Prism / 1 / 2008

11

Black Swan Events –

Should you be concerned?
Philip W. Beall, Rodolfo Guzman, David Lyon, Geoff Stevens,
Eleonor Kramarz

The discovery of black Black Swan events have entered the jargon following the
swans in Western Aus- success of Nassim Nicholas Taleb’s book ‘The Black Swan:
tralia was a shock for The Impact of the Highly Improbable’. Taleb writes about
scientists. Today those very low-probability / very high-impact events with three
unexpected birds have characteristics:
become a symbol for
the disruption of the bell 1. They lie outside the realm of regular expectations. In
curve that is used for statistical terms these events are not in the main body
most forms of variation. of the bell curve used to account for many common
Bhopal, Exxon Valdez and forms of variation. Black Swan events are outliers in the
Société Générale have tails of the distribution and interestingly these tails are
shown that, in business, often “fat” because there are more events than expect-
the highly improbable ed in the extremes.
can occur, with devastat-
ing consequences. In this 2. A Black Swan event carries an extreme impact. One of
article the authors take Taleb’s contentions is that nothing in the past can con-
a closer look at some of vincingly point to the possibility of such events.
these incidents, examine
the effect on companies 3. In spite of its unforeseen character, human nature
and show what you can makes us concoct explanations for a Black Swan occur-
do to prevent a Black rence after the fact, making it explainable and predict-
Swan event happening to able.
you.
When Black Swan events occur in the context of company
operations, they can put the affected company in a poten-
tially unsustainable position, as illustrated by the position
of Union Carbide following the Bhopal disaster in 1984.
But if they are so rare and unpredictable, are such extreme
events something for companies to be concerned about?
And if so, is there anything that they can do proactively to
help prevent them? In this article we review some of the
consequences of Black Swan events and explore possible
preventive approaches.

Fat tails and unexpected major losses

In order to explain the concept of a “fat tail”, it is useful to

consider the catastrophic losses resulting from accidents,
such as large fires or explosions, suffered by oil and gas
 Black Swan Events – Should you be concerned?
12

companies. The corresponding loss data that insurance

companies collect can be plotted to show the probability of
losses of various sizes (see Table 1).

Table 1 Loss history: losses vs. probability of occurrence

60%

50%
probability of occurrence

40%

30%

20%

10%

0%
0 100 200 300 400 500 600 700
Loss (million USD)
Source: Insurance Company Loss Data, Arthur D. Little Analysis

The graph illustrates that 90 per cent of these serious

events result in losses of less than $100 million. It is inter-
esting to note that when we work with operators trying to
estimate the potential size of losses, very few are willing
to consider individual events with losses of this magnitude.
Typically loss-ranking matrices adopted by such companies
do not envisage losses greater than $10 million. But the
When Black Swan events loss history shows that not merely are the majority of such
occur in the context of com- losses 10 times larger, a small proportion are larger still and
pany operations, they can are true “Black Swans”.
put the affected company in
a potentially unsustainable Yet these events are not as rare as one might imagine. The
position, as illustrated by data in the graph include 165 major fires and explosions
the position of Union Car- that occurred over a 40-year period; in other words, on
bide following the Bhopal average more than four such events occur in the industry
disaster in 1984. each year. Of course the “on average” does not happen;
the characteristic of these very large losses is that they do
not occur on average and (to the consternation of insurers)
often come in flurries.

In recent years there has been increasing attention paid by

financial modellers to these rare extreme losses under the
general description “Extreme Value Theory”. Particular atten-
tion has been paid to the shapes of the tails in the loss profile.
The attempts to fit the data in Table 2 illustrate the point.

