A Behavioural Finance view of the Deepwater Horizon Disaster of 2010
INTRODUCTION
The Deepwater Horizon disaster of 2010 was the largest accidental marine oil spill the oil industry has ever seen. It claimed 11 lives and caused huge environmental damage on the shoreline of the Gulf of Mexico by discharging circa 4.9 million barrels of oil. There have been numerous enquiries and studies into the accident to try to understand what happened, the goal of the enquiries was to establish where the fault and therefore guilt lay in order to ascertain compensation. This report attempts to apply key concepts from behavioural finance with a view to understanding the thought processes behind the decisions that were made. In the last decades there have been a lot of studies documenting the impact of psychological traits on the decisions made by managers. This report while touching on them, will not go into detail regarding the various technical difficulties encountered on the Deepwater Horizon and will instead focus on understanding the thought process of the various players involved. We will also briefly cover BP’s historical safety record as this will help us understand some of the behavioural factors in play within the organization. While BP was the principal on the Deepwater Horizon rig there were also other parties like Transocean , the rig operator, Halliburton, who operated in a consultancy position, Anadarko and Schlumberger. These companies played a part in the crisis and their performance and relationship to BP will be explored as well.
BP’S ORGANIZATION STRUCTURE AND CULTURE
BP under Lord Browne was a very silo based organization. Business units were under direct control of asset managers and they and their teams were incentivized by their assets performance. I worked at BP between 2007 and 2008 and it was very clear to me that the various assets were in direct competition with each other. The competitive nature of the organization encouraged rivalries between units that almost resembled the rivalries between football clubs supporters. With each unit under pressure to offer higher returns a culture of cost cutting spread across the firm. This cost cutting compromised processes and in turn compromised safety across the board. This was highlighted in the Baker Report in the wake of the Texas City tragedy. BP’s senior executives sought to view safety through the prism of incident numbers and also to question employees recording incidents on whether they were true incidents. This method of measuring safety is a combination of the availability heuristic and also framing. The information was easy to gather, it’s easy to record individual incidents, and therefore was over weighted as senior managers used to demonstrate that safety was indeed getting better in the individual assets. The act of questioning employees who attempted to record incidents is an example of framing where one would attempt to ask the question a different way to get an answer that was more favourable. BP’s method of determining safety standards is also a consequence of the confirmation bias where evidence is interpreted in the way one wishes to see it. Confirmation biases are normally a consequence of system 1 thinking. In “Thinking Fast and Slow” Daniel Kahneman defines System 1 thinking as “ System 1 operates automatically and quickly, with little or no effort and no sense of voluntary control”. I wouldn’t for a moment suggest that BP’s senior managers didn’t think when set up their flawed method of measuring safety but they used System 1 thinking when they attempted to simplify it to a metric that was easy to measure and easy to understand. Safety is a far bigger issue than simply incident numbers, as we saw with Texas City, Prudhoe Bay and finally Deepwater Horizon there were underlying systemic flaws that only System 2 thinking would enable to be discovered. In “Thinking Fast and Slow” Daniel Kahneman defines System 2 thinking as “System 2 allocates attention to the effortful mental activities that demand it, including complex computations. The operations of System 2 are often associated with the subjective experience of agency, choice and concentration”. In other words System 2 thinking doesn’t encourage snap judgements which would be enabled under System 1.
BP’S RECENT SAFETY RECORD
In March 2005 an explosion at BP’s Texas City refinery killed 15 people and injured a further 180. The financial losses to BP exceeded $1.5 billion. The political fallout of this disaster could be seen in the fact that a former Secretary of State, James Baker, was appointed to lead a committee to investigate what had happened. The committee’s findings were shocking as they highlighted serious safety and maintenance shortcomings which were driven as a result of cost cutting. Shefrin (2008) profiled BP as an organization that possessed all the psychological weaknesses of high firms such as over confidence and excessive optimism. In figure 1, below, we can see the various psychological pitfalls and processes in a 4x4 Pitfall-Process framework. Shefrin (2008) in his study of the Texas City incident applied this framework to analyse the disaster. This framework helps us understand the different psychological attributes and their interaction with business processes - which in an ideal world operated in a bias free environment.

Texas City wasn’t the only big accident BP suffered in the 2000’s. In March 2006 an oil spill was discovered at a pipeline owned by BP in Prudhoe Bay Alaska. 267,000 barrels of oil had spilled into the bay. The spill originated from a hole in a pipeline. BP was fined $20 million.
The Baker Report
The findings of the panel were summarized under three headings: Corporate Safety Culture: Process Safety Management Systems and: Performance Evaluation, Corrective Action and Corporate Oversight. I will now briefly look at these three areas using behavioural finance concepts
Corporate Safety Culture
BP’s senior managers adopted a safety measure totally reliant on personal safety. As discussed above this was due to a confirmation bias and relied on an availability heuristic as it is easy and quick to measure. What this safety methodology didn’t do was examine process safety which would have pointed out inherent flaws in BP’s operating methodology. Relying on personal safety statistics, which were ever improving, gave senior management over confidence in their own ability to manage risks. In “Nudge” Richard Thaler argues that when we are presented with complex choices we attempt to simplify the decision making process. One strategy is “elimination by aspects” (Amos Tversky, 1972). Using this strategy one decides what is important, applies a cutoff level and then eliminates all alternatives that do come up to this standard. This is a version of the confirmation bias where one decides on an answer that one likes and then looks for the data that supports that.
