Monday, November 10, 2008

AIG's Failed Quantitative Risk Models

Last Monday's Wall Street Journal featured a page one piece on the failure of AIG's risk management models.

It definitely caught my attention, because I had written recently on the fallacy of composition in risk management on Wall Street in this post just over a month ago.

"Here we have an unintended consequence affect the situation, due to the fallacy of composition.

Such a downgrade leads each counterparty of such an affected, downgraded firm to require more collateral on each position held with that firm as a counterparty. Thus, in the blink of an eye, or the stroke of a pen at a rating agency, the financial collateral requirements for a firm's book of positions with other trading partners rises significantly.

This is what actually drove AIG into ownership by the US Treasury. Its downgrading by a credit rating agency caused AIG's counterparties to require more collateral for swaps and other insurance arrangements it had sold than the firm had capital available to provide for such needs.

Do you suppose any of the quantitative, computer-based risk management systems of any of AIG's counterparties had the capability to model, forecast and integrate into their risk estimations such an occurrence? Did any of AIG's counterparties have enough knowledge of AIG's exposures to allow it to reasonably estimate the effects on that firm's capital position, and, thus, its risk as a counterparty, if it were downgraded?

I doubt it.

Thus, in yet another perverse way, the individual, similar actions of many financial service players, collectively, lead to a result which causes more risk and uncertainty in the system, even as each individual party seems to act to reduce its own risk to its positions and its counterparties."

Well, the Journal's AIG piece basically confirms my suspicions. Specifically, the article begins by describing how AIG's risk modeling was dependent upon the work of an outside consultant, Gary Gorton, who teaches at Yale's management school. It notes,

"Mr. Gorton, who teaches at Yale School of Management, is best known for his influential academic papers, which have been cited in speeches by Federal Reserve Chairman Ben Bernanke. But he also has a lucrative part-time gig: devising computer models used by the giant insurer to gauge risk in more than $400 billion of devilishly complicated deals called credit-default swaps.

AIG relied on those models to help figure out which swap deals were safe. But AIG didn't anticipate how market forces and contract terms not weighed by the models would turn the swaps, over the short term, into huge financial liabilities. AIG didn't assign Mr. Gorton to assess those threats, and knew that his models didn't consider them. Those risks have cost AIG tens of billions of dollars and pushed the federal government to rescue the company in September.

The swaps expose AIG to three types of financial pain. If the debt securities default, AIG has to pay up. But there are two other financial risks as well. The buyers of the swaps -- AIG's "counterparties" or trading partners on the deals -- typically have the right to demand collateral from AIG if the securities being insured by the swaps decline in value, or if AIG's own corporate-debt rating is cut. In addition, AIG is obliged to account for the contracts on its own books based on their market values. If those values fall, AIG has to take write-downs.

Mr. Gorton's models harnessed mounds of historical data to focus on the likelihood of default, and his work may indeed prove accurate on that front. But as AIG was aware, his models didn't attempt to measure the risk of future collateral calls or write-downs, which have devastated AIG's finances."

Thus, AIG used a fairly simple set of assumptions which neglected to consider both the effects of actions by other investors in similar assets, and the resulting changes in collateral, should those investors' actions lead to downward price spirals.

It should give you pause to realize that there has been a fairly widespread knowledge of the cascading effect of position liquidation ever since the failure of 'portfolio insurance' in the 1987 Crash. That's over twenty years ago.

Where have Mr. Gorton and AIG's senior and risk management executives been during that time? How could they have possibly believed that a quantitative risk management system which deliberately omitted effects of panic selling and attendant collateral increases due to other position value declines would have much relevance to the real world facing AIG and its trading and investment books?

Yet, even with such glaring holes in its approach to risk measurement and management, AIG paid Mr. Gorton a small fortune to do an inadequate job. Nice work, indeed, if you can get it. You can bet that Mr. Gorton will not be liable for any damages to AIG.

To understand how out of touch with real market behavior Gorton is, the article quotes his comments to one of his classes at Yale recently,

"On a rainy morning last week, Mr. Gorton briefly discussed with his Yale students how perplexing the struggles of the financial world have become. About 30 graduate students listened as Mr. Gorton lamented how problems in one sector caused investors to question value all across the board. Said Mr. Gorton: "There doesn't seem to be a fundamental reason why." "

Of course there is a fundamental reason why. That 'one sector' just happened to be one involving sophisticated investors and traders valuing structured and arcane financial instruments. When the CDOs, MBSs and swaps all went untradeable due to real, identifiable underlying trends, e.g., the cooling of the mortgage origination market and the observable rise in alt-a and subprime mortgage defaults and delinquencies, the broader markets began to realize that the so-called 'sophisticated' investors had gotten valuations terribly wrong.

In effect, the fact that many institutional investors who were presumed to be so astute were seen either holding untradeable paper, worth little in a 'mark-to-market' world, or dumping that paper at large losses, when possible, caused bystanding investors to wonder what the value of anything now was. Especially the equity and debt of the publicly-held institutions visibly undergoing value evaporation due to having misunderstood the values of these exotic structured instruments.

This whole mess didn't start because your grandmother worried about the value of Google's equity.

It began because sharp-elbowed asset managers and traders at opportunistic investment banks and hedge funds, such as Bear Stearns, Lehman, Goldman Sachs, and Citadel, began to react, logically, to the news of economic trends in mortgage markets which had all-too-obvious implications for holders of structured finance instruments and swaps.

That's why it only took one sector's actions to cause the resulting cascade of value destruction and, eventually, AIG's demise amidst a sea of poorly-understood risks on its balance sheet.

No comments: