WAPO on AIG III

Robert Waldmann

ended my last post 5 minutes ago wishing for part III of the saga and here it is !
The final act for AIG by Robert O’Harrow Jr. and Brady Dennis

The collapse was, of course, quick when it came. One interesting fact is that AIG Financial Products (AIGFP) stopped writing new CDSs in 2005. Another is that, until then, they had little idea what they were doing.

In fall 2005, Eugene Park was asked to take over Alan Frost’s responsibilities at Financial Products. Frost had done exceedingly well in marketing the credit-default swaps to Wall Street, and was getting a promotion. He would now report to Cassano directly on other strategic projects.

Park …was worried about the subprime component of the CDO market. He had examined the annual report of a company involved in the subprime business. He was stunned, he told his colleagues at the time.

What a fascinating new idea — how about examining the annual report of a firm whose securities one insures. I mean not all of them (what a bore) but just one.

From this passage it seems that AIGFP just assumed that mortgages were what they had been in the past. I mean it’s not their problem. The non bank mortgage lenders were loaning all they could, because they figured they could pass on the risk (well sometimes risk sometimes certainty of default) to people making CDOs who could pass it on to people buying CDOs who could pass it on to AIGFP which doesn’t seem to have wondered what they were insuring.

I give you a hint. If you tell the world that you are glad to bear all of some vaguely defined kind of risk, then it will get riskier.

AIG was a major factor in the market:

Financial Products had $2.7 trillion worth of swap contracts and positions; 50,000 outstanding trades; 2,000 firms involved on the other side of those trades; and 450 employees in six offices around the world.

That’s $ 6 billion in exposure per employee!

Obviously AIGFP had decided that to write CDSs on AAA rated securities was to get money for nothing. They couldn’t possibly examine the securities they were insuring. Notice how Frost’s role was described: “Frost had done exceedingly well in marketing the credit-default swaps to Wall Street.” You can do very well at selling something if you are charging too low a price.

I guess they can blame the credit rating agencies. In fact they do, but not for rating toxic sludge AAA but for rating them AA. The final words “There was no system in place to account for the fact that the company might not be a Triple A forever.”

In fact AIG hasn’t paid out on CDSs, they ran $150 billion short of ready cash, because they had to post collateral: ‘If additional downgrades occurred, either in AIG’s credit rating or in the CDO ratings, Financial Products would have to come up with tens of billions of dollars in collateral it did not have.’

AIG was downgraded from AAA to AA when Eliot “socks on” Spitzer caught him cooking the books.

On March 14, 2005, Greenberg stepped down amid allegations about his involvement in a questionable deal and accounting practices at AIG. The next day, the Fitch Ratings service downgraded AIG’s credit rating to AA. The two other major rating services, Moody’s and Standard & Poor’s, soon followed suit.

I guess that explains why he didn’t go after anyone with a pitchfork.

Back to my usual anti CDS rant after the jump.

I’d say that was crazy. Given AIGs exposure they shouldn’t have been rated AA. I ask for the nth time, what is the point of a CDS ? Why can’t AIG issue AIG bonds and use the proceeds to buy assets rather than insuring them ? I think it is clear. If AIG had issued 3 trillion in debt they wouldn’t be rated AA. CDSs written appear on the balance sheet at market value. This is nonsense. In the long run, the two actions (write a CDS on assets or sell debt and buy those assets) have the same impact on AIG’s book equity, in the short run the only difference is that AIG can be forced to post collateral (which can be seized in full even if it is in chapter 11 and mere bond-holders have to wait and get cents on the dollar) if it writes CDSs.

I think that it is strictly better for bondholders if AIG issues debt and buys assets than if it insures those assets (I am considering the premia paid on CDSs). I think the only reason to do it with CDSs is to trick regulators and ratings agencies about AIGs liabilities.

On a basically different topic, I cut some boring stuff out of one of my posts and put it here.

Now I think it is fairly likely that, if they they had stuck to CDSs on corporate bonds, AIGFP would have done very well just as they would have done issuing AAA bonds of their own and buying AA bonds.

My guess is that, at first, the money was there to be made because of excessively prudent rules and regulations. It is also there, because prudential regulations do not consider covariances so a diversified pool of AAA bonds is treated as if it is as risky as one AAA bond. This is necessary as there was no covariance rating agency which was worthy of the trust earned by the credit rating agencies. Thus the only variable which was independently estimated with some reliability (back then in 1998) was the risk of default of a single instrument.

My current view is that this would be a reasonable approach to prudential regulation if there weren’t firms like AIG financial products. The entities subject to the rules and regulations are large enough that they can diversify their portfolio at a very modest cost. They don’t bear risk for the fun of it. It is safe to assume that they will diversify without being specifically required to do so … unless there is a financial engineering industry which sets up special purpose entities with diversified portfolios and issues single securities whose ratings are high because of that diversification. This means that the diversification (by the SPE) suddenly relaxes the prudential requirement.

By pre-diversifying the financial engineers reduce the further reduction in risk available from diversification. The limiting case would be reached if all securities were put in a huge pool which was cut into tranches. At that point, the risk of default on any portfolio of, say all existing AA securities would be as high as the risk of default on a single security, because there would only be one AA security. Thus for the same prudential regulations, banks would be allowed to bear much more risk.

This is not socially useful. If prudential regulations are optimally adjusted to take into account the increased correlation of different assets, then nothing is accomplished. Otherwise the regulations are effectively changed by agents who can pocket the expected value of future public bailouts.