Computers & Economists - A Disastrous Combination; Part I of II

This week's article describes how computers have been misused by economists and Wall Street.  In the three examples briefly discussed below, a completely misplaced confidence in the sagacity of computational sophistication to manage enormous amounts of leverage and peer into the economic future was the root cause behind three enormous economic debacles; all of which produced enormous and long-running repercussions.   

• Long Term Capital Management
• Changes to the Net Capital Rule
• AIG's Trade in Mortgage Credit Default Swaps

Long Term Capital Management (September 1998))
Long Term Capital Management, (LTCM), developed mathematical models to determine what prices should be.  It used the result of the calculations to enter into derivatives contracts.  Because the prices predicted by its model were often only slightly different than the prices which prevailed in the market, the only way for LTCM to make serious money was to use enormous amounts of leverage with its derivative trades.  (In these trades, LTCM would essentially 'bet' that the market price would move toward the price predicted by its model.)  For every $1 in capital, LTCM might borrow $30.  With this amount of leverage, very small losses could wipe the firm out.  Stripped of its mathematical complexity, the geniuses at LTCM thought they could make a fortune by picking up pennies in front of freight trains. 

Some idea of the complexity - and the enormous intellectual arrogance - behind LTCMs mathematical models can be gleaned by a review of the educational background of some of LTCM's partners. 

• Greg Hawkins, Larry Hillibrand, William Krasker, Robert Merton, David Mullins & Eric Rosenfield all had MIT PHDs.
• Robert Merton was also a professor at MIT and won the Nobel Prize in economics. 
• Eric Rosenfield and David Mullins both taught at Harvard.
• David Mullins was also vice-chair of the Fed.  (1)

The mistake all these MIT educated geniuses tripped over was merely the first rule of counting cards.  LTCM made enormously large bets via their derivatives contracts.  However, the basis of their bet - their mathematical models - was never accurate enough to justify the high leverage they routinely used.  It didn't matter that LTCM's models were correct 90% of the time when it only took a 5% level of inaccuracy for the firm to collapse!  LTCM would collapse in September 1998, and this would prompt the biggest mistake of the Greenspan era at the Federal Reserve; the emergency rate cut of October 1998.  (2)

Net Capital Rule (August 2004)
The net capital rule was a regulation that essentially limited the amount of leverage a Wall Street firm could employ.  The 'rule' was within the power of the SEC to change.  The SEC chair while much of the back and forth on the net capital rule took place was William Donaldson.  Donaldson never spent a day of his 45-year professional life working in this country's wealth producing industries.  Instead, he became the typical Rockefeller republican and carried water for both Richard Nixon and Nelson Rockefeller.  Later, he was president of the NY Stock Exchange and founded a Wall Street firm.  By the time he became SEC chair in 2003, Donaldson was completely incapable of looking at a problem objectively.  Instead, he was hard-wired to advance the interests of the people he had gone to school with and spent his entire working life with.  Unsurprisingly, this is exactly what he did. 

In 2004 and as a result of industry “urgings” from the likes of Henry Paulson and others, the net capital rule was changed - by voice vote no less.  In its place, Donaldson and the SEC allowed the largest Wall Street firms, the broker-dealers, to use “an alternative risk-based approach to satisfy the Commission’s regulatory capital requirements, instead of using the current net capital rule.”  (3)  Basically, the Wall Street firms argued that now that they had computers, computers would allow them to manage much greater amounts of leverage.  The SEC, unanimously, agreed.

Avoiding the legalese of the regulation, the practical impact of the changes to the net capital rule were they allowed much higher amounts of leverage.  In fact, Merrill Lynch, Bear Stearns and Lehman Brothers were likely leveraged at least 30:1 just before they each went down in flames in 2008, just four years after the net capital rule was changed.  The SEC’s tolerance for much higher levels of leverage seemed to be based in large part on supposedly more accurate and more sophisticated computer models.  The idea was these models would allow the broker-dealers to better measure and assess the risks they were running, and greater amounts of leverage could thus be used.

The changes to the net capital rule and the attendant reliance on a “risk-based” approach was an unmitigated disaster.   Of the five broker-dealers who qualified for the relaxed standards resulting from the changes to the net capital rule – Bear Stearns, Lehman Brothers, Goldman Sachs, Merrill Lynch and Morgan Stanley – only Goldman and Morgan Stanley would survive the financial crisis intact.  In the immediate aftermath of Lehman’s collapse a former SEC official, Lee Pickard, called the 2004 changes to the net capital rule, “the primary reason for all of the losses (among broker-dealers) that have occurred."

