Why not subscribe?

Friday, November 09, 2007

Risk in finance: a few questions [following Tyler Cowen]

Tyler Cowen, at Marginal Revolution, writes on the surprising nature of current risk taking in financial markets:

"We are finding more and more ways to (implicitly) write naked puts in highly leveraged forms. Yes this has brought us new products but it all seems to be new mortgage products. Could those products possibly justify the financial carnage we have seen? That is the critical question but I suspect the answer is "no," that in this sphere we stepped beyond the bound of optimal risk-taking."

On risk: Tyler's clearly on the right track in terms of looking at various incentives. There seems to be considerable widespread carnage from the US mortgage boom of the past few years, and this carnage extends to a variety of surprising areas. Why, for example, are banks in the UK at serious risk from their investments in US mortgages?

Some questions that might be asked:

1. Do financial managers really have good models of risk? Or do they just think we do? (insert obligatory "Black Swan" reference here)

2. Is there some fundamentally bad assumption that was made here? For example, if the probability of failure of loan A is 5%, and loan B is 10%, the joint failure is .5% -- IF they are independent. But CDO's made up of mortgages aren't independent (surely I don't have to explain why here). Did they underestimate the covariance? That's what I mean by a fundamentally bad assumption.

3. Are these financial packages really offloading risk / spreading risk around / diversifying, or are they creating a large Rube Goldberg structure in which we aren't lowering risk by diversifying it, but creating so many connections that a "margin call" creates an overall risk of crashing?

4. Citi seems to be in trouble in commercial paper for lending out long term but borrowing short term. Isn't that a pretty basic error? Isn't that fundamentally what happened to many S&L's only a couple of decades ago?

5. Why do we allow companies to keep so much "off-book" that clearly creates obligations (Enron, Citi)? Is a large part of the problem one of transparency?

6. Do the rating agencies (Moody's, etc) really understand risk? Are there good statistics about how good they are?