Wolfgang Munchau provides nice succinct overview of some of the recent debate surrounding a source of financial system risk that has suddenly captured the popular imagination: credit default swaps. For those new to the concept, credit default swaps are effectively insurance. A protection seller (think insurer) takes the risk of default on a reference entity (a corporation, an index) in return for periodic payments (think insurance premiums).
But unlike insurance, this market is unregulated, and is about to be tested for the first time in a serious way (single names, like Adelphi, have gone bankrupt, but the market has yet to experience the consequences of a downturn-induced deterioration in credit quality).
Munchau’s piece does not break new ground, but it makes a tidy, high level summary of key issues. From the Financial Times:
If this had been a mere subprime crisis, it would now be over. But it is not, and nor will it be over soon. The reason is that several other pockets of the credit market are also vulnerable. Credit cards are one such segment, similar in size to the subprime market. Another is credit default swaps, relatively modern financial instruments that allow bondholders to insure against default. Those who such sell such protection receive a quarterly premium, based on a percentage of the amount insured.
The CDS market is worth about $45,000bn (€30,500bn, £23,000bn). This is not an easy figure to imagine. It is more than three times the annual gross domestic product of the US. Economically, credit default swaps are insurance. But legally, they are not, which is why this market is largely unregulated.
Technically, they are swaps: two parties swap payments streams – one pays a regular premium for protection, the other pays up in case of default. At a time of low insolvency rates, many investors used to consider the selling of protection as a fairly risk-free way of generating a steady stream of income. But as insolvency rates go up, so will be the payment obligations under the CDS contracts. If insolvencies reach a certain level, one would expect some protection sellers to default on their obligations.
So the general health of this market crucially depends on the rate of insolvencies. This in turn depends on the economy. The US and Europe are the two largest CDS markets in the world. It is now widely recognised, including by the Federal Reserve, that the US economy is heading for a sharp downturn, possibly a recession. The eurozone, too, is heading for a downturn, but possibly not quite as sharp.
According to the National Bureau of Economic Research, the average length of US recession, excluding the 2001 recession, was 11 months. The 2001 recession was shorter, which brings down the average to about 10 months. The US has been quite lucky. Germany, for example, suffered a downturn at the beginning of this decade. It lasted a near eternity – 15 quarters – and included two separate technical recessions. Interestingly, and perhaps most relevant to today’s debate, is the fact that this downturn was also aggravated by a national credit squeeze. German banks cleaned up their balance sheets after a decade of binge-lending.
The German experience has taught us that persistent problems in financial transmission channels cause long economic downturns. Today, the really important question is not whether the US can avoid a sharp downturn. It probably cannot. Far more important is the question of how long such a downturn or recession will last. An optimistic scenario would be a short and shallow downturn. A second-best scenario would be for a sharp, but still short, recession.
A truly awful scenario would be a long recession. The US did experience some longish recessions in the past, for example from November 1973 until March 1975, but there was no CDS market around at the time.
So what then would be the effects of these scenarios on the CDS market? Bill Gross of Pimco*, who runs the world’s largest bond fund, last week produced an interesting back-of-the-envelope calculation that received widespread publicity. He projected that the losses from credit default swaps caused by a rise in bankruptcies could be $250bn or more – which would be similar to the expected total loss as a result of subprime.
This is how he arrived at this estimate. His calculation assumes that the corporate insolvency rate would return to a normal level of 1.25 per cent (measured as the default rate of all investment grade and junk debt outstanding). As the entire CDS market is worth about $45,000bn, $500bn in CDS insurance would be triggered under this assumption. The protection sellers would probably be able to recover some of this, so the net loss would come to about half of that. This estimate is very rough, of course. Most important, it is based on the assumption that the hypothetical US recession would not turn into a prolonged slump. In that case, one would expect corporate default rates not merely to return to trend, but to overshoot in the other direction.
So one could take that calculation as a starting point. A downturn lasting two years could easily trigger payments streams of a multiple of $250bn.
At this point we might be tempted to conclude that this all is irrelevant, since this is only insurance, which is a zero-sum financial game. The money is still there, only somebody else has got it. But in the light of the current liquidity conditions in financial markets, that would be a complacent view to take.
If protection sellers were to default en masse, so too could some protection buyers who erroneously assume that they are protected. Given that the CDS market is largely unregulated there is no guarantee of sufficient liquidity behind each contract.
It is not difficult at all to see how the CDS market has the potential to cause serious financial contagion. The subprime crisis came fairly close to destabilising the global financial system. A CDS crisis, under a pessimistic scenario, could produce a global financial meltdown.
