Risk assessment and black swans

by Michael Smith
NBS Information Specialist

Most events involve some degree of uncertainty; for instance, we may know an unplanned event is absolutely guaranteed to occur, but not when. Risk assessment requires calculating the probability of a specific result and assess whether it's acceptable to carry on as before. If we're dealing with day-to-day events, that's fine, but what about rare and extreme events? A risk manager needs to know not only what can go wrong, but how wrong it may go.

Category of event

The likelihood of any future event happening can be broken down into three categories:

• Known known
• Known unknown
• Unknown unknown.

With a known known event risk, enough is understood to be able to predict how likely it is that the event will occur. With a known unknown event risk, there is an understanding that the risk exists, but not enough information to make accurate predictions about the event or why it may happen. For instance, the mechanics of earthquakes are well understood, but they are almost impossible to accurately predict.

The final category, unknown unknown, made famous by the then US secretary of defence, Donald Rumsfeld, takes a little more effort to understand. An unknown unknown risk is one that has not been considered at all and probably not even thought about.

[T]here are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns – the ones we don't know we don't know. (Donald Rumsfeld – 2002)

While the above statement sounds strange, it does make sense when worked out.

Simple? No. Ponder this, then: until the discovery of Australia, swans were considered to be white. This did not signify that all swans were white, only that no one had ever seen a black one; it was an unknown unknown.

The 'Black Swan problem' states that, "no matter how many swans you find, the fact that they are all white, so far, can't prove that other colours don't exist".

Black Swan events

To be a true Black Swan event, an event must have three distinct features:

• The event is a surprise to the observer
• The event has an extreme impact
• After it happens, the event can be predicted with hindsight.

Black Swan events are unknown unknowns, and thus impossible to accurately predict or risk assess, but they can be prepared for. The ability to rationalise a Black Swan event in hindsight separates it from being pure chance. After the event, the cause can be seen or calculated and the precursor evidence will be seen to have been there all along.

There is an ongoing argument about whether recent events in Japan were Black Swan or not because of hindsight: it was well known that earthquakes of magnitude 9 or more were possible and it was also known that fault lines lay off the coastline of Japan. Given a tsunami of a certain size, it was obvious which areas would be inundated and that these would include the standby generators at Fukushima. If a full risk assessment had been carried out looking at geological, seismic, reactor safety and emergency services factors, it may have concluded that the chain of events which occurred on 11 March were entirely possible, in effect a Black Swan event.

Micromorts

What of the people involved in these events? What is the risk to them? The human brain is not good with large numbers and will often see 'once in a million years', as 'not for another million years'. Looking at this another way, if there was a one millionth chance of being in a car crash, out of every million people, statistics say that one person would die in this way. Statisticians call this one millionth chance of death a 'micromort'. The overall chance of unnatural death in the UK currently stands at 350 micromorts per year. Other micromorts of risk include:

• Drinking a litre of wine – 1
• Spending a week in New York – 3
• Spending a night in a UK hospital – 75 (source - London School of Economics).

Predicting risk

Traditional risk management relies on identifying risks based on the experience of the teams involved in the enterprise. If the risk is outside the experience of the group it is unlikely to be considered, and even if it is, it is likely to be prioritised as low by being allocated an extremely low probability rating.

Risk management concentrates of managing the risks to the enterprise that would have a significant impact and have a reasonable probability of occurring. This is simply a way of prioritising potentially 'bad' events so that time and resource can be allocated. Brainstorming risks is also highly unlikely to capture Black Swans. The exercise will either be too narrow, by staying within the comfort zones of the participants, or too broad by considering risks that are not relevant to the business.

Another method of predicting risk of this catastrophic nature is by using 'ABCD Risk Management', which considers the strategic assumptions of the business: those things which must happen for the business strategy to succeed. Analysis by ABCD turns to thinking assumptions rather than risks, which helps to focus on the objectives of the business.

ABCD Risk Management

Assumption Based Communication Dynamics (ABCD) is a novel, formal methodology that enables the capture of differing knowledge and viewpoints from stakeholders, in a form that facilitates communication of issues, assumptions and ensures pro-active management of risks. By dramatically improving communication, risks are avoided or managed proactively and objectives are delivered on time and to budget.

ABCD is considered to work where other risk management approaches fail. One of the stated main reasons for this is that the system focuses on positives (assumptions) rather than negatives (risks) and is therefore seen as a positive exercise by people who would normally feel uncomfortable with a risk assessment.

The core of ABCD is Assumption Analysis. This method uses structured techniques to analyse plans and associated documentation and identify the most sensitive assumptions that are potentially unstable, and therefore the source of greatest risk.

Conclusions

It can be argued that because of their nature, there is no point in trying to predict black swan events in business, as it is impossible to predict purely random events; unfortunately it is these events that often have the greatest impact. Instead, risk management should concentrate on building robust general purpose response plans, able to deal with any eventuality.

Further information

Trying to quantify uncertainty
A lecture at the LSE by Professor David Spiegelhalter
http://www2.lse.ac.uk/publicEvents/events/2010/20101117t1830vHKT.aspx

De-Risk: ABCD Risk Management theory
http://de-risk.com/page.php?7.

July 2011

 

ViewAllArticles

BackToTop