There are problems with using historical cost accounting, and the move to accounting based on market values seems at first glance to be more accurate. Market values change every day, and so they are not as prone to a particular tendentious bias, especially, the hope that bad asset will rebound in price even though everyone knows it is severely impaired.
Yet, as every critic of this crisis notes, the market is hardly perfect. The same market priced these same mortgages as if they had basically zero credit risk in 2006, is predicting massive future default rates. If one thinks that the marking to market gives the best assessments, then one can not really say that regulators, or anyone, failed pre 2007. The market suggested no risk, as implied by the market value, or spreads, of these instruments.
Currently, many closed-end funds with market-traded instruments trade at a substantial 20% discount. That is, you can trade there constituents, and they are worth 20% more than the entity that holds them. Above we see Nuveen's convertible bond portfolio (JPC), and this trades at a 20% discount to the value of its portfolio. Is it wise to therefore value these assets at their net asset value, or the value implied by the closed-end fund? Further, this is not specific to bond closed end funds, as closed end funds with stocks have similar discounts (see Adams Express, ADX). Market values, clearly, are not merely unbiased present valuations, because this discount varies over time. If the market has a 20% risk premium assigned to mortgage related products, is it optimal to impute this into book asset valuations?
The bottom line is that if one argues that the market should set book asset values, and thus risk parameters for a solvency exercise, this goes both ways. No one thinks that the market correctly priced mortgage risk in 2006, however, so what does one do?