What Might Artificial Intelligence Have Told Us About 'The Smartest Guys in the Room?'
Enron, the infamous US energy and commodities company that failed in 2001 is the gift that keeps giving. It continues to attract interest and deliver valuable lessons in corporate governance 18 years later. As I will describe, however, it is possibly an even more interesting a case study now we can view it through the lens of artificial intelligence (AI).
Firstly, though, some background. In 2002 while working with a client board in Texas, I was able, one evening, to watch a C-Span replay of a session of the Senate Committee inquiry into Enron’s sudden descent into bankruptcy. In Washington, earlier in the day, the former Enron board committee chairs had been paraded in front of the Committee. One-by-one they attempted to convince Committee members that Enron had been governed by the best board they had ever been a member of. Committee members had difficulty keeping a straight face, but the assertions of those Enron directors had some basis. For one thing, the company had recently won an award as one of the top five boards in the US. That was recognition that Enron had ticked many of the ‘boxes’ touted as corporate governance best practice. For example:
- the roles of chair and chief executive had been separated (very rare in the US at that time);
- the Audit Committee chair was conspicuously well-qualified for the position;
- compared to many US corporate boards the Enron was relatively small.
A prominent visual backdrop to the hearing was an exhibit prepared by staff assisting the Committee. Senators routinely referred to this in their questioning of each of the Enron directors. It was a timeline of the company’s recent history on which a sequence of decisions by either the board or the Audit Committee had been red-flagged. Each of the highlighted decisions was thought to have contributed in some material way to the company's demise. Each was contrary to a board policy or inconsistent with the company's much-touted values. The most damaging of these decisions appeared to have been approvals of various special purpose entities (SPEs) proposed by the company’s CFO, Andrew Fastow. SPEs were mechanisms for additional borrowing backed by Enron’s own shares that did not show up on the balance sheet. The company’s true financial position was distorted further by the company's practice of booking profits (‘mark-to-market’) that were not yet earned (and might never be).
Enron went from share market darling to bankrupt in a matter of months once various commentators started questioning the substance of the company’s reported earnings and the SEC started investigating. It was hardly surprising that in concluding the session of the enquiry I was watching, the Senate Committee Chair, observed to the Enron board representatives that “…you not only fiddled while Rome burned, you toasted marshmallows over the flames!”
Ultimately, the spectacular failure of the company proved, contrary to their own self-assessment, that the Enron folk were not ‘the smartest guys in the room’. (1) Malcolm Gladwell concluded that ‘Enron’s S.P.E.s were, by any measure, evidence of extraordinary recklessness and incompetence’. He also referred to the conclusion of the Powers Committee, a panel charged with investigating Enron’s demise, that even though its directors sat in meetings where those deals were discussed in detail, many on Enron’s board failed to understand the economic rationale, the consequences, and the risks of their company’s SPE deals. (2) In Kurt Eichenwald’s ‘Conspiracy of Fools’, arguably the definitive story of Enron, he contended that even Fastow, the principal designer of Enron’s various financial manoeuvrings, didn’t understand the full economic implications of the tangle of financial obligations he created. (3)
In 2004 Andy Fastow, ‘CFO of the Year’ in 2000, was sentenced to ten years imprisonment for wire and securities fraud. He served six years of that sentence and now speaks around the world about what went wrong at Enron and the lessons to be learned. He also provides consulting services helping to identify where a company may be technically complying with rules but is being materially misleading. Ironically, he even teaches classes on business ethics.
The problem, he routinely tells audiences, was that while the deals he carried out at Enron were ultimately materially misleading and fraudulent, they were fully disclosed to and approved by Enron’s accountants, outside auditors, its inside and outside attorneys and the Enron board. The deals were, arguably, legal but they were not right. As Fastow has suggested his real role at Enron would have been better described as ‘Chief Loopholes Officer’. (4)
Something Fastow says he is concerned about today is that most directors have limited information to work with. They are given only a few metrics such as earnings and credit ratings, and [selective] information from the CEO. Consequently, directors he says, don’t know when they need to get involved. While some directors might try to do too much, Fastow asserts that most do too little because they don't have the tools. (5)
It is possible to see, however, that the ‘tools’ situation may be changing. In a recent article in the AICD’s Company Director, John Green reported on a development that shows how artificial intelligence (AI) may have predicted the Enron failure. (6) KeenCorp, a Rotterdam based company, has developed an AI system that uses psycholinguistics to examine employee engagement in real-time. The system analyses the tone of internal communications looking not for what employees are saying but how they are saying it. Firstly, a base is created through an analysis of historical email traffic to give a sense of what is a company’s normal ‘tone of voice’. Using this datum, the system can then identify pattern changes that may indicate a shift in the collective mindset.
In testing their system, KeenCorp researchers applied it to years of publicly available emails from Enron’s top 150 employees. They found a dramatic plunge in confidence occurred in June 1999, more than two years before the company’s collapse. This was surprising – and counterintuitive – given, at that time, that Enron’s star was shining brightly, and its share price and market capitalisation were still increasing rapidly.
Concerned there may be a flaw in the system KeenCorp contacted Andy Fastow. He was able to tell them that, contrary to their fears, their analysis had identified the exact day – 28 June 1999 – when he persuaded Enron’s board to accept one of the most controversial of the structured finance deals. Fastow identified it as the deal that eventually led to the SEC investigation. (7) The AI analysis had uncovered the negative reaction to this decision among Enron’s top 150 people. Even though they knew the board would be making the wrong decision, none of those executives had spoken up about it. What the AI system picked up though was that this was how that group of executives was feeling about the decision. Apparently, it showed tension levels that were almost as high as on the day Enron declared bankruptcy.
This example shows us how a smart analysis of internal communications could have been a powerful tool to alert the Enron board to what was really going on. If this kind of real-time AI had been available to them in 1999, the board would have been aware of the plunge in key employee engagement caused by the board’s decision on what others saw as a particularly dodgy deal. Possibly the Enron board may have been brave enough to try to unwind that deal – and others like it. This information may have saved the company.
(1) This phrase stems from the book (Bethany McLean and Peter Elkind. The Smartest Guys in the Room: The Amazing Rise and Scandalous Fall of Enron. Portfolio Trade, Reprint edition, 2004) and subsequent documentary film.
(2) Malcolm Gladwell. ‘Open Secrets’. The New Yorker, 8 January 2007. (https://www.newyorker.com/magazine/2007/01/08/open-secrets-3)
(3) Kurt Eichenwald. Conspiracy of Fools. New York, Broadway Books, 2005
(4) Anon. ‘Ex-CFO of Bankrupt Enron Offers D&O Lessons from Accounting’s ‘Gray’ Areas’. Insurance Journal, 15 February 2017 (https://www.insurancejournal.com/news/national/2017/02/15/441619.htm)
(5) Anon. ibid.
(6) John M Green. ‘Early Warning System’, Company Director. Australian Institute of Company Directors, November 2018, p. 44-45.
(7) Anon. ibid.