Tags
Adjustable-rate mortgage, AIG, American International Group, Collateralized debt obligation, Complexity, Credit default swap, Fannie Mae, Hank Greenberg, Howard Sosin
Preface
This post is the latest to consider aspects of the 2007-2008 Financial Crisis, which illustrate the difficulties we face in finding its “root causes.” My previous analysis considered information systems and culture, offering insights into shortcomings that contributed to the catastrophe. This series focuses on the impact that increasing complexity in global markets had on AIG and its Financial Products group.
This aspect of modern systems is becoming increasingly important to our understanding and management of risks and opportunities. Unfortunately, complexity and its effects are often overlooked, in part because we do not have the tools, and some would say the language, to effectively work with it. I trust this post will give you some things to think about as you consider the growing complexity of the increasingly interconnected world we live in.
An Odyssey Begins
AIG‘s Financial Products Group had its modest beginnings shortly after a meeting between a team of Drexel employees, headed by Howard Sosin, and Hank Greenberg, AIG’s legendary chairman. The resulting joint venture proved successful as Sosin’s team reaped $60 million in profits during FP’s first six months of operations.[1]
Those profits came courtesy of the strategies and system (PASS) that Sosin’s team conceived and later brought to life at FP. They were able to “monitor the minute fluctuations in various rates” within the derivatives market, allowing the company to “place offsetting trades on all sides of a transaction, so it almost didn’t matter what the markets did. That was the beauty of their evolving machine: The firm won either way, as long as it stuck to its commitment to keep hedging its bets.”[2]
So if FP initially avoided real estate and was so successful with derivatives, how did it nearly bankrupt AIG with subprime mortgages? From a complexity perspective, the failures were caused by market interdependencies, disruptions and adaptations. The first major disruption was the 1993 Greenberg–Sosin dispute, Sosin’s departure and Tom Savage’s promotion to lead FP.[4] By then, competitors had adapted to FP’s dominance by developing their own strategies to close the gap, putting pressure on the FP team to find new business opportunities. The expanding subprime market proved an alluring target.
Beginning in the early 1990s, the Federal Government started pressing banks to approve more loans to people with a poor or nonexistent credit history—another disruption. By 2001, the market had adapted by expanding the subprime market to over $100 billion. A related adaptation was the bundling of mortgages (including subprime) and their securitization as components in Collateralized Debt Obligations (CDOs). CDOs were structured in layers or tranches, with the highest being the the least risky and lower paying, while the lower ones had higher risks and offered higher returns. These adaptive strategies created complex interactions and new interdependencies with secondary mortgage market entities like Fannie Mae and Freddie Mac.
An Added Complexity
In 2001, Savage approached Greenberg with a strategy to offer CDO originators default insurance, called Credit Default Swaps (CDS), for the most secure CDO tranche. Given the apparent safety of these tranches, the risks were considered minute, although there were two areas of concern. The first was that individual CDS contracts were difficult to fully model for risks; the second was that no hedging strategy was available to offset risks. Both represented major departures from Sosin’s original risk-containment strategies. If CDOs failed in large quantities, AIG would be at risk. After considering the issues, Greenberg gave the concept a green light, and FP began offering CDS products to CDO originators, quickly becoming the go-to CDS source.[5]
Four years later, Eugene Parks was asked to take over CDS marketing. By this time, the complex financial and real-estate markets had adapted to the Federal Government’s pressure to approve more loans by increasing subprime mortgage originations by 400 percent, from approximately $150 billion in 2001 to over $600 billion in 2005.
The rapid growth created another disruption because loan originators needed more capital to lend, which triggered adaptations that included securitizing more mortgages through CDOs loaded with a single family of debt: real estate in general and lots of subprime loans in particular. This high-risk structure affected all tranches, including the top ones that FP had concluded were safe and for which AIG had issued tens of billions in Credit Default Swap contracts.
The loan origination market had also adapted by reducing or eliminating down payments on subprime loans and expanding the use of Adjustable Rate Mortgages, which offered an initial period of very low-interest rates and low payments, followed by increases starting about three years later; the premise was that increasing home values and established credit would allow borrowers to secure fixed 30-year, FHA-guaranteed loans before the increases made payments unaffordable. The elimination of down payments meant that there was no buffer in case of wide drops in real estate values, which increased the risks posed to originators and, by extension, to FP.
Problems Pile Up
The explosive increase in subprime mortgages also triggered disruptions in the real-estate valuation and underwriting market, which struggled to meet demand. The market adapted by shifting the balance between efficiency and quality of valuations and underwriting towards greater efficiency and throughput at the expense of quality. Poor valuation was another unexpected risk emanating from unforeseen disruptions and adaptations, but loan originators, regulators and FP failed to recognize the implications.[7]
Finally, the widespread use of subprime mortgages across the country undermined regional variations that loan companies had long relied upon to contain the impacts of downturns. Regional markets historically moved independently of each other based on regional economic conditions, thus providing a kind of hedge against risks from national downturns. Government policy had accidentally turned the American real estate and subprime mortgage markets into national markets, something that existing risk models had not taken into account.[8]
The Fix That Failed
The rapid growth in the CDOs market triggered the need to speed up traditionally time-consuming risk assessments, without which CDOs could not be priced. This potential bottleneck disruption found a solution in the theoretical work of David Li, a Chinese-Canadian scholar, who in 2001 had proposed that a relatively simple mathematical formula could be used for this purpose.[9] The Gaussian Copula was widely adopted by mortgage originators and used by financial managers, most of whom did not understand underlying models, to calculate risks and price CDOs.
