Amir Bhinde, writing in Bloomberg, has an interesting post which makes a few points. To begin, he makes the case that the growth of securitization was heavily influenced by state sponsorship:
The securitization revolution that really stifled traditional banking was led by Fannie Mae and Freddie Mac. The government-sponsored agencies paid banks a fee for originating mortgages that conformed to certain criteria, sparing banks the expense of in-depth analysis and losses from bad loans. Fannie and Freddie sold securitized bundles of the mortgages by suggesting they were as safe as Treasury bonds. Regulators then encouraged banks to buy the securities. Capital requirements for the residential mortgages that a bank kept on its books were more than twice those for mortgage-backed securities that had AA or AAA ratings.
This is broadly accurate; the GSEs were the parents of securitization, and (for reasons not wholly unconnected with the Pfandbrief market), RMBS are cheap for regulatory capital purposes.
The Bloomberg article then criticizes securitization for its reliance on ‘a few abstract variables’, and certainly we have seen that there is more to mortgage risk than stated income, LTV, FICO, and DTI. That said, if honestly sourced, these variables often do quite a good job at predicting the risk of a securitization. Often, though, is perhaps not good enough.
Amir’s next suggestions are more controversial:
To fundamentally reform the financial system, we need to end state sponsorship of securitization.
First, the federal government must stop guaranteeing mortgage securities. If lawmakers feel impelled to divert credit to homebuyers, the Small Business Administration’s approach of offering partial guarantees for housing loans would do less harm. Let the private sector securitize the loans if it can.
Second, banks should be required to evaluate the creditworthiness of every individual or business they directly or indirectly lend to, rather than outsourcing credit analysis to ratings companies or relying on reductionist statistical models. Allowing banks to buy securitized assets with only superficial knowledge of the ultimate borrowers is folly.
There are two issues here.
A correctly functioning government guaranteed (or for that matter covered) bond market provides a useful source of safe assets. That safety is worth more to investors than the cost to the state. What you do with the lower tranches – and who pays for them – is a different story*, but at least for the top tranches, securitization is performing a useful service for investors as well as borrowers.
The second problem is how credit is extended. Like it or not, the large banks have more or less killed off the loan officer in favour of statistical models for most retail and SME credit extension, not just for the securitized piece of it. If you want to go back to the prelapsarian world of local credit for local people, then banking is going to get a whole lot more expensive.
*Whether the state should subsidize home ownership or not, and by how much, is essentially a political rather than a financial question.