued by XYZ Corporation starts to suspect that the
probability of a default on that bond is increasing. The seller can hedge its risk by selling short either
XYZ Corporation stock or other XYZ Corporation bonds. In the event of a default, the loss that the
seller will take by having to purchase the defaulted bond at par will be offset by the gains on the short-selling.
The problem with dynamic hedging is that it only works in a relatively stable market, in which few
others are attempting similar strategies. When everyone is trying dynamic hedging at once, the result is
a wave of short-selling that overwhelms markets.
Overall, then, if dynamic hedging is used by sellers of credit default swaps, they generate systemic risk.
The individual swap sellers form contingency plans which, in the aggregate, are not compatible. When
swap sellers perceive an increase in risk, they all seek to short securities simultaneously, creating the
equivalent of a bank run.
Arnold, did it ever occur to you that if you’ve written credit protection on a credit, and you want to hedge that position, the easiest and most obvious way of doing so is to simply buy credit protection on the same credit? The CDS market is just that — a market. You make bets on whether spreads are going to widen out or tighten in, and you make money if you bet right, and you lose money if you bet wrong. Just as in the stock market, you can unwind your position at any time simply by taking an equal and opposite position. There’s really no need whatsoever for this kind of dynamic hedging.
Kling then moves on to what he calls "liquidity risk" — which is really nothing more than standard margining requirements in any derivatives market. If the market moves against you, you’ve lost money, so you need to put up that money as collateral. Kling is very exercised about the fact that when a protection seller loses money on a trade, they have to put up collateral, even if there hasn’t been an event of default. But if I buy an out-of-the-money equity put or call, and the markets move against me, I still need to put up margin or collateral, even if the option hasn’t moved all the way into the money. It’s no different.
"The seller may lose liquidity due to margin calls or lose solvency due to the change in
the value of the swaps," writes Kling: well, yes. That’s how derivatives markets work. If this is an argument against CDS, then it’s an argument against all derivatives: equity options, commodity futures, interest-rate swaps, you name it.
If you think Kling’s argument is a little bit funny so far, just wait till you see where he goes next:
The demand for safe collateral has two adverse effects. First, it increases the demand for short-term
Treasuries, artificially raising their price (lowering their interest rate), while driving down the prices
(driving up the interest rates) on other securities. Second, it squeezes the liquidity of sellers of credit
default swaps, threatening the viability of those firms, which in turn triggers even more demands for
collateral in the system and even further flights to safety.
First: No. Treasury bills are the single most liquid market in the world, and a the fact that a few of them are being used as collateral in the CDS market has zero effect on T-bill interest rates. And I don’t even understand why increasing demand for T-bills would drive up interest rates on other securities.
And if I’m a seller of credit protection these days, the amount of collateral I need to put up is fixed, as per the Isda master agreement: it has nothing to do with my own viability. If I find myself with a lot of collateral calls, well, I’ll either need to unwind my CDS trades and take a loss, or else I’ll have to find that collateral. In neither case is the "demand for collateral in the system" increased.
All I can think of is that Kling is thinking of a situation where my own dubious solvency means that CDS written on me widen out, and that people who have sold protection on my debt have to put up further collateral. Which might have happened with the monolines, but it’s hardly a systemic problem. After all, CDS, like any other derivatives market, is a zero-sum game: if some people are losing money, other people are making money. That kind of thing doesn’t trigger "even further flights to safety".
Kling then wanders off into a casino metaphor I don’t understand, before eventually circling back to the real world:
We have had Mr. Bernanke and Mr Paulson running around with huge bags
of money, frantically dumping it on the tables in casino. $30 billion to cover Bear Stearns’ bets, $100
billion to cover AIG’s bets, $300 billion to cover Citigroup’s bets, and so forth.
But two of these things are not like the other. Bear Stearns’s losses had nothing to do with credit default swaps. The government backstop on Citigroup assets is on just that — real assets — not on credit default swaps either. The only one of these three cases which is CDS-related is AIG. And yes, as I’ve argued before, AIG and other triple-A rated insurers did indeed pose a nasty systemic risk, precisely because they didn’t need to put up collateral on their CDS so long as they retained their triple-A ratings. In other words, the system of collateral calls doesn’t make things riskier; it makes them much safer. Yet Kling gets this entirely the wrong way around:
The way I see it, we should have punished
the impatient grabbers and instead rewarded firms that were willing to sit back and let the contracts
Not at all. By far the most systemically damaging attitude in the CDS market was the one held by the triple-A insurers: we don’t need to worry about the market value of our CDS because we only need to pay out if and when "the contracts play out" and there’s an actual event of default. The discipline of posting collateral to "impatient grabbers" would have prevented tens if not hundreds of billions of dollars in losses at AIG, MBIA, Ambac, and elsewhere.
So that’s Arnold Kling’s Congressional testimony. I should also briefly address his blog entry, about me. Kling says that if I’m right and it’s pretty much impossible to back out implied default probabilities from CDS prices, then there can’t be any use to them at all. That’s just silly. The purpose of the CDS market was never to give people implied default probabilities. But actually, if the Default Recovery Swap market takes off, maybe we will, finally, be able to get a better idea of implied default probabilities than we’ve ever had in the past. Which would be a nice ancillary benefit to the CDS market, but hardly the main one.