As recently as last summer, the central bank’s entire vault of reserves — about $800 billion at the time — was in Treasury securities. By last week, the Fed’s holdings of unencumbered Treasuries had dwindled to just over $300 billion. Much of the rest of its assets were in the form of loans to banks and investment banks, which had pledged riskier securities as collateral.
A few years from now, there will be a magazine cover with someone we've never heard of who bought all of those mortgages and derivatives for next to nothing on the correct assumption that they were indeed worth quite a bit
The person who will be on that cover is someone we've all heard of, Ben Bernanke. As Andrews points out, Bernanke is selling treasuries to buy the assets that Karabell thinks are undervalued. Meanwhile, the market is falling all over itself to go in the opposite direction.
The best case scenario is that the U.S. government hedge fund makes a big profit for the taxpayers over the next few years. The worst case scenario is that house prices go into free fall and the taxpayers take a loss. But in that case, you could argue that the Fed's actions were at least countercyclical.
The accounting issue is a tough one. When the market prices are obviously right, then firms that refuse to mark to market so they can keep adding to their risk are a menace. The S&L crisis is the notorious example.
When market prices are obviously wrong, then mark-to-market is a bad thing. Karabell makes that point eloquently.
I think that market prices are more likely to be right than wrong, and it is particularly difficult to identify when they are obviously wrong. Even now, when I think the odds favor the Bernanke hedge fund making a profit, that's just my opinion. Others think that he is going to take a bath, perhaps an awful one. Ken Rogoff, in the piece I linked to yesterday, seems to be in that camp. So, I still think that, in an imperfect world, mark-to-market is the best choice.