When I first read the Buttonwood column on Collateral Debt Obligations (“Of Scorpions and Starfighters”, Economist, January 31, 2006), I disagreed strongly with it but put it aside without much further thought. But then Anuradha suggested that I should blog about it; so here I go.
Buttonwood paints a picture of CDOs as being dark and mysterious things and so I began to wonder what is it about CDOs that creates unease in the minds of many. A CDO is a fairly straightforward and legitimate instrument. After all, a commercial bank is at bottom nothing but a CDO — though doubtless it is a rather crude and old fashioned way of creating a CDO.
I therefore went through the Buttonwood column replacing CDOs by banks and bank loans. A large part of the column goes through quite nicely. This is a sample of a few paragraphs:
If most of the borrowers stay solvent, the bank makes good money. If more than a handful default, then depositors and investors begin to take a hit ... The precise mixture of risks and payouts depends on how the bank is managed.
Moreover, the value of a bank loan portfolio depends not just on expected rates of default, but also on what might be recovered from defaulting companies’ assets. ... Bank loan portfolios are dynamic: they are in the hands of managers who can weed out the exposure to companies before they default, or trade credit risk with the aim of improving the portfolio.... Banks slice themselves into tranches of differing risk — deposits, (subordinated) debt and equity. Thus in theory investors can pick the collection of risks that suits them. They are helped by the existence of credit ratings, at least for the safer tranches (the riskiest equity tranches, which bear the first loss in the event of default, usually have no rating). But they must also consider the likely market price of the tranche they invest in, both for accounting reasons and in case they want to sell before maturity.
Bank loans are not that actively traded .... So it is often near impossible to establish a market price for them. Accountants have a horrible time when auditing books of illiquid bank loans, being forced to use numbers that they know are nearly meaningless.
One can go on with much of the rest of the column. For example, Buttonwood decries CDOs of CDOs, but in reality a CDO squared is not very different from a bank lending to another bank or investing in the subordinated debt of another bank. But let me not belabour the point.
Buttonwood also seems to think that cash settlement of credit derivatives is a bad thing that somehow disconnects them from reality. This is not true at all — the only difference between cash and physical settlement is one of transaction costs. Buttonwood is also worried about the notional vlaue of credit derivatives exceeding the total amount of debt that the company has issued. Again this is quite common in derivative markets. People are quite willing to take a little basis risk to operate in a more liquid market and therefore the largest derivative contracts usually attract a volume and open interest that is much larger than the direct risk exposure to the underlying of this contract. It is natural for people to use Delphi related CDSs or CDOs to hedge exposures to the entire auto component industry and also to cross hedge some General Motors or even auto industry risk. It is not at all surprising that the notional far exceeds the outstanding debt of Delphi
Let me end with a provocative question. Having invented banks first, humanity found it necessary to invent CDOs because they are far more efficient and transparent ways of bundling and trading credit risk. Had we invented CDOs first, would we have ever found it necessary to invent banks?