Saturday, September 27, 2008

The wrong-headed media coverage of AIG

The media has been writing all kinds of garbage about the fall of AIG. The general themes are that 1) AIG didn't know how to handle risk. 2) They should not have stepped outside of their familiar territory, insurance, and ventured into credit derivatives and 3) They were brought down by massive CDS losses related to mortgages.

Lets start with the first one that AIG does not know how to handle risk. These CDS derivatives that everyone is blaming for the collapse in not what brought them down. They have experienced very little actual losses on these instruments. What brought AIG down was the confluence of mark-to-market accounting with overly aggressive credit rating agencies combined with a panicked credit market leading to a liquidity crisis. I will get back to this later.

Now on to the second one, that AIG should not have ventured into the market for these credit default swap, CDSs, that were the specialty of investment banks like Goldman Sachs. People say, they should have stuck with what they know, insurance. The reason this is absurd is that CDS ARE insurance contracts. The opposite conclusion should be reached that ONLY insurance companies should be writing these instruments NOT banks. CDS are contracts where the writer agrees to make a payout when some event of default occurs. This is no different from when an insurance company agrees to make a payout when a fire occurs. They just need to estimate the chance of such an event and be sure that they are adequately capitalized to handle the losses. In fact, Eric Dinallo, the New York, Superintendent of Insurance is now pushing for CDSs to be classified as insurance and demanding that anyone writing such contracts get an insurance license.

Now let me explain why CDS losses did not bring down AIG.

AIG had $441B worth of CDSs as of last quarter, June 30, 2008. $307B (3/4 of the portfolio) of these are "regulatory capital" related. These are not risky by any means. In short AIG agreed to accept some of the risk of bank loans in order to allow banks (about half of them European banks) more room to lend until the new Basel II regulatory capital changes go into effect in about a year. Generally speaking, AIG will start to suffer losses on this portfolio when loss rates on these loan portfolios hit 23%. This is for bank loan portfolios. Since banks are generally leveraged about 10 to 1, a 10% loss on a portfolio are usually fatal for banks. Losses like this generally only happen during major depressions like the Asian financial crisis or the Great Depression and have never happened all over the world at the same time. Losses as large as 23% are that much rarer. It has never happened and short of nuclear armageddon, never will. Losses to date on this $307B regulatory capital portfolio is $125 million or 0.04% of the portfolio. To put this in perspective, AIG makes about $10B/year from insurance and probably more than a billion dollars from the premiums on these very contracts.

So that takes care of 3/4 of their CDS portfolio. There are three other parts of this CDS portfolio that I will discuss, corporate credits, CDOs with subprime and CDOs without subprime.

CDSs on corporate credits of which they have $54B of exposure. These are CDSs which will pay out in case certain companies go bankrupt. Losses to date have been about $1B or less than 2% of the exposure, which is about the same as losses experienced on a typical corporate bond portfolio. This can easily be made up by having yields 2% higher than risk-free treasuries. So far these have likely been very profitable just like the regulatory capital related CDSs discussed above.

Then there is the CDSs on the multi-sector CDOs which can be broke up into subprime related and not subprime related. The exposure of the part not related to subprime is $22B. The "losses" to date have been $3.5B. The CDSs related to subprime have an exposure of $57B with "losses" to date of $21B.

Now I will discuss these "losses" and explain why I put this in quotes. One needs to understand a bit about how these losses are calculated. I will also explain a bit about how these CDSs work. The CDSs are insurance contracts against defaults in things called CDOs. The CDOs are the instruments which receives the cash flows from underlying mortgage backed securities which are pools of mortgages. I won't go into the details here but the key thing to understand is this. If the CDOs start producing losses then the writer of the CDS, that is AIG, has to pay the CDO holder the cash that was not received. So far, hardly any of these CDOs have defaulted. That has to do with the fact that these are "super-senior" CDO tranches. This means basically that there are other less senior tranches that need to be wiped out completely before they lose any money at all. That is the issue of subordination. It is like being in the 10th row of the British army. If you get attacked, the people in front of you will get the bullets and you won't get shot at until all of them are dead. It affords protection.

So if the CDOs have not defaulted, then why do they have "losses". This is due to so called mark-to-market accounting. Basically this is because the CDOs are marketable. Parties occasionally, but rarely, sell them. The accounting rule FAS 157 requires that the CDOs be marked to the market prices. The CDS writer needs to assume that these prices are correct in predicting what the losses will be in the future. So regardless of what the losses actually will be, the CDS writer can be brought down simply by the market panicking and selling off these CDOs which is exactly what happened to AIG. So the trouble is not really that AIG underestimated the risk of losses on these instruments. It is that they did not foresee the effect of market panic of the value of their insurance contracts because usually this has no bearing. For example, if there is a major hurricane like Katrina, people often and irrationally get more risk averse to hurricanes. They think that maybe global warming will result in more dangerous and more frequent hurricanes. This may be true but the event of a hurricane does not change the truth of it.
If insurance contacts traded on the market, you would find that they would increase in value and that the writers would experience large losses. But they don't trade on the market and insurers do not account for hurricane insurance contracts in this way. Insurance companies are allows to estimate their own losses and keep reserves for what they think they will be. This has been the way insurers have accounted for losses as long as there has been insurance and accounting. So far, it has worked pretty well as long as they are regulated by insurance commissions.

If CDSs were only written by insurers, they would likely be accounted for in the same way. If they were accounted for like insurance contracts then AIG would not have taken large losses, not yet anyway. AIG has estimates that their losses from CDSs related to subprime loans would be less than $8B even in extreme circumstances. They have been forced to mark down $25B due to the market panic and sell off in CDOs. I won't even get into all of the facts that explain why AIG's losses will be much smaller than people think.

What happened to AIG was this. Due to mark-to-market accounting, they have to write down their CDS portfolio to unrealistic levels. Because of this, they had to raise capital to avoid credit rating downgrades from companies like Moody's and Standard and Poor's and Fitch. These companies being accountants cared only about the numbers and the rules and not the reality of the situation. Credit rating downgrades would force AIG to come up with cash collateral to give to those holding the CDS contracts, mostly New York investment banks like Goldman Sachs and Morgan Stanley. The credit ratings agencies realized that AIG might not be able to raise the cash to do this. Because of this, they saw this as a major business risk. A downgrade and a collateral call would cause a default which would cause a bankruptcy event at AIG. Because of this new risk, they felt AIG needed to be downgraded anyway. It was a catch-22 of sorts. If they were vulnerable to a liquidity run, then they were not deserving of their current rating. Because of this the downgrade the liquidity run was assured. So the credit agencies gave AIG a short amount of time to find the cash. Once the markets became aware of AIG's situation, AIG because radioactive. No one would be first to put in any cash if they were not sure that others would join in. If you put in cash and it wasn't enough, then your money might get locked up in bankruptcy and you might not even get it back. So the liquidity run began on AIG. S&P gave AIG just three days to sell off some of their subsidiaries to raise the cash to avoid a downgrade. AIG was not able to do this in time and as a last resort went to the Federal reserve. The rest is history as the Fed seized 79.9% of the common shares in return for a high interest collateralized loan. Such was the fate of AIG.

So in summary, the media is reporting all kinds of non-sense about AIGs fall. The reality is that AIG did not anticipate the difference that mark-to-market accounting would have on their CDS insurance contracts as opposed to their typical insurance contracts. They did not underestimate the risk of losses. They underestimated how a market panic could create fictitious losses that would have real effects, ultimately fatal ones, on their business.