Process Safety Management Systems
BP’s leaders failed to provide process safety standards and adequate process safety training. Process safety in an oil refinery is a complicated discipline. However, just like anything else constant practice helps makes any discipline easier to understand and execute (Nudge, Richard Thaler, 2009). By not setting standards and then properly training staff to follow those standards BP’s leaders were guilty of lowering safety standards. This lack of effective communication could have been brought about by the silo based culture that was fostered across BP where power was transferred to assets and assets were made to compete against each other.
Performance Evaluation, Corrective Action and Corporate Oversight
BP’s safety philosophy has been described in detail above where it has been illustrated that BP focused on personal safety and not process safety. Personal safety was easier and quicker to measure and therefore was useful in confirming the narrative that BP was improving its safety record. Narrative is a useful way of getting a story across. In this case confirming to regulators and the markets that BP’s safety record was good was part of the narrative to the markets that BP was an efficient, profitable organization. By framing the safety record around personal safety BP was able to convince shareholders that the company was doing a good job of risk mitigation. The report also found that BP’s process safety audits seemed to focus on making sure that BP’s processes met all legal requirements. By doing this BP “anchored” their audit process to a belief that meeting the legal requirements meant that adequate precautions were in place to deal with any contingencies. In truth most legal requirements are the bare minimum that companies need to adhere to. By anchoring their process audit outcomes to a legal requirement BP showed that they were unable to react to new information and operated in a narrow confidence band which led to over confidence among senior managers.
DEEPWATER HORIZON DISASTER
The Deepwater Horizon was an ultra-deep water offshore oil rig. The rig was owned and operated by Transocean and leased to BP. It had operated in the profitable Thunder Horse oil field in the Gulf of Mexico and had been described as one of the most powerful rigs in the world (Kaskida, April 2007). In 2009 the Minerals Management Service hailed its safety record as the standard for the industry. It had operated in the Tiber oil field in the Gulf of Mexico which was at the time the deepest oil well in the world and was still deeper than the Macondo well.
The Macondo well was located in the Gulf of Mexico. Initial estimates showed that well may have had 50 million barrels of oil (Klump, Edward, 2010). The rights to drilling the Macondo well had been purchased by BP in 2008 and they owned 90% of it with rest owned by MOEX Offshore.
The Macondo oil well was located 40 miles offshore and 130 miles from New Orleans.
On the day of the explosion (20th of April 2010) there were teams from BP, Transocean, Halliburton, Anadarko and Schlumberger on the rig. The team from BP were responsible for managing the drilling, Transocean for operating the rig under guidance from BP and Halliburton were operating in an advisory role to BP.
The explosion on the night of the 20th of April 2010 took the lives of 11 workers and injured 16 others. It also caused the largest accidental marine oil spill in the world.
The explosion was caused due to a buildup of gas during the well sealing and closing process. Once a well has been drilled it is sealed and closed to enable the engineers to install a pipeline to draw the oil out. The process of closing a well is complex due to the underwater location of the well. One of the key parts of the sealing process is the well casing process.
Well Casing Process
Once the engineers have completed the drilling process the well has to be cased and sealed. There are two ways of casing the well, one is the liner/tieback casing option which provides four barriers or protection. The other is the long string casing which only provides two barriers. The liner string casing method is obviously quicker and cheaper to carry out but also carries higher risks as there are fewer levels of protection. A BP Forward Plan Review document recommended against using long string casing, however internal communications showed that BP was moving towards a long string casing option on the Macondo well. Communications from Brian Morel, the drilling engineer, show how the long string casing option was selected on the basis of it “saving time” while Mark Hafle another BP drilling engineer made the same argument and also suggested it would save money.
Centralizers
As part of the casing process the well would have to be filled with cement to hold the steel pipe in place. Imagine a tall glass with ice at the bottom, if you pushed a straw between the ice so that it stood up straight would have to use the position of the ice in order to make sure that the straw stood up straight otherwise the tip of the straw wouldn’t be perfectly flat against the bottom surface of the glass. In this example you were able to use the ice to position the straw, in the case of an underwater well engineers normally use something called centralizers to ensure that the pipeline that has gone into the well is perfectly straight and once it is straight they fill up the well with cement. If the gap between the pipe and the surrounding rock hasn’t been filled up with cement you would get gas building up on the sides. Halliburton’s role on the Deepwater Horizon was to run a simulation model to ensure that the pipe was straight and that the cementing had been effective to stop a gas build up. Halliburton used a simulation system called OptiCem that took data from engineers and returned results showing the effectiveness of the well design. It was considered highly reliable. In the build up to the explosion the Halliburton running the OptiCem model found that BP would need to run twenty one centralizers to effectively seal the well, BP used six in order to save time. Brian Morel, a BP engineer, emailed John Gagliano (a Halliburton employee) telling him that it was too late to request new centralizers and that they would use the six they had. Tellingly, in his email he says “It’s a vertical hole so HOPEFULLY the pipe stays centralized due to gravity”. Hopefully? This sounds like the optimistic bias raising its head again. Instead of listening to the results of a highly regarded simulation system, OptiCem, Brian Morel uses simple rule of thumb to satisfy himself that he had done the right thing. This is what Herbert A Simon calls satisfice – this is when people seek out a satisfactory solution rather than an optimal one (Simon, Herbert, Models of Man, 1957). We see this every day where people choose to do the bare minimum. They do this because they are inherently lazy, because of time/cost pressures or any other reason. The point is they seek out an acceptable solution rather than the best solution possible.
