Freddie and Fannie were extremely thinly capitalized and they had a bunch of mortgage-backed securites on the books. Existing regulations forced them to mark to market. When the MBS market started tanking, Fannie/Freddie book value took a hit.
It was more than just that. They also were carrying higher LTVs than 80%. In particular Fannie. If they had stayed with the traditional 80% LTV, both of them would be in better shape.
In your point of view, CRA has been a time-delay fuse that was lit in 1977 and finally brought down Fannie and Freddie 31 years later, in 2008. Got it.
In a way (combined with later revisions to it as well).. but also that CRA achieved the opposite of what it was intended. The effect of the change in realized demand was ignored in housing prices. If you increase LTVs for low income, you also implicitly increase LTVs for investors through other channels. I would like to see all loans go to 80 LTV for Freddie and Fannie, but I just don’t see it happening.
My argument is, 20% down is not equally hard for everyone, it’s much harder for a low-income household than for a high-end household. It should be okay to give 3% down loans to low-income homebuyers.
I disagree here. If you consider both higher income and lower income buying the same house, it is true. If you consider buying houses at ratio to gross income, it is actually harder for the higher income person because taxes take a greater proportion of income… all other things proportionately.
If you remember, there was a neck(bend) in the ratio between quality of house and amount paid in 2007. 500k might get you a so-so house or condo. If you get up to 800k, you get a decent house. The improvement in sq footage was more than the ratio between the prices. Not only that, but the quality was much better. This had to do with the availability of easy financing and high LTV loans. They were more available under 500k because they could get handed over to Freddie and Fannie. You needed a high FICO for high LTVs over 500k and/or would pay high interest rates.
