Also published on the Atlantic Monthly’s Business Channel.
Behavioral Polemics
In times of crisis, society develops a need for a theory of the crisis itself. That is, there must be a set of logically interconnected ideas that explains what happened before, what is happening now, and what will happen afterward. This is a laudable instinct, since once we can explain what brought us here, we can agree to not do it again. In the narrative of our current crisis, we as a society have forsaken the need for a coherent theory of events and have settled for a confused and hazy mythology inspired by the very real suffering of millions and the fear of more suffering to come. Like many other mythologies, its development was organic. Along the course of its development, the collective thinking of millions hashed together a cast of demons, sages, and saviors. Bombarded with images, sounds, and stories of calamity, we embarked on a psychological witch hunt, succumbing to the primal need to identify and destroy the cause of our suffering. Somehow, the derivatives market made the cast of demons in this mythology. And it seems that even some of the sharper tools in the box believe, in my opinion without any real theoretical basis, that the derivatives market has earned its place in that cast.
A Tall Order That Came Up Short
In a recent Financial Times article, George Soros explained his own theory of the crisis. While certainly less mythological than most others, it is still theoretically unsound and factually inaccurate in several respects. In this article I take issue only with his treatment of derivatives. I happen to agree with many of his opinions on the Efficient Market Hypothesis and in a limited sense, what he calls “reflexivity,” which holds that a market price is not only a reflection of fundamentals but can also affect fundamentals.
Soros begins his argument by explaining how the risks of shorting a stock differ from the risks of owning a stock. Although there are other ways to short stock, let’s assume that we are talking about the method that involves borrowing stock, selling it, and then repurchasing the stock (hopefully at a lower price) and delivering it to the lender. Soros correctly points out that by shorting a stock, you leave yourself exposed to an effectively unlimited amount of risk, since the stock price could rise to arbitrarily high levels (back when those things used to happen), and you would be responsible for going out to the market, repurchasing it, and delivering it to the original lender, leaving you on the hook for the difference between the sale price and the repurchase price. He says that this means that ownership of stock and shorting a stock have “asymmetrical” risks. He then states that this asymmetry “serves to discourage the short selling of stocks.” While that statement is a bit unclear and incomplete, I get what he’s trying to say, and I can live with that.
He then incorrectly claims that shorting a bond through credit default swaps creates similar “asymmetrical” risks between buying protection and selling protection. He states that protection buyers have “unlimited profit potential” while protection sellers have “unlimited risks”. That is categorically false. Those who read my blog often know that the maximum risk exposure of a protection seller is capped at the notional amount of the CDS, which also limits the maximum reward of the protection buyer. Without the jargon, this means that both sides agree to the amount of protection bought at the outset of the contract. The protection seller never has to pay more than that amount and the protection buyer will never receive more than that amount. Thus, both sides of a CDS have a cap to the amount of credit risk they are exposed to.
This pulls the cornerstone out of Soros’ argument that CDSs over incentivize buying protection. There may be other arguments to that effect, though I doubt it given the bilateral nature of the contracts and the rule of no free lunch. In any case, his is certainly not one of those arguments.
Betting On Failure
Soros then argues that both the shorting of stock and the buying of protection through CDSs contributed to a lack of confidence, by lowering the price of stocks and raising the cost of protection respectively, each of which accelerated the demise of several institutions. This second argument is much more difficult to debunk as a practical matter. But that doesn’t imply that it is sound. The validity of this argument, and its inverse, depend on which occurred first: the lack of confidence or the shorting/protection buying. My instinct is that we may never know.
Moreover, the cyclical self-fulfilling dynamic that Soros is alluding to can be created without derivatives, shorting, or any of those fancy techniques: see e.g., bank runs. Whenever the dynamic running a market is a high level of demand for short term capital (which is what occurs during a liquidity crisis) coupled with the fear that you will be the last to exit a market, prices will plummet. This is because each participant has an incentive to maximize the short term value of its assets. And each participant knows that if it doesn’t convert its assets to cash or other low risk short term assets, the collective selling of others will erode the value of any assets that it doesn’t sell. So, even if a given participant doesn’t want to sell and values its assets at a price that is higher than the current market price, it will sell anyway if it needs capital in the short term.
Note that this logic creates an exception to the EMH. That is, when there is insufficient short term capital, there isn’t enough available cash to bring prices back up to efficient levels. This logic also explains the recent mad rush into short term Treasuries.
In short, even if Soros is correct that short selling and protection buying exacerbate self fulfilling market dynamics, which I doubt, they can occur anyway.
