“Unfortunately, in a space where traders became so dependent on their models, the chain was circular …”

This would be Soros’ reflexivity, no?

To the first, briefly, I think that modelling default probabilities exogenously is far more appealing from a conceptual standpoint – it happens to be one of the features I like in the SFGC model – but we must realize that without the feedback mechanisms that would be otherwise captured by an endogenous model, we will systemically underestimate risk. For this reason I’m partial to models that incorporate a jump component; seems like a relatively simple way of capturing the chaotic distribution exogenously (assuming your joint distribution is well specified, of course!). At the risk of contradicting myself, the Merton model is a wonderful tool for illustrative purposes, and a deceptively cruel thing to implement. It’s too limited, in much the same way the SFGC is. In particular, leverage kills it… so like so many models, it fails right when you need it to work.

And to the second, I am actually of the rare opinion that the CDS markets remained liquid – and displayed the voice of the broader market – while the cash markets fell apart. In my mind, this led to the rise of the negative basis trade. As people fled the markets, the potentially unlimited OTC contracts were the ones that could be generated as needed; finding buyers for cash products that were perceived as doomed was another story, and bargain prices resulted. I firmly believe that if banks like AIG had been forced to post collateral, much of the systemic impact would have been avoided. Lehman may still have gone – even AIG may have gone down – but they would not have so grossly levered themselves as to take down everyone standing nearby. We all know that a catastrophe will destroy an insurance company; our goal should not be to save these companies per se, but to make sure that whatever insurance they do sell is done so with proper risk management in place. Nonetheless, because CDS insurance differs from real insurance in that one security may be insured multiple times, I am strongly in favor of a clearinghouse solution supported by a diverse group of banks. I do not see a need for a government backstop, I think it can be handled privately and the diversity of the backing group would provide assurance that the counterparty will always stand.

My apologies for the length of these responses… I’m too used to the blog, I guess.

Two additional questions:

- What’s your take on endogenous/exogenous PDs? Statistical measures, or Merton model?

- Do you think there was a bank run on the CDS market? If so (or even not), do you think there should be an issuer of last resort, someone with the credibility to backstop the systematic CDS market, like the Fed?