AIG's (NYSE: AIG ) announcement that it had "material weaknesses" in its accounting for credit default swaps on collateralized debt obligations (CDOs) is like yelling "Fire!" in a crowded theater -- you can be certain there will be mass panic. The stampede for the exits left the stock at a five-year low. Let's take a closer look.
AIG's stock got pounded as the company upped its loss estimates. To simplify the problem, AIG thought its credit losses on CDO swaps would be X, but it turned out to be much, much more than X. If this sounds familiar, it's because virtually every bank, including Citigroup (NYSE: C ) , Bear Stearns (NYSE: BSC ) , Wachovia (NYSE: WB ) , Morgan Stanley (NYSE: MS ) , and Income Investor pick Bank of America (NYSE: BAC ) , has had something on its balance sheet go toxic.
According to an SEC filing by AIG, the net cumulative decline in the valuation of AIG's positions was estimated to be $352 million at the end of last September, $899 million one month later, and a whopping $5.2 billion as of the end of November. Not exactly the trend you'd like to see.
In the end, the culprit is complexity. AIG uses what it calls a Binomial Expansion Technique. From the SEC filing:
In doing so, [AIG] employs a modified Binomial Expansion Technique ("BET") model that currently utilizes, among other data inputs, market prices obtained from independent sources, from which it derives credit spreads for the securities constituting the collateral pools underlying the related CDOs. The modified BET model derives default probabilities and expected losses from market prices, not credit ratings. The initial implementation of the BET model did not adequately quantify, and thus did not give effect to, the benefit of certain structural mitigants, such as triggers that accelerate amortization of the more senior CDO tranches.
Huh? Just to give an idea of the complexity involved, a residential mortgage-backed security (RMBS) is a pool of thousands of home loans. The RMBS is sliced and diced into several classes, where higher classes have "more" security, but also pay higher prices. There are typically a whole host of other complicated rules that govern an RMBS, such as trigger points, and over-collateralization tests.
Now, that in and of itself is pretty complicated. So just for kicks, let's make it even more complex. Why don't we make a pool of loans, and instead of using individual home loans, we'll make it out of the riskier slices of RMBS, which we'll call a collateralized debt obligation (CDO). Then, as usual, we'll slice and dice that into different classes and throw in another set of complicated rules.
But we're not done yet. Instead of just investing in slices of CDOs, why don't we write some insurance against the default of the senior portions of CDOs, and call it a credit default swap? So here you're insuring a slice of a slice of pools of thousands upon thousands of individual home loans, where each layer of securitization has its own rules thrown in.
The truth is, it's pretty hard to value these things. AIG partially marks its positions to market but messed up when it forgot to add in some of the complex rules governing its swap book. Whether that was an honest mistake or a careless error, I've no idea. What I do believe, though, is that perhaps AIG should not even be in this business in the first place.
Just as Stock Advisor recommendation Berkshire Hathaway's (NYSE: BRK-A ) Chairman Warren Buffett refuses to get involved in businesses he doesn't understand (or can't value properly), perhaps others should follow his lead. After all, if AIG's accountants, with their granular data, probably involving hundreds and hundreds of pages of documents and detailed data on each individual position, cannot properly value their own book, then it's virtually impossible for an outside investor, who has 1/100th of the information upon which to make a decision.