What Mr. Soros meant is perfectly obvious:
1) Banks and investment banks that sell credit protection do so on the basis of enormous leverage. While the total amount of the loss is capped by the notional value of the CDS, the loss can easily exceed the capital of banks. And at that point the taxpayer steps in. Conversely, if the bank makes money on the short CDS, the taxpayers are not invited to the party. The risk is indeed asymetrical — taxpayers have unlimited risk and zero upside. I know, I know, the folks at Goldman Sachs and Morgan Stanley are absolute genisues who never have had a losing trade and we should be grateful that Hank Paulson used our grandchildren’s tax money to keep these firms afloat.
2)As Deutsche Bank recently discovered, there is need not be any short-term relationship between the CDS price and the price of any underlying bond. The price of a corporate bond reflects not just a credit premium but other factors including a liquidity premium. The liquidity premium can soar leaving the credit premium unchanged. A long in the corporate bond hedging with a CDS will have much to explain. This ‘premium’ only goes away on maturity of the bond.
Barry,
That is not even close to what Soros said. He was making statements about the risk profiles of investments. He did not mention tax payers at all.
Bonds are illiquid, and therefore prices do not move as fast in the bond market as they do in the CDS market. Big deal.
Apparently it was a big deal to Deutsche Bank.
Porsche!
I think we can assume that the asymmetry Soros is alluding to is not to be understood on the basis of a one-to-one contract, but on one of one-to-many, which, in practice, is the way CDS contracts are: One protection seller against many protection buyers. Under this practice the exposure of the seller becomes extremely high indeed.
Now to consider that Soros does not know about the capping of CDS, is not very serious, in my opinion.
I don’t think we can assume that. You can feel free to assume that.
If he had just made the statement that CDS offer unlimited upside, you might have a point. But his analysis, comparing shorting bonds with shorting equity, suggests he believes that individual CDS contracts offer unlimited upside, just like shorting equity.
When someone as influential as Soros makes inaccurate statements, I’m going to call him out.
I agree with you on a stylistic point of view, the comparison is poorly chosen.
I just prefer crediting Soros with poor writing skills than an ignorance of what CDS contracts are at a very basic level.
I have to agree with the general thrust of your two previous commenters. though I’d state it differently. One can criticize How it was stated by Soros, but the examples in the market of naked protection writers (AIG, DB)– naked in the sense that their capital position was never sufficient to actually pay off their exposure certainly illustrate the point.
The one-to-many aspect which allows more CDS notional to be outstanding that the face of the bonds they theoretically stand in for, allows the scale of any specific credit event to be magnified to a degree many times larger than the underlying event. If like AIG, you believe(d) the risk in a given name is essentially zero, you can happily write protection without limit. This was certainly true for AAA writers who did not have to post collateral. It remains to be seen how much exposure the remaining counter-parties can tolerate and how well-implemented the collateralization of exposure is for the non-AAA writers. This year and next should be enlightening in that regard.
Regarding the issue of a liquidity spread for CDS vs. physical bonds, yes that’s a reasonable explanation and no doubt represents most of the spread most of the time, but it’s also reasonable to wonder whether the full spread is properly attributed to liquidity or whether the lower yield for CDS sellers vs. bond buyers has at any time reflected the competition among writers who underestimated actual default risk and were therefore willing to reach their revenue objectives by increasing position size — making it up in volume.
Last thought: It strikes me that naked CDS writing is tantamount to, but the inverse of, naked shorting of common stock. In each case there’s no initial cost to the seller, the seller’s counter-party ends up with a fail (though this isn’t discovered with CDSs until an event). Both trades rely on counterfeiting the underlying value the instrument it is supposed to represent. In the case of equities the results have been understood sufficiently to be made illegal (though the exceptions are interesting).
Is this what Soros meant? I can’t say. But at the underlying effects seem at least to be compatible with his claims.
“naked protection writers (AIG, DB)– naked in the sense that their capital position was never sufficient to actually pay off their exposure certainly illustrate the point.”
Insurers never have enough to cover the total face value of their policies. That’s how they make money. They work with probabilistic models that predict how many of their policies will require payment.
“The one-to-many aspect which allows more CDS notional to be outstanding that the face of the bonds they theoretically stand in for, allows the scale of any specific credit event to be magnified to a degree many times larger than the underlying event.”
CDS do not create actual losses to the economy. For every losing end of a contract there is a winning end. So even if people buy and sell more notional amount than previously existing par value of a particular debt, the CDSs will not create losses. They simply shift the allocation of cash from some parties to other parties.
“In the case of equities the results have been understood sufficiently to be made illegal.”
Naked shorting of stock has nothing to do with what has been mistakenly called naked shorting through CDS, which is shorting without owning the underlying bond.
[...] pundits, and perhaps most ironic, financiers! Even that giant of finance, George Soros has loused up explanations of how credit default swaps work. I’ve called out economists in the past for their mumblings on credit default swaps and the [...]