As strange as it may seem, banks' shitty assets are their smaller problem. Like the AIG saga is showing, the real problem is the size of their liabilities.

Thanks to the CDS market, the big financial institutions have taken to making very, very, very large bets on supposedly highly improbable events (ie, they's get a fee upfront and would commit to make a big payout if the unlikely event, say the bankruptcy of Lehman Brothers or AIG, happened). CDS were initially created as a risk-mitigation instrument, and they can certainly be used that way (for instance, if a transaction depends on a large payment by Lehman Brothers, it may be useful to buy the additional protection to guarantee the amount of that payment from someone else, typically a highly rated entity like AIG (used to be), should Lehman fail). But they turned out to have two great advantages:

• they were a totally unregulated way to release regulatory capital: since you don't take the risk on Lehman anymore, you can re-use the amount you'd have normally needed to set aside for that exposure (apparently the regulators forgot to note that you took exposure on someone else instead for that amount, even if that someone else was highly rated); by using capital more than once, you can boost your returns;
• they were a totally unregulated way to make bets: you did not need to have an actual need to hedge a position to purchase (or sell) them, which allowed you to bet mutliple times the amounts you could have under normal regulations on a given risk. By taking risks that were considered extremely low (like AAA rated risk) in high enough concentrations, you could make good income for (what was perceived as) no risk - or at least for no cost in equity;

So lots of financial players started taking those huge bets on supposedly unlikely things happening - with commitments to pay huge amounts should these things actually happen.

For those players, the CDSs are not assets, they are potential liabilities. They booked the upfront fee right away, are maybe getting a smallish yearly commission as income, and have this potentially huge payout to make if something bad happens to some company or asset.

Again, to get an idea of what kind of leverage we're talking about, read this article about John Paulson in last week's Economist:

Another motivating factor for Mr Paulson was the alluring asymmetry of shorting credit. The most you can lose is the spread over some benchmark rate. Yet if the bond defaults, the gains can be mouth-watering. He targeted BBB-rated tranches, the lowest in subprime securities. With credit spreads so low because of a liquidity glut, *his possible upside as a buyer of protection using credit-default swaps (CDSs) was as much as hundred times the potential downside*. One $22m trade is said to have netted him $1 billion when Lehman Brothers went bust.

And well, there were three problems:

• One was that these unlikely things were not that unlikely (or, in any case, not as unlikely as suggested by the price used to take the risk). People tend to have trouble allocating proper probabilities to rare events, and bankers seem to be no better (go read Nicholas Taieb's *Black Swan*);
• the second was that the "virtuous circle" effect of bubbles further distorted perceptions (asset prices are rising, people borrow more, they buy more assets whose prices go up; they can borrow more to repay earlier loans by backing that with rising collateral, thus reducing default rates, which encourages banks to lend yet more, etc...). All the securities that were seen as not risky at all in a bubbly environment were maybe riskier than they seemed. How could there be only 12 AAA-rated companies in the world, and 64,000 mortgage-backed securities with that same rating?
• the third was that the very use of their instruments, and their heavy concentration in the hands of a small number of players actually created new risks that did not exist before. AIG could have survived underwriting a few billion of CDSs, but not 500 billion-worth of the stuff.

And of course it now turns out that, as should have been obvious (and was to people like John Paulson) these CDS were, in more than a few cases, very, very bad bets. And the number of cases is growing rapidly as the crisis gets worse and hits more companies and reduces more assets' values. Which means that those that underwrote these CDSs are faced with large - and mounting - bills.

Thje size of these bills potentially dwarfs the size of the toxic assets they are also saddled with.

And as the risks are increasing (or seen as increasing which may or may not be the same thing), those that underwrote the CDS are seen as increasingly weak, and those that bought the protection, or took naked bets (but how can you tell - this is all unregulated anyway), suddenly worry about their CDS counterparty in addition to worrying about (or hoping for, in the case of naked bets) the underlying insured event to happen. That CDS counterparty being, of course, a (until recently) highly rated financial institution.

CDS typically have mechanisms whereby the CDS underwriter may need to post collateral to guarantee its obligations: this is what brought AIG down: as signs of trouble started to build, it lost its AAA-rating, which triggered obligations to pay collateral to all its counterparties on its portfolio of CDSs. It is those obligations, all coming at the same time on a large pile of CDSs, that bankrupted it and led government to step in to make the necessary payments.

Note that initially, the government did not even use taxpayer money to make actual payments under the CDSs - just to post collateral. Now, as CDSs are triggered, full payments need to be made, thus the successive bailouts.

The justification for these bailouts is that not paying out on these CDSs could make some of the buyers of protection bankrupt themselves, thus triggering more CDS payments, giving birth to further liabilities for the institutions that underwrote the CDSs. In addition, givne that many of the buyers of CDS protection were banks or hedge funds, there is worry of a domino effect.

It is that liability crash which is the cause of the continued lack of trust in financial institutions by the markets themselves: they know that financial bombs are littered all over the landscape.

Buying toxic assets is "nice" for banks, but solves nothing. Bailing out AIG, oddly enough, could be seen at least as a step in the right direction - the problem of course being that if you're going to take care of all potential liabilities, the total bill might be in tens of trillions, rather than _mere_ trillions - with a lot of that money going to the smart hedge funds that bet on things going badly in various markets and for various institutions (cf Pauslon above).

Given all that, we have several routes:

• one that gives a lot of money to banks that do not deserve it to solve their asset problem, but still do not make them creditworthy (the current Geithner plan), which gives stock markets a temporary boost, taxpayers permanent pain, and solves nothing;
• one that does help them get rid of their real problem (huge contingent liabilities on bets that are turning sour), but is vastly more expensive than the mind-numbing numbers we're throwing around already, and gives all the money to hedgies: the AIG route, multiplied ten or hundred-fold;
• one that acknowledges that the issue is liabilities rather than assets, and that focuses on the fact that a lot of these liabilities are wholy unrelated to any economic or financial activity, and are contingent rather than actual - ie nobody loses anything if they are cancelled. If a 100:1 bet you made is cancelled, your actual loss is not 100, it is 1 - something that could be paid back to you.

So far, the second route has been used when an emergency beckoned (AIG et al); the first route has been used massively but the Treasury does not seem tired of it yet, and the third one seems anathema.

Of course, it means taking the shiny toy away from the hands of the hedgie kids.

Why is that a bad thing, again?