There are already a lot of replies, so maybe it is pointless adding my own . . . My understanding of the problem with mortgage backed securities was that the calculations assumed the risks on the low-quality debt was uncorrelated. If that were true, all of the math works out, and there is no problem with mortgage backed securities. The reality was/is, however, that the risks on those loans were, in fact, highly correlated. The conditions that caused one loan to default also made it much more likely that others mixed up in the pot would also default. In that case, the math was wrong and the magic failed. I suppose we could argue all day about whether that was an honest mistake or deliberate manipulation.