Prism / 1 / 2008
 13

Table 2 Loss history: Fitting a curve to the “fat tail”

60%

Loss Histroy
50%

probability of occurrence
Exponential fit
“Beta Fit”
40%
“Bell Curve”

30%

20%

10%

0%
0 100 200 300 400 500 600 700
Loss (million USD)
Source: Insurance Company Loss Data, Arthur D. Little Analysis

It may be seen from the loss history line that the normal
Taleb notes that, in spite of distribution (the often-cited bell curve) does not fit the bill
the unforeseen character of - the tail of the loss history curve clearly shows that very
a Black Swan event, human large losses occur with greater probability than would be
nature makes us concoct expected. Skewed distributions are required to model the
explanations for its occur- curve, and in this example a much better fit is achieved us-
rence after the fact. ing an exponential distribution. This point is not just about
esoteric statistics. The important message is that if compa-
nies wish to predict the scale of losses that can occur (for
example by Monte Carlo simulation), appropriate “fat tail”
distributions should be used for modelling.

The share price impact of a Black Swan event

The third of Taleb’s three characteristics of Black Swans

helps us to understand the way in which stock markets
respond. Taleb notes that, in spite of the unforeseen
character of a Black Swan event, human nature makes us
concoct explanations for its occurrence after the fact. The
good news is that this suggests that one completely unex-
pected loss can be accepted and confidence in the affected
company can be quickly restored. The bad news is that a
sequence of unexpected losses eventually does impact
share price negatively because it may point to systemic
shortcomings in the company’s operations.
 Black Swan Events – Should you be concerned?
14

Table 3 Loss history: losses vs. probability of occurrence

100 Investor discussions
BP (USD/share) Alaska Criminal Prosecution September 2006
90
June 2006
Crude Oil (USD/barrel)
80
Adjusted Monthly Price (USD)

gasoline (USD/litre)
70

60

50

40
Threat of Criminal Prosecution
30
December 2005
20

10 Grangemouth Incidents Texas City Incident

June 2000 March 2005
0
2000 2001 2002 2003 2004 2005 2006 2007
Source: BP Investors Relations, USEIA and Arthur D. Little analysis

The history of the BP share price is a good illustration of

the point (see Table 3). It can be divided into two periods,
the first from 2000 to 2005, and the second after 2005.

Table 3 shows the response of the BP share price to ac-

cidents at its Grangemouth refinery in the UK and the sub-
sequent legal proceedings. At the time of the accidents,
which attracted national coverage, in June 2000 there was
a decline in value of BP stock but that was matched by sim-
ilar falls in other oil stocks. At the time of the court hearing
18 months later, and the imposition of a heavy fine, there
was, if anything, a strengthening of the BP price relative
to crude oil and gasoline prices. Despite the accident, the
BP price tracked the crude oil price for several months. The
pattern suggests that, for this very large company, both the
accidents themselves and the court judgment and fine had
a negligible impact on company valuation.

Table 3 also shows the response of the BP share price to

an accident in Texas City three years later. A major cata-
strophic hydrocarbon release and consequent explosion at
BP’s Texas City Refinery led to the deaths of 15 staff and
injured 180. This major incident led to a significant drop in
relative share value of about 10 per cent, which persisted
for about eight weeks before recovery was complete. Even
the announcement of a fine of $21.4 million in Septem-

Prism / 1 / 2008
 15

ber 2005 appeared to have no depressing effect on the

share price. Indeed the reaction was somewhat similar to
that noted at Grangemouth, where defining the penalty
seemed to lift the market. (In both cases the level of the
penalty was dwarfed by the cash flow generated by the
company.)

But things started to change from then onwards. When

in December 2005 a criminal prosecution was announced
concerning the Texas City incident, the relative share price
showed a drop of up to 10 per cent, which was not fully
recovered. Six months later another criminal prosecution
was announced, this time concerning an oil pollution inci-
Recovery is possible from dent on the Alaska North Slope. This occurrence depressed
a Black Swan event, even the share price by a further 15 per cent and there was no
though the short-term evidence of recovery. Three months later, after persistent
consequences can be very concerns about BP management performance, one-to-one
severe. Single catastrophic discussions were held between major investors and the BP
events are often “forgiven”, board. The news of these meetings depressed the relative
as a result of the process share price by a further 20 per cent, again with no signs
Taleb calls “concocting of recovery. The sequence contributed to the unplanned
explanations for Black Swan resignation of the BP chief executive in May 2007.
occurrence after the fact”.
The insight to be drawn from this story is that, for such re-
source-rich corporations, recovery is possible from a Black
Swan event, even though the short-term consequences
can be very severe. Single catastrophic events are often
“forgiven”, as a result of the process Taleb calls “concocting
explanations for Black Swan occurrence after the fact”.

