Ed Glaeser has a nice piece about the debate over whether securitization should get the blame for the subprime mess. But it doesn’t address one of the problems created by securitization: hidden leverage.
When banks (commercial and investment) sold off mortgage backed securities, they got them off their balance sheets, and so there was a pretense that they were no longer liabilities. But in order to sell the MBS, the lenders had to offer repurchase agreements, which said that if there was something materially wrong with the loan underwriting, the investor could return the mortgage backed security to the lender at par. Lenders also often kept residual positions of mortgage backed securities, meaning that to reassure investors, the lenders (i.e., the sellers of the securities) would take first loss positions.
Both repurchase agreements and residuals effectively increased the leverage taken on by lenders. Let me illustrate: suppose a lender has an whole asset and capital of ten percent, and the asset loses one percent of its value. The lender takes a ten percent hit against capital, because it is levered at 10 to one. But now suppose it takes a first loss position of ten percent on residuals, and the mortgage underlying the residuals lose one percent of value. The residual loses ten percent of its value, which means it wipes out the capital that is implicitly backing it. The combination of ten percent capital and a ten percent first loss position implies actual leverage of [updated: 100 to 1].
Ironically, the fact that financial institutions ate some of their own cooking–something that should have mitigated moral hazard–made them more vulnerable.