AIG and Credit Default Swaps (November 2008)
In the years of the housing bubble, AIG was, by far, the biggest player in mortgage credit default swaps, (CDS).  In its CDS trade, AIG issued “insurance” on mortgage bonds.  This insurance – which was issued not as a conventional insurance policy but provided though complicated trades in financial derivatives – protected the purchasers of the insurance against a specific mortgage bond losing value.  However, and evidence of the otherworldly aspect of the “insurance” provided by AIG via CDS, the people who purchased the insurance rarely owned the mortgage bond that was being insured!  Basically, the CDS issued by AIG was a bet or wager on whether the mortgage bond would lose money or not.  By issuing the insurance AIG was betting that the bond wouldn’t lose money; the purchasers of the insurance were betting that the bond would lose money.  AIG’s CDS strategy - when stripped of its Ivy League pedigree – was virtually identical to that employed by LTCM.  Specifically, both were the investing equivalent of trying to get rich by picking up pennies in front of a runaway freight train; the risks overwhelmed the potential reward.

The “bet” that was being made with CDS was not an even-money bet.  The bonds that were being insured were considered of very high quality and the risk of the bonds ever suffering losses was considered very low.  A useful analogy for the market for mortgage credit default swaps in the early 2000s is the heavyweight championship boxing match between “Iron” Mike Tyson and James “Buster” Douglas in February 1990.  (4) In the same way that mortgage bonds were judged very unlikely to ever lose value, Mike Tyson was considered very unlikely to lose to Buster Douglas, and was an overwhelming favorite in the fight.  The final betting line had Tyson as a 42:1 favorite to beat Douglas.  To put these odds in perspective, a winning $50 bet on Tyson would only pay about $1.20!  In much the same way, and because the bonds it was insuring were believed to be of such high quality, AIG had to provide enormous amounts of loss coverage for relatively small premium payments.  It was not unusual that for as little as $35-million in annual premium payments, AIG was willing to expose itself to as much as $1-billion in potential losses on each CDS trade it entered.  

To manage this enormous amount of risk, AIG heavily relied on computer models developed by an Ivy League finance professor, Wharton's Gary Gorton.  These computer models proved no more accurate than those used by LTCM or those used by Bear Stearns and Lehman Brothers as part of the new net capital regulations.  The last people to realize the error-riddled nature of these models was AIG itself.  As late as December 2007, when mortgage bond bombs had already started to detonate all over Wall Street, both Gorton and senior AIG management expressed complete confidence in the computer model and AIG's CDS trades.  (5)

• Gorton claimed  the models “… are guided by a few, very basic principles, which are designed to make them very robust and to introduce as little model risk as possible.  We always build our own models.  Nothing in our business is based on buying a model or using a publicly available model.”  
• AIG’s CEO, Martin Sullivan, claimed Gorton’s models gave the company a “high level of comfort.”
• Joseph Cassano, the head of AIG Financial Products, believed the models were “simple, they’re specific and they’re highly conservative."

Concluding Remarks
I have two degrees in mechanical engineering and have spent my entire professional working life as an engineer.  So, I hope this keeps me from being accused of being a Luddite.  Computers are a tool, period.  They can - and certainly have - had profound impacts on how work is done.  However, computers can't alter the work that still needs to be done.  Too many economists and financiers confuse numbers stored with 15-signficant digits to be accurate within 15-significant digits.  They are not and the three examples discussed here provide mute testimony of this confusion.  In next week's article, we will see what happens when mechanical engineers start thinking like economists and Ivy League professors. 

 

Peter Schmidt
October 04, 2020
Sugar Land, TX

P.S. As always, if you like what you read, please consider registering with the site.  It just takes an e-mail address, and I don't share this e-mail address with anyone.  The more people who register with the site, the better case I can make to a publisher to press on with publishing my book!  Registering with the site will give you access to the entire Confederacy of Dunces list as well as the Financial Crisis timeline.  Both of these are a treasure trove of information on the crisis and the long-running problems that led to it. 

Help spread the word to anyone you know who might be interested in the the site or my Twitter account.  I can found on Twitter @The92ers

Endnotes:
1.  When Mullins left the Fed for LTCM, he would be replaced by Alan Blinder, another MIT PhD.  Some idea of Blinder's view on economics can be gleaned from the fact he was an ardent supporter of the "cash for clunkers" program.
http://www.the92ers.com/dunce/alan-blinder

2.  http://www.the92ers.com/dunce/alan-greenspan

3.  Alternative Net Captial Requirements for Broker Dealers that are Part of Consolidated Supervised Entities, Securities and Exchange Commission, https://www.sec.gov/rules/final/34-49830.htm

4.  http://www.the92ers.com/blog/tyson-douglas-mirage-hotel-sportsbook-and-aig-debacle

5.  Carrick Mollenkamp, Senea Ng, Liam Pleven and Randall Smith, "Behind AIG's Fall, Risk Models Failed to Pass Real World Test," Wall Street Journal, October 31, 2008 https://www.wsj.com/articles/SB122538449722784635