Punishment can be hard, however, when a sequence of

events points to systemic and consequently manageable
failures. The Baker Panel investigations into the Texas City
disaster, released in January 2007, brought to light sys-
tematic failure in BP’s safety management. This could be
interpreted as an “explanation” for an event that the Texas
City operators would have probably regarded as incredible
prior to its occurrence. However, the critical factor was
that during this period other incidents occurred, none of
which was as catastrophic as that at Texas City but which
were able to be painted, perhaps unfairly, as apparent cor-
roboration of the “explanation”. Where the “explanation”
developed post-event to account for the Black Swan event
is apparently validated by subsequent short-term incidents,
 Black Swan Events – Should you be concerned?
16

damage to market capitalisation may not be recoverable

until a management change occurs.

How to prevent Black Swan events

It is perhaps all too easy to point a finger at BP, but com-

placency would be unwise. Whilst this is one of the more
recent examples of a corporate Black Swan event, others
have already taken place since and will continue to occur in
the future – there are not many who would have predicted
the losses at Bear Stearns in the US and Northern Rock in
the UK, for example.

In theory, Black Swan events cannot be identified and

therefore cannot be prevented. In practice, Black Swan
events are typically a result of a long chain of smaller
sub-events that in series lead to catastrophic experiences.
If one or more of the sub-events are removed from the
series, then the chain is broken. This gives us a possible
means of preventive action.

To illustrate this point, let’s consider a Black Swan event

In practice, Black Swan experienced by Exxon. On March 24, 1989 the Exxon Val-
events are typically a result dez tanker hit a reef in Prince William Sound, Alaska, and
of a long chain of smaller released approximately 262,000 barrels of crude oil. Exxon
sub-events that in series spent billions of dollars remedying the release and monitor-
lead to catastrophic experi- ing potentially long-term impacts.
ences. If one or more of
the sub-events are removed As with the BP Grangemouth experience, the Exxon stock
from the series, then the value was little impacted (see Table 4). Unlike BP, however,
chain is broken. This gives Exxon and later ExxonMobil have avoided subsequent
us a possible means of Black Swan events. Its adjusted stock consistently in-
preventive action. creased in value even during times of stagnant crude oil
and retail gasoline prices during the 1990s and has tracked
crude oil values since the large run-up in prices since 2002.

Prism / 1 / 2008
 17

Table 4 Evolution of ExxonMobil share price

100
ExxonMobil (USD/share)
90
Crude Oil (USD/barrel)
80
Adjusted Monthly Price (USD)

gasoline (USD/litre)
70 Exxon Valdez
March 1989
60

50

40

30

20

10

0
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Source: ExxonMobil Investor Relations, USEIA and Arthur D. Little analysis

ExxonMobil has avoided Black Swan events for nearly 20

years. It seems hard to believe that it has just been “lucky”,
although of course there will, necessarily, always be a
certain element of luck. It is fair to assume that the com-
pany’s strong record after Exxon Valdez is largely the result
of enhancing its Operational Integrity Management System
(OIMS) through a nearly religious devotion to continual
improvement in performance – perhaps a level of devo-
tion that can only be achieved after suffering a catastrophic
event such as Exxon Valdez.

OIMS is an 11-element safety, health, environmental and

security management system. Element number 2 strongly
emphasises risk identification, assessment, management
and monitoring (see Table 5 next page). Ultimately, Exx-
onMobil believes that OIMS leads to increased reliability
and to superior safety, health, environmental and security
performance, thus providing a competitive advantage.

