In Are US Banks Worthless?, James Kostohryz raises a valid point: Accounting is indeed one-sided, and you have to bring losses forward and postpone gains. And in the case of banks, that means credit losses on your bad assets are valued, but net interest margin on the good assets is not. And, if I understood the article correctly, they could add up to something that has a positive net present value.

At its core, however, this argument presumes net interest margins are stable through time - and that a failing bank can maintain a positive net interest margin into the future - or at least as long as it takes to write off bad loans over time. History says that depositors won't stick around long enough for that theory to work.

And the lack of deposits kills banks, not capital. (And in that regard, unlike the liquidity crises of last summer and fall, the "stress-test banks" -- such as JPMorgan (JPM), Citigroup (C), Wells Fargo (WFC) and Bank of America (BAC) -- don't appear to be experiencing any debt-funding issues right now - even though there are many unanswered questions on capital.)

James' argument does explain why it feels like bank regulators are slow to close down banks: Net interest margin is a great mop for credit losses; the longer regulators can drag things out, the longer the "mop" can be used. And if you look at bank failures, they tend to occur only when the flow of losses causes depositors to flee, and net interest margin to collapse.

The bad news in this crisis is that so many of our largest banks chose to expand their balance sheets by buying mark-to-market assets, not held-to-maturity assets - and the reason for this choice was capital: Mark-to-market assets didn't require reserves.

And now that things have turned against these banks, they're caught with too little capital and not enough time to make it up with net interest margin.