1. Mark to market. This is too severe, because the adverse selection problem means that the price of, e.g. formerly AA subprime securities put on the market is lower than the value of ALL AA subprimes including those retained by the original owners.

2. Par value subtracting expected failure to repay by December 2010. This is what the stress tests did. As the op-ed says, this is a stupid method, because many loans are due later and we know now that they won’t all be repaid. The excuse that the recession will be over by 2011 is pretence. The loans were bad before the recession started, and they’ll bad after it ends. They are bad because of the housing bubble, not because of the recession. So this tells us we can’t trust the regulators– they are trying to cover up the problem.

3. Estimate how many of the loans will be repaid, not exempting any of them. This is the obvious method, the method of conventional accounting. It is what banks ought to be doing themselves, except that the difficulty of making such estimates gives them enough wiggle room to avoid their duty.

What ought to be made clear is that (3) is not the same as mark-to-market. It is not as big a change from par, but it might well be big enough to show immediate insolvency, not just undercapitalization.