There are plenty of other examples from history that show

that rigorous and robust risk identification, assessment,
management and monitoring may not identify all potential
Black Swan events but could well mitigate their effects. For
example, it was inconceivable that the Titanic would sink
in the North Atlantic on its maiden voyage after striking an
iceberg. However, the chain of events that led to the unim-
 Black Swan Events – Should you be concerned?
18

Table 5 Risk Management at ExxonMobil OIMS

Comprehensive risk assessments can reduce safety, health, environmental
and security risks and mitigate the consequences of incidents providing
essential information for decision-making.

2.1 Risk is managed by identifying hazards, assessing consequences

and probabilities, and evaluating and implementing prevention and
mitigation measures.

2.2 Risk assessments are conducted for ongoing operations, for prjects
and for products in order to identify and adress potential hazards to
personnel, facilities, the public and the environment.

2.3 Periodic risk assessements are performed by qualified personnel,

including expertise from outside the immediate unit, as appropriate.

2.4 Risk assessments are updated at specified intervals and as changes

occur.

2.5 Assessed risks are addressed by spcified levels of management

appropriate to the nature and magnitute of the risk, and decisions
are clearly documented.

2.6 A follow-up process is in place to ensure that risk-management

decisions are implemented.

Source: TonenGeneral

aginable loss of life and property could possibly have been

prevented or mitigated using a risk identification, assess-
ExxonMobil has avoided ment, management and monitoring programme. The Titanic
Black Swan events for was designed to be its own lifeboat in the event that it was
nearly 20 years. It seems sinking. We know now that the Titanic split in two pieces
hard to believe that it has just before sinking, which vastly accelerated the vessel’s
just been “lucky”, although demise. Did the Titanic’s designers evaluate potential
of course there will, neces- reasons why the vessel sank faster than designed and did
sarily, always be a certain they improve subsequent designs? A more robust expan-
element of luck. It is fair to sion-joint design would not have prevented the Titanic from
assume that the company’s sinking, but it could very well have allowed the ship to fulfil
strong record after Exxon its design feature of being it own lifeboat and floating for
Valdez is largely the result several more hours, allowing additional passengers to be
of enhancing its Operational rescued.
Integrity Management
System (OIMS) through a Société Générale’s trading loss is a more recent example
nearly religious devotion to demonstrating that Black Swans can occur in all industries
continual improvement in and that risk management principles should and can be
performance. applied to all fields. On current knowledge the Kerviel affair
appears to relate to a long sequence of small events within
Société Générale, where a former back-office technician
circumvented trading controls and managed to establish
trades for more than the value of the company. As with BP,
the chain could have been broken by applying the principles

Prism / 1 / 2008
 19

of a robust risk management system involving complete

independence of the control team from the corporation.
Indeed, Kerviel was from the back office, where he had
acquired in-depth knowledge of the control procedures.
Ongoing investigations will probably reveal the nature of
the specific Black Swan event that led to the biggest trad-
ing loss in financial history.

With increasing awareness and concern of corporate

responsibilities, people are less and less forgiving of those
Black Swan events. The public now assumes that risk
should be fully managed by corporations, which have the
economic capacity to implement robust risk identification,
assessment, management and monitoring throughout the
company.

Application of a process safety management system

Example 1: Transport industry
Process safety management system elements can be applied to the transport industry,
including highways and trains. Very similar risk identification, assessment, management
and monitoring techniques as used for the oil and gas and chemical industries can
also be used for transport. For example, while advising a major railroad project on the
implementation of a risk management system, we helped to look at both individual risks
and risk sequences that could impact the project. While many of the individual risks
did not have major consequences, the approach was able to help the project identify a
sequence of reasonably probable/ low-consequence risks that, if they were to combine
simultaneously, would have resulted in very major delays for the project, far larger than
the aggregate impact of the risks taken individually. Although the project believed that
such an event could not occur, a plan was developed to monitor this chain of risks so
that appropriate risk avoidance and mitigation actions could be taken if needed.

