I’m not sure why everything has to be so complicated. To me, the solution seems quite simple:

1. Fund a new government entity with $500bln to $750bln. Let them draw it down with $100-$200bln increments if it makes it easier.
2. Define a what they are allowed to buy: financial instruments (I think these would be CDOs, but there are so many different things floating around I’m not really sure) that are back by real mortgages. The organization would not be authorized to buy derivatives (since these are basically just a form of gambling in many cases.) Just to be clear: the purchased securities must be backed by real assets (ie: mortgages).
3. Set the price: This is the hardest part in theory, since the fact that there is no market for these things that is the root of the problem. I think paying full price (as has sometimes been discussed) is stupid. I see three options:
   • Buy them at the last mark-to-market value of the selling firm. In this model, the seller would be required to sell a portfolio of securities that seemed, on the whole, to be reasonably valued.
   • If valuation is in question and probably worth less, pay book value but require additional warrants in the seller. This gives the government (taxpayer) some upside in case the assets are worth less.
   • If the valuation is in question and probably worth more, pay book value + 20% and take some warrants. This gives the seller some more case and still gives the government some upside.

   The big trick here would be setting the warrant coverage, but I’m pretty sure it wouldn’t be that hard.

4. As a bonus, it would be ideal if the agency doing this could somehow either acquire, unwind, or force, these purchases to re-consolidate loans. This would mean buying up all the tiers of a security across which a number of loans have been subdivided. In this scenario, the agency would effectively roll the securities back up and, in the process, unwind them back to the original loans. Holding a giant portfolio of loans would basically make this new agency fannie/freddie 2.0. No private entities are capable of doing this because of the sheer scale, but the government could – either with money or with legislation or both.

Why would this work?

If these securities are backed by actual mortgages they have some value. It’s unclear what it is, but it is based on how many of the mortgages in the portfolio default. So if the government is able to buy these securities at $0.20-$0.60 on the dollar – which would appear likely since everyone is desperate and there have been open market transactions in this range – it would be a good deal in the long, long term. The problem is banks have to mark to market (ie: reduce the value) of these assets which hurts their capital ratio and forces them to get more money, thus creating a vicious cycle. The government has no such short-term issues (in theory). So if the government can wait 10-30 years and basically hold the mortgages for their effective lives, it should recoup $1 on the dollar (plus interest) minus whatever the foreclosure rate is. This is interesting because it seems to me that a foreclosure rate of > 40% seems really unlikey and foreclosures don’t result in $0 return. So even at a 40% foreclosure rate where foreclosures recouped 25% of the loan value, that would be a 30% off. So $0.70 on the dollar. (And that’s not including interest on the loans that get paid.) That’s not a bad deal if you can be patient.

And only the government has this kind of patience.