This paper investigates the extent to which securitization trusts, as significant purchasers of subprime mortgages, increased the overall quantity of mortgages being originated or simply displaced the roles of existing purchasers/holders of mortgages. In order to investigate this, we exploit a natural experiment from 2004 in which the S&P rating agency changed its rating methodology for securities issued by securitization trusts. This rating change effectively increased the cost of securitizing loans made in certain states but not others. The states affected by this rating change were those whose laws created the possibility of securitization trusts being held liable for actions committed by loan originators. The success of our empirical strategy depends crucially on our analysis of a complex web of overlapping state and federal lending regulations, allowing us to pinpoint precisely which loans were and were not affected by the rating change. Through the use of panel regressions and a difference in differences methodology, we find that the S&P rating change did lead to a significant reduction in the quantity of certain types of subprime loans being securitized within the affected states. We find a smaller effect on the total amount of subprime lending in these states, indicating that some but not all of the subprime mortgages that would have been originated and securitized were instead simply held by other entities. Together, these results suggest that while securitization may have increased the number of subprime mortgages being originated in the lead up to the financial crisis, a significant portion of the demand may have existed even apart from the presence of that particular investment channel. This is joint work with Kay Giesecke.
Back to Workshop IV: Forensic Analysis of Financial Data