Sidestepping the editorial : The basic problem with NINJA loans was that the historical payment/default characteristics were very encouraging. When the rating

Sidestepping the editorial : The basic problem with NINJA loans was that the historical payment/default characteristics were very encouraging. When the rating agencies modeled up new asset classes, they naturally gathered as much historical data as possible. For NINJA loans, the historical dataset was pretty small, since banks don't normally lend to such hopeless cases. However, when they did lend, the facts of the particular loan normally went something like : House buyer works for himself as a builder/developer, gets paid in cash, plays golf with the bank manager - and everyone knows that he's good for repayment (even if his tax returns don't show any income). The major problem with this is that the statistical data did not contain these qualitative factors, and showed only that people with No (declared) Income, No (on-the-books) Job and No (visible) Assets paid back their loans consistently. That became codified into a simple fill-in-the-form application which missed the entire point... And loan brokers went to town, with the loans eventually packaged up and dumped into a pool. At it's heart, the root cause of the problem is that rating agencies did a horrible job of understanding the risks that were embedded in the different types of loans. And no-one had an incentive to let them in on the 'trick' (and, perhaps, the rating agencies were willfully blind to what was going on). Since the middle-men were only short-term holders of each mortgage, the risk got transferred from modeling spreadsheet to investors in pools who trusted the ratings agencies within months or weeks. Frankly, the politicians were completely outside the loop, and the regulators trusted that the Rating Agencies were doing a good job. Terrible assumptions.