This is not a prediction of what will happen, merely a contingent scenario. But it is contingent on an event – a nasty and long recession – that is not entirely improbable.
11 comments
Comments are closed.
isn’t the back of the envelope calculation a bit optimistic ? after all, 1.25 applies to all outstanding bond, including blue chips, while CDS outstanding are probably skewed toward the riskiest parts of the bond market (it is likely there are more CDS on, say, GM than on GE)
Credit default swaps can be hedged by shorting the stock, so I think it’s wrong to suppose that only the recovery value will be recuperated; selling the stock when the stock goes to 0 will also bring some good ol’ p&l.
I made a mash of that. Let’s put it this way.
Suppose you write a CDS that company W will go bankrupt, i.e. you pay 100 if there is a default event.
And suppose there is now a bond of Company X which trades 90. You calculate that, if X defaults, the bond will fall to 40. Then a very simplistic calculation shows that you can hedge the default risk on the CDS by shorting two bonds. Or the stock is worth 10. If the company defaults, the stock will fall to 0, so you can sell short 10 of the stock to hedge the risk of default.
So the point is: CDSs are not a closed system. They can be hedged, with bonds or with stocks. Simply treating them as naked positions will result in incorrect calcuations as to how much people will lose, because they have (or they should have) offsetting positions which will earn them money should they have to pay out on their CDSs.
The risk in this kind of thing is the models, and unexpected events which cause jumps (eg Bank of America buying Countrywide). Every meltdown, being different than the last, shows yet another problem with the models, and losses (and sometimes gains) result.
a,
In theory that is true, but in practice, the volume of CDS is so far in excess of the underlying that most of the positions are unhedged or are being “hedged” with other CDS.
In the Adelphi bankruptcy, the amount of CDS written was ten times the amount of cash bonds. And remember, you can short stock only to the extent you can borrow it. There is not way that the market value of Adelphi was ever 10x the amount of its bonds, and even if that had even been true, no way that 100% of it could be used to short.
In fact, at the time of the Adelphi bankruptcy, the rules in the contracts were that you had to submit collateral (the bonds) in order to collect on your CDS. The rules were waived to prevent the collapse of the market. I assume subsequent contracts call only for cash settlement.
Since CDS are used for many reasons, including creating synthetic debt exposures (in fact many investors reportedly prefer to use CDS to buying bonds), the amount greatly exceeds the amount of underlying debt (I remember reading the 10X number is generally operative, but I can’t recall where I saw that).
And per Gillaume’s comment (which I believe is correct) if you have more CDS in lower credit quality stock, there will be even less stock (in the sense market value is lower) available for shorting.
“At this point we might be tempted to conclude that this all is irrelevant, since this is only insurance, which is a zero-sum financial game. The money is still there, only somebody else has got it. But in the light of the current liquidity conditions in financial markets, that would be a complacent view to take.
If protection sellers were to default en masse, so too could some protection buyers who erroneously assume that they are protected. Given that the CDS market is largely unregulated there is no guarantee of sufficient liquidity behind each contract.”
In this sense, Bill Gross’ estimate is a ‘gross’ estimate. Is there any way of scoping the potential net damage, taking into account the ‘0 sum game’ characteristic?
By halving the final number after assuming a return to trend, Gross is effectively attempting to give a “net” estimate.
Yves:
Regarding the Delphi bankruptcy, there was an auction to determine a final price for the bonds and that was a physical settlement auction.Any seller of protection retained his right to demand physical delivery before paying so his counterparty needed to buy bonds at the auction.The rules haven’t changed, CDS work mostly on physical settlement and in case of imbalance, there will a physical settlement auction whose price will be used for those who want cash settlement.In practice this means that physical settlement applies on a net position basis instead of a gross position basis and has the additional benefit of preventing “squeezes.”
You are right that the gross amount of CDS oftentimes exceeds the amount of cash bonds (12 times for Delphi), but the net amount is much more balanced.
“And per Gillaume’s comment (which I believe is correct) if you have more CDS in lower credit quality stock, there will be even less stock (in the sense market value is lower) available for shorting.”
I’m not in charge of any such book, but c’mon there have to be risk limits on this kind of thing. Really, if there aren’t, the entire risk department at the bank has to be fired.
Gross’ halving is net of recoveries; not net of inverse counterparty effects.
a, read Bookstaber’s blog sometimes. Nothing that goes on would surprise me. Traders most places can bully risk managers as long as they appear to be making money.
It’s quite odd that Bill Gross doesn’t seem to understand the difference between liquidity reserves and capital reserves. That doesn’t add a whole lot of credibility to the rest of his analysis.