This post lacks a nifty news peg, but I've been thinking a lot about Joshua Coval, Jakub Kurek, and Erik Stafford's paper "The Economics of Structure Finance." In particular, people talk about the "systemic risk" of big firms and massive banks. The banks, in other words, that are too big to fail, and whose very existence thus poses a threat to the system. That's where you get the talk about capping bank size. But the argument of the paper is that the turn from bonds and loans to structured finance was itself a contributor to systemic risk. And I've not heard as much discussion about that.

Coval, Kurek, and Stafford is clunky to write, so henceforth, I'll just call them CKS. Either way, they argue that risk needs to be understood not only in terms of how often an asset fails but when it fails. Namely, is it likely to fail amidst a chain of other defaults and a broad economic collapse? Or is failure basically uncorrelated to other aspects of the economy? That matters for an individual portfolio, of course, because you don't want your investments to drop all at once. But it matters even more, as we're finding out, for the system as a whole, which can't handle everyone's portfolios dropping all at once.

According to CKS, the key here is in the shift from traditional bonds to structured finance. Traditional corporate bonds, they argue, are "largely driven by firm-specific considerations." Microsoft may be somewhat likelier to go out of business if GM is collapsing, but if I told you GM was collapsing, your first move wouldn't be to check on Microsoft. Structured finance, however, plays differently: Banks were mainly interested in holding AAA tranches because that lowered their capital requirements. Structured finance was a way of building a lot of AAA tranches. And so, by 2007, the plurality of Moody's business was rating structure finance. The problem is, unlike a traditional bond, damage to safest tranche of a CDO probably happens at the same time as damage to the safest tranche of just about every CDO. An ill tide sinks all boats.

"Credit ratings," write CKS, "are silent regarding the state of the world in which default is likely to happen." An asset that has a one percent chance of default gets the same rating whether the default is likely to happen in a recession or simply randomly. That's because the bond rater is pricing risk to individual firms, not risk to the system. And since portfolios biased towards structured financial products decreased the risk of individual firms but increased the risk of the system as a whole, risk was underpriced in structured finance. Which meant there could be more structured finance products sold with a AAA rating. And even in the downturn, that worked out: The suffering has been systemic more than it's been firm-specific.

I've not seen so much on how to fix this. Maybe ratings agencies will develop a standard formula for assessing how likely a firm's portfolio is to go bad all at once. Maybe the government will regulated how much correlated risk a bank can hold at one time. But it does seem to be part of the puzzle.