Over on his blog, Felix Salmon attempts to shoot down the defenders of securitization. He cites the academic debate (also recently commented on by Megan) about whether securitization necessary causes loan quality to deteriorate. Since the academics seem to be in disagreement, Salmon wants to appeal to common sense instead. He says:

At heart, it all comes down to information: loans are stronger and more desirable than bonds, because a bank intends to hold its loan to maturity and does a lot of underwriting, shoring it up with covenants. Bonds, by contrast, are often held only briefly, and are often bought by investors who do precious little fundamental analysis; what's more, they simply don't have the kind of granular information that bank lenders have. And securitizations are even worse than bonds -- no one really knows what's in them, and they're ultimately based more on models than on shoe-leather underwriting. So it's entirely predictable that the boom in mortgage securitization was bad for the overall quality of the debt. And attempts to show otherwise are ultimately doomed.



The italics on "really" are his. I think wonks' attacks on securitization are resulting in a sort of throwing the baby out with the bathwater situation. Securitization certainly contributed to the mortgage market overheating. Excessive exuberance caused irrational expectations leading to lower debt quality. But that final result wasn't caused by some sort of unavoidable feature of securitization.

Has anyone taken note what's happened in the auto loan securitization space? How about with credit card securitizations? What about the dozens of other assets classes that don't rely on mortgages? Why hasn't underwriting suffered for those products? Real estate is the symptom here, not securitization.

Salmon's claim that "no one really knows" what's in a securitization is common perception. It's also completely false. Originators, investment banks and rating agencies know exactly what's in them. There are computer database files with all loan-level information including, but not limited to: loan attributes (term, interest rate, principal, loan-to-value), borrower characteristics (geographic location, FICO score, debt-to-income) and other important statistics that that these parties use to model the deals. The data is generally verified through third-party due diligence completed by accounting firms.

There are, of course, problems. They center on how process works currently. The first is that investors generally don't have this level of detail to work with. As a result they must rely on rating agencies to ensure that the data results in a given rating. I have cited this as a problem ad nauseum. What makes this situation more ridiculous is that technology would make providing this data to anyone who wants it pretty easy. And the models aren't so complex that investors couldn't make their own. Given ratings agencies' loss of credibility, all investors should want to do their own in-depth analysis in the future. This would solve the problem that Salmon appears to worry about concerning investors not having enough information. If they don't have the chops or desire to do the analysis, then they shouldn't invest in securitization, and the problem of too much issuance would solve itself.

So I agree with Salmon that the lack of information was a problem. That's why they bought junk, thinking it was gold. It was investors' lack of prudence that allowed excessive demand and caused the market to overheat. And whenever a market overheats, quality deteriorates. Had investors exercised sharper reason, fewer would have invested and those who did would have required better collateral. More widely disseminated information would have resulted in the market correcting the problem. But since rating agencies failed to blow the whistle on poor underwriting, and kept giving bad deals their stamp of approval, the securitization market failed. But it was a problem of circumstance, not function.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.