"The AIG Fiasco or How Not To Manage your CDO Exposure
I remember sitting in on a meeting with an AIG portfolio manager sometime in early 2007 where the topic of conversation was the corporate CDS market. It was a standard chat where we talked about the market environment as well as the most recent product innovations.
Though mostly unmemorable, there was one moment in the meeting that I will never forget. As the marketing guys were pitching mezz tranches to the PM, I threw in a comment that if credit spreads were to widen the delta of the tranche would go up thus increasing the mark-to-market (MTM) sensitivity, and thus net credit exposure, of the trade. This the PM calmly brushed aside responding “we are not MTM sensitive” as he reached for another piece of fruit.
How about them MTM-apples now?
In this post I thought it would be interesting to touch upon a couple of issues that were brought to light from the AIG fiasco. These are: wrong-way counterparty risk management plus the actual (if, surprising to some) nature of risk that AIG was underwriting.
My basic points are the following
- From comments made by AIG executives it appears that the company fundamentally misunderstood the nature of risks that it was underwriting. Those risks were
a) much more highly correlated than they assumed (due to the nature of bonds in CDO structures as well as the likely performance of super-senior tranches in event of impairment)
b) actually mark-to-market risk, not default risk which actually made AIG’s business much riskier than it thought. This is because long before super-senior tranches became impaired (the only risk AIG was worried about), AIG will have had to post more collateral than the cash it had on hand effectively guaranteeing its bankruptcy long before it had to “make the policy holders whole”
- The logical consequence of the previous point is that buying protection from AIG on ABS CDO’s is horribly wrong-way (discussed below) or, to use an analogy, akin to buying deep out-of-the-money puts from a company on its own stock. In other words, that protection is worthless. The consequences of this point are that
a) internal risk management groups inside investment banks were massively short AIG to compensate for the wrong-wayness of this exposure and
b) investment banks who bought protection from AIG, while fully aware of the zero value of the protection they were buying, were continuing the charade only in order to continue originating CDOs.
- The new CDS clearinghouse will fundamentally change the nature of counterparty risk.
Brief Outline of What Happened
The broad outlines of the story are the following. As part of an effort to expand its insurance underwriting business, AIG (more precisely, London-based AIG Financial Products) began writing protection on supersenior (senior to AAA) ABS CDOs. By the time lax underwriting standards led AIG to get out of this business in 2005, it had sold some $560bn of protection.
By 2007 spreads had widened enough that counterparties started to demand that AIG post collateral on the trades, which by mid 2008 totaled over $16bn. Following its first and second quarterly losses of $5.3bn and $7.8bn, AIG, under pressure, adjusted the valuation methodology for its CDO portfolio (word at the time was the company was not mark-to-marking the trades) - leading to a further $8bn writedown. On September 15th - the Monday following the Lehman default, AIG’s rating was cut, effectively guaranteeing a bankruptcy of the company. Concernerned about the effect on world markets, the government stepped in with a bailout.
The Enabling Factors of the AIG’s CDS Underwriting Business were the following
- AIG did not have to post collateral on the trades which, combined with their view of these trades being “free money”, meant they sold protection in astounding size
- US Investment Banks needed entities to sell them super-senior protection so they could “complete the capital structure” and continue originating CDOs. Monolines and CPDC’s were the other enablers.
- European Banks needed AIG to sell them protection to provide regulatory capital relief under Basel II
The Misunderstood (by AIG) Nature of AIG’s Risk
AIG believed that the likelihood that super-senior tranches take losses is so infinitisimal that it was “money good”, according to Joe Cassano, the former head of the unit. As I mention above there are two interesting issues to consider here: one of correlation and the other of mark-to-market vs. default risk.
On the first point, and taking housing as an example, it’s likely that AIG underestimated the probability of super-senior tranches suffering losses since for this to happen you would need a nationwide fall in housing prices (something we havent seen too often) as actual mortgage bonds making up the CDO tend to be well diversified by region. If you look at the data, you would think the likelihood of this happening is nil, however, this is exactly what happened. Although AIG avoided the worst vintage years, it clearly underestimated the risk that a banking crisis followed by a recession would have a nationwide impact on housing prices . Continuing the logic, if one super-senior tranche gets hit, indicating we are in a nationwide slump in housing prices, that means other super-senior tranches are very likely to be hit as well. Selling protection on ABS CDOs on a diversified nationwide portfolio of mortgages is quite different from holding a well diversified portfolio of fire insurance policies. The former is much less diversifed than the latter
Mark-to-Market vs Default Risk
The second point about AIG’s risk has to do with the fact that instead of taking a well-justified massive punt on super-senior tranche default risk (infinitesimal), it instead took a massive punt on credit spread (i.e. mark-to-market) risk and it did so close to the top in the credit market. Simply backtesting this strategy and assuming a high corelation between all credit products (well-justified), AIG would have blown up several times in the last 20 years. By the time default risk became a possibility, AIG will have run out of cash posting collateral to its counterparties, which is effectively what happened this time around.
