Subprimes and The Efficient Market Theory

Many active investors who don't believe that markets are fully efficient (if they did, of course, they wouldn't be active investors) nevertheless believe that markets are mostly or frequently efficient (see, for example, Warren Buffett making this point in his 1988 Berkshire Hathaway Chairman's Letter). How then to explain the yawning gap between the market prices of certain mortgage-backed securities and their supposed hold-to-maturity or intrinsic values?

Brian Wesbury, chief economist of First Trust, reiterated on CNBC today a point he made in a recent Forbes column: if 100% of the mortgages in a subprime mortgage CDO defaulted, owners of the security would recover something -- perhaps 40 cents on the dollar -- from the sale of the houses. And yet, two months ago, Merrill Lynch unloaded some CDOs for about 22 cents on the dollar (Barry Ritholtz argued at the time that, since Merrill was financing about 75% of the sale itself, the actual sale price for the assets was about 5.47 cents on the dollar).

If what Wesbury says is right (and it seems reasonable), why aren't institutional investors lining up to bid on subprime-backed paper for 22 cents on the dollar?