The Circularity of Mark to Market Accounting

The present crisis has seen a lot of debate about mark-to-market accounting. Traders tend to favor mark-to-market, as do academics who mistrust banks' motives and accounting. Bankers and efficient market skeptics argue for more flexibility. The usual problem, of course, is that in a crisis prices can plummet for liquidity reasons, which can cause institutions to become insolvent on a mark-to-market basis, causing bankruptcies, liquidations, further price declines, and more insolvencies. Very few people are comfortable actually allowing this dynamic to play out, but it is a struggle to find a credible alternative accounting regime. For large institutions, like the banks that hold troubled mortgages, there's an even more fundamental problem with mark-t0-market accounting: it is circular and therefore not well-defined.

Mark-to-market accounting says that assets should be valued at their market prices. The market price for an asset depends on the large institution's demand for the asset -- that is well-known. But the large institution's demand depends on the market price, and if the institution is levered, its demand may in fact depend positively on the price of the asset.

So the question is: when the bank is valuing its assets, should it consider the market price including its own demand, or not including its own demand?

The problem is even more acute when there are speculators driving down the price of an asset solely because they anticipate that they will force a large, levered investor to liquidate. In other words, the market price may only be low because investors believe liquidations are coming, and those liquidations are only forced by the investors' belief. It becomes an extremely destructive self-fulfilling prophecy.

What's the right solution here?