John Gagliano the Halliburton executive team leader had expressed concern with BP that the OptiCem model had indicated that there was high risk of channeling occurring, channeling being when the pipe is not straight and the cement hasn’t therefore completely sealed the space between the pipe and the rock. Brian Morel however questioned the model’s output claiming some of the earlier outputs were different to what BP’s crew engineers had measured onsite. So, instead of listening to a reliable tool BP’s engineers decided to query its results based on their findings. This framing may have been driven by the affect heuristic where the engineers believed in their own analysis and therefore didn’t want to spend any more time or money to appease the models findings.
Circulating Mud
Before cementing the well it is industry practice to circulate the mud around the pipe. This enables engineers to check if there is any gas leaking in. According to the American Petroleum Institute it is best practice to circulate the mud at least once. BP estimated that circulating the mud would take between six and twelve hours. According to the drilling logs for April 19, the mud was circulated in just thirty minutes. Is this a case of satisficing – do the bare minimum to convince yourself that you have done enough? It certainly seems like it, there is a big disconnect between what BP estimated and what actually happened, was the initial estimate wrong? It seems unlikely as we know it was a difficult well and with hindsight we know that gas did leak and cause the explosion. This goes back to the process safety shortcomings that were prevalent throughout the organization. If there had been a “nudge” in the form of a checklist or in some other form then the engineers would have been forced to circulate the mud for longer and in the process may have discovered the leak.
Consequences of poor processes
As a result of the short cuts taken in the three main processes above there was a gas leak which led to the explosion that sank the Deepwater Horizon and caused the huge oil spill. It was a consequence of engineers’ satisficing themselves and being overconfident in their own abilities regardless of what a non-emotional simulation model was telling them. The Challenger Space Shuttle disaster of 1985 had similarities with Deepwater Horizon where engineers knew that a certain part (an O Ring) was faulty years before the crash but did nothing about it because they felt that it would be ok. Changing the part would have been expensive and may have led to other parts being more closely investigated so they carried out enough simulation to convince themselves that the part worked well. Richard Feynman famously disproved their analysis in a committee hearing by damaging the material used to make the O ring. All he used was a glass full of ice ! Once again framing and the availability heuristic come into play.
AFTERMATH
In the immediate of the accident the BP CEO was pictured sailing, he made unfortunate statements that the spill was literally a drop in the ocean and that he famously wanted his life back. He blamed the accident on nature and the difficulty of the well as opposed to any systemic errors. Going sailing just after an oil spill is certainly an example of hubristic behavior. Like Dr Faustus Tony Hayward refused to show contrition in the face of a serious accident. Some of the qualities that Babiak and Hare use to spot a corporate psychopath are: Considers others stupid, Cold and Calculating, Show no regret. These could be applied to Tony Hayward to show that he was exhibiting typical psychopath behavior in the aftermath of a serious accident. There were process issues at BP which they ignored due to their confirmation biases and they continually used framing and satisficing to convince themselves and other they were making the right decisions. The overall goal was to continue to sell the narrative to investors that BP was an efficient organization that had mitigated risks, unfortunately as Deepwater Horizon illustrated their failure to debias tore that narrative to shreds.

References
Daniel Kahneman, Thinking Fast and Slow, 2011
Hersh Shefrin, Behavioural Corporate Finance, 2007
Richard Thaler, Nudge, 2009
Hersh Shefrin and Enrico Maria Cervellati, BP’s Dailure to Debias, 2011
Amos Tversky, 1972
Kaskida, 2007
Baker Report, 2007
Joseph J Trento, Prescription for Disaster, 1987
Bibliography
Daniel Kahneman, Thinking Fast and Slow, 2011
Hersh Shefrin, Behavioural Corporate Finance, 2007
Richard Thaler, Nudge, 2009
Hersh Shefrin and Enrico Maria Cervellati, BP’s Dailure to Debias, 2011
Amos Tversky, 1972
Kaskida, 2007
Baker Report, 2007
Joseph J Trento, Prescription for Disaster, 1987
Malcolm Gladwell, Outliers, 2008
Michael Macooby, The Productive Narcissist, 2003