Counterparty Credit Risk (technical)
Counterparty credit risk is concerned broadly with the risk that a counterparty to an OTC trade will be unable to make the payments as required under the trade (say, due to a default). Let’s say a bank does a trade with a corporate and after some time the corporate defaults. We have two cases:
- The trade is mtm-negative to the bank: - the bank closes out the trade and pays the corporate the mtm with a net loss of zero to both parties
- The trade is mtm-positive to the bank: - the bank closes out the trade, receives nothing on the trade and becomes a creditor in the corporate’s workout process.
This suggests that the bank’s current exposure to the corporate is the maximum of the contract’s market value and zero.
While the current exposure is known, future exposure can be obtained by calculating the expected exposure at each simulation date. Different types of products will have different exposure profiles. For example, amortizing products like interest rate swaps will have a decreasing exposure profile because as time goes on a decreasing amount of cashflows remains to be exchanged between counterparties (see below).
On the other hand, an FX forward has a rising exposure profile because the MTM of a single cashflow at maturity is expected to drift away from current value.
While we know our potential future exposure, what we are really after is the counterparty credit exposure. This number represents the possibility of loss of value in the trade due to a counterparty default and is basically equal to the difference in values between a risk-free trade and a risky trade i.e. a trade that takes into account counterparty default risk.
In practice, this amount is reserved against the trade or, in other words, not recognized immediately.
The credit exposure calculation above is essentially the expectation of discounted exposure contingent on a default by the counterparty. So far we haven’t said anything about a dependence between the counterparty credit quality and the exposure.
When that dependence exists, it is known as right-way/wrong-way risk. Wrong-way risk means the exposure increases as the counterparty credit quality worsense. Clearly, wrong-way risk is undesirable as the counterparty is more likely to default just as when our trade is more likely to be mtm-positive to us.
Wrong-way risk trades include the following:
- A bank receives fixed and pays floating oil price to an oil producer (wrong-way because an oil producer’s credit quality will suffer when oil prices are low just when the bank’s trade is positive-mtm to the bank)
- A bank buys usd/brl forward from the brazilian government (wrong-way because the trade is positive to the bank when BRL depreciates, which suggests a lower government credit quality – see Russia’s experience defending the RUB)
- A bank buys CDS protection from an insurance company (wrong-way because credit spreads tend to be correlated suggesting that when the trade is positive mtm to the bank, the insurance co’s credit spread is wider)
Did you catch that last one? This is what happened with AIG.
In fact, I would argue that the credit quality of AIG was not just somewhat correlated to the credit quality of the insured CDOs but was in fact 100% correlated, especially in the case that matters i.e. impairment of super-senior tranches. By the time this happens AIG will have gone bankrupt posting collateral and even if it survived up to this point the very high correlation between the super-senior tranches it wrote protection on means AIG would have to pony up an unbelievable amount of cash.
So, where does that leave us? Making the back-of-the-envelope assumptions above of 100% correlation in credit quality between AIG and its insured CDO as well as zero recovery, the value of protection that investment banks bought was zero. Remember that the correct value of the trade is the risk-free valuation less the credit exposure. In our case, the credit exposure would be equal to the risk-free value of the trade.
Why did Banks buy Protection from AIG?
Did the banks realize the value of its protection held against AIG was zero? Of course they did - they aren’t as dumb as the media suggests. The reason they continued to pay the full market CDS offer (rather than a much lower level due to AIG’s massive wrong-wayness) to AIG was because they considered it a cost that allowed them to continue originating CDOs. If they could not offload super-senior risk to someone, their originating desks would be effectively shut down.
So, while the trading desks continued to buy super-senior protection from AIG, the risk management desks, realizing that the protection was effectively worthless, bought protection on AIG itself from the street and clients in large size. In fact, I would imagine the size they needed to buy was too large and they likely ended up buying puts on the AIG stock or just shorting outright. Let’s hope the Fed unwinds of AIG’s trades took into account the huge gains these banks took on the AIG hedges.
Onwards and Upwards: the CDS Clearinghouse
In the better late than never column, market participants are establishing a CDS Clearinghouse whose members will face the clearinghouse on all trades, rather than each other as is the case now. This will help in assigning trades, posting collateral, unwinding trades etc. This will hopefully do away with zero-collateral posting by AAA counterparties, which means that selling protection in massive size will be less of a “free money” trade than before."
Don the libertarian Democrat says:
“Did the banks realize the value of its protection held against AIG was zero? Of course they did - they aren’t as dumb as the media suggests.”
I couldn’t agree with you more. Maybe people will finally get this.