Example 2: Medical industry

Risk identification and management processes can be applied to any industry where
processes can be mapped. For example, we advised a manufacturer of vaccines on
major risks to business continuity, and helped identify areas where it could strengthen
its current risk controls, including disaster recovery. Process safety management techni-
ques employed included:
Potential risk identification
– Development of risk register
– Risk assessment and ranking using Monte Carlo
probability assessment
 Black Swan Events – Should you be concerned?
20

Insights for the Executive

Can potential Black Swan events be prevented? The an-

swer is a qualified yes. However, usually it is impossible to
prove, since Black Swan events are by definition inconceiv-
able and rare. The key to preventing Black Swan events
is breaking the chain of sub-events that allows the incon-
ceivable to happen. Bhopal, BP-Texas City, Exxon Valdez,
the Titanic and the Hindenburg all had their own chains of
events that led to the catastrophes.

The following insights can be drawn from considering the

history of Black Swan events:

• Black Swan events should be considered the results of

long sequences of low-impact/high-probability hazards.
Although it appears that they cannot be predicted, the
development and passionate application of rigorous risk
identification, assessment, management and monitor-
ing programmes across company operations can poten-
tially remove elements of those sequences. It allows
company processes to remain within design and control
parameters, hopefully breaking the chain of events.

• If company operations can be set down as a detailed

process flow, then process safety management (PSM)
systems, including risk management, developed for the
chemical and petroleum refining industry, can be used.
Our experience shows us that PSM systems can be
used effectively for many industries, not just oil and gas
(see box text for two examples).

• A well designed risk identification, assessment, man-

agement and monitoring programme allows the com-
pany to look beyond a specific incident to understand
the outcomes, not just the outputs – for example, what
does the incident mean for levels of fines, protests,
investor questions, licence to operate? How should we
adjust risk assessments in other parts of the business?
Effective risk management programmes integrate the
engineer’s and external affairs executive’s views of
the world, linking technical and stakeholder issues to
ensure that outcomes are clearly recognised.

Prism / 1 / 2008
 21

• Risk management programmes can improve your ap-

proaches to identifying and understanding potential sys-
tematic failures. For example, are there specific issues
at a site level that have implications at business unit or
corporate level? To what extent do risk-based manage-
ment systems address issues in any asset around the
world? What are company-specific key performance
indicators forewarning that a Black Swan chain of events
could be initiated?

You might still think that it could never happen and, indeed,
the most likely outcome is that it will not. But if you take
the relatively modest time and effort needed to take these
precautions, catastrophic events can potentially be prevent-
ed in our very unpredictable and uncertain world.

Philip W. Beall
... is a Principal within Arthur D. Little’s Sustainability and Risk practice. Philip has over 30 years
of work experience including 15 years technical and management experience in refining, gas, and
petrochemical facilities and more than 17 years international experience in environmental, health
and safety technical assessment. E-mail: beall.philip@adlittle.com

Rodolfo Guzman
… is a Director in the Global Energy Practice based in Houston. He has over 15 years on consulting
experience in the oil and gas industry and has provided strategic and organizational advice to many
leading international and state-owned companies around the world.
E-mail: guzman.rodolfo@adlittle.com

David Lyon
… is a Senior Manager in Arthur D. Little’s Sustainability and Risk Group in Cambridge, UK.
He focuses on enabling companies derive business value through sustainability including use of social
and environmental driversfor revenue protection, cost reduction, revenue growth and lowering risk
exposure. E-mail: lyon.david@adlittle.com

Geoff Stevens
... is a Principal in Arthur D. Little’s Cambridge office in the sustainability and risk practice. He has
spent some 18 years at Arthur D. Little. His client work covers risk identification, assessment and
management in Energy and Transportation industries. E-mail: stevens.geoff@adlittle.com

Eleonor Kramarz
... is a new Consultant within Arthur D. Little’s Sustainability and Risk practice. Since joining Arthur
D. Little she has taken a great interest in the assessment of risk and prevention of safety incidents.
E-mail: kramarz.eleonor@adlittle.com