Monday, 1 March 2010

Indecent exposure: putting clothes on a CDS

There's been a movement - reminiscent of the anti-short-selling campaign - against buying "naked credit default swaps". The idea of a CDS is to provide insurance against default on a corporate or national bond; and a naked CDS is one which is bought by someone who doesn't own the original bond.

The objection to this is based on a kind of moral hazard argument - the idea that if I buy insurance against your house burning down, I'll have an incentive to leave piles of leaves and open cans of petrol near it.

So one proposal, according to this article in the FT, is to prohibit people from buying a CDS if they don't own the underlying bond. (Update: Felix Salmon points out Wolfgang Munchau's more in-depth article proposing the same thing)

Sounds like it makes sense at first glance. But one big question arises: what if I want to sell the bond afterwards?

Do you prohibit me from selling it without also selling the CDS? If so, I am suddenly left with a much less liquid instrument which is harder to sell - because I need to find a new investor who wants both the bond and the CDS. I will probably end up selling the pair at a discount, reducing the ability of the market to find accurate asset prices. Or more likely, I will end up not selling them at all, which similarly reduces liquidity, shrinks the market and increases volatility.

Or do you let the CDS expire as soon as the bond is sold? This is equivalent to an even bigger (forced) discount on selling, meaning that again I am highly unlikely to sell the bond. Or indeed to buy the CDS in the first place.

A variant on this is to allow (as we do with motor insurance) the bond owner to redeem the unused period of the CDS when they sell the bond. However the difference with motor insurance is that the policyholder's risk is unlikely to have changed between signing up for a policy and then selling their car. With CDS, there would be a huge incentive for people to buy CDS on a risky country, wait for the risk to decline, sell the bond at a profit and then cash in the CDS at the expense of the issuer. If the risk doesn't decline, they just keep the bond and CDS at par value - it becomes a free put option.

You could just say that people shouldn't sell bonds if they are going to take out a CDS on them. But remember that the whole point of bonds is that they can be sold: if I were planning to hold a bond forever, there are other ways to invest money and get a better return. The reason governments and companies can raise money relatively cheaply is because they can sell bonds which investors know they can sell again when needed. Thus, we need to allow bondholders to sell their bonds on.

Or finally, do you allow the bond to be sold as normal, and let the seller still keep their CDS? This looks at first to be the most natural and fair solution, and fits most naturally with the standard economic models of property rights and pricing. But it opens up an obvious arbitrage opportunity: anyone who wants a naked CDS just buys a bond, buys the associated CDS and immediately sells the bond again. So this solution would probably be forbidden by the new regulation.

So the proposal, while eliminating a form of moral hazard, would probably also eliminate the whole CDS market. Perhaps that would be a good thing (though I'm with Felix on this - it's probably not) and perhaps that is what people want. But it ought to be done overtly and not by pretending to make "sensible" moral hazard regulations.


Min said...

To me the more important aspect of CDSs is not the moral hazard, but the explosion of risk. It is as if a house worth $500,000 burns down, and insurance is paid to the tune of $50,000,000! Or more. And mostly to bystanders who have made bets, not really taken out insurance.

After the bursting of the Tulip Bubble, the courts refused to honor tulip futures contracts, as being gambling debts. While that is not the only consideration, there is a certain wisdom there, isn't there?

Anonymous said...

I'm just not sure about the magnitude of this problem for bonds that are very liquid, like corporates.