The underlying individual assets didn't really turn out to be poorly rated. The problem was they assumed a low-level of correlation between the risk of individ

The underlying individual assets didn't really turn out to be poorly rated. The problem was they assumed a low-level of correlation between the risk of individual assets. Basically, they believed default risk for a house in Milwaukee essentially an independent variable compared to default risk in South Carolina. They made that assumption because there had never been a nationwide (or global really) housing crash. That assumption actually drove nationwide demand for mortgages, which lowered rates, thereby creating the correlation they assumed didn't exist.