Unfortunately, while Li’s methodology worked under controlled conditions, it ultimately failed to accurately estimate risks within the interdependent, complex markets of the global economy, leaving in its wake hundreds of billions of poorly assessed CDOs, many of which had been insured by FP. This is an example of an adaptive strategy leading to disaster.[10]
No Way Out
Back at FP, Eugene Parks had taken the time after his 2005 promotion to personally analyze the mortgage components of new CDOs. He quickly determined that FP’s models had not accounted for the increasing risks posed by subprime mortgages and within weeks convinced FP’s leadership to stop issuing new CDS contracts. Unfortunately, FP had already issued over $80 billion in CDS contracts and was thus fully exposed to the risks that Savage’s team had originally identified—AIG had no exit strategy available.
The final key disruption came courtesy of the Financial Standards Board, a body that sets accounting standards for the U.S. The Board issued a new rule in 2006, FSB-157, that was to go into effect in November 2007 and was intended to more accurately estimate asset values held by institutions such as banks. Banks had traditionally valued their real estate assets using a Price to Model strategy that assigned values based on an internal model. This meant that real estate based assets generally kept their book value, even when their market value dropped.
Why was Price to Model a standard practice? Because it was assumed that real estate based investments were long-term and would recover their true value in time, even after a temporary market downturn. The new rule had the effect of reducing the value of these investments and associated reserves containing real estate in general and subprime mortgages in particular. The change put pressure on banks to price, and sometimes sell, real estate related assets at bargain basement prices in order to shore-up their balance sheets.
The rule’s effects further reduced the amount of capital available for mortgage lending, which depressed real estate prices even more; that, in turn, further lowered the institutions’ asset worth, a downward spiral that triggered defaults in CDOs all the way to the top tranches insured by AIG. The predictable result was worsening market conditions in which CDS counterparties (investment firms like Goldman Sachs and banks holding default insurance) began demanding payment on their insured CDOs. AIG could not come up with the tens of billions in capital necessary to honor those contracts, which led to the largest corporate rescue in U.S. history.
In the next post of this series, I will summarize how the adaptive, interdependent nature of complex systems combined to create a perfect storm for AIG. I will also discuss some of the implications for business and government in our increasingly complex markets.
Please Comment
Have you seen examples in your work of complex systems adapting to change in unexpected ways? What were the results?
Questions?
Visit my website at DecisionsToLead.com or contact me to find out how this information can be applied to benefit your organization.
References
[1] O’Harrow, Robert, Jr.; Brady, Dennis. “The Beautiful Machine.” Washington Post, December 29, 2008, page 4. http://www.washingtonpost.com/wp-dyn/content/article/2008/12/28/AR2008122801916_4.html?sid=ST2010062905395
[2] Ibid.
[3] Form-10K SEC Filing, AIG, December 31, 2010, page 4. http://www.aigcorporate.com/investors/2011_February/December_31,_2010_10-K_Final.pdf
[4] Boyd, Roddy. Fatal Risk – A Cautionary Tale of AIG’s Corporate Suicide. John Wiley & Sons, Inc., 2011, Nook Edition, page 48.
[5] O’Harrow, Robert, Jr.; Brady, Dennis. “A Crack in the System.” Washington Post, December 30, 2008, page 3. http://www.washingtonpost.com/wp-dyn/content/article/2008/12/29/AR2008122902670.html?sid=ST2010062905395
[6] The Subprime Mortgage Market, National and Twelfth District Developments, 2007 Annual Report, The Federal Reserve Bank of San Francisco, page 8. http://www.frbsf.org/publications/federalreserve/annual/2007/subprime.pdf
[7] O’Harrow, Robert, Jr.; and Brady, Dennis. “Downgrades and Downfall.” Washington Post, December 31, 2008, page 4. http://www.washingtonpost.com/wp-dyn/content/article/2008/12/30/AR2008123003431_4.html?sid=ST2010062905395
[8] Pages, Scott E. “Complexity Versus Uncertainty,” Chapter 11 of Understanding Complexity. The Teaching Company, 2009.
[9] Li, David X. “On Default Correlation: A Copula Function Approach.” Journal of Fixed Income, 2000, 9 (4): 43–54. DOI:10.3905/jfi.2000.319253. http://www.defaultrisk.com/pp_corr_05.htm.
[10] Dekker, Sidney. Drift to Failure. Ashgate Publishing Ltd., 2011, Kindle Edition, location 211-275.