Handling Risk and Uncertainty in project analysis...and

beyond.

Dr. Cameron Gordon,

Associate Professor of Economics,

University of Canberra, Australia

Texas A&M 2013

What do we know…• …and when will we know it and how it

will affect us?

• This is one way of thinking of risk.

• Risk deals with the probabilities of 'known' possible events.

• A 'full' analysis should identify the (major) relevant risk types and the expected probabilities of each type over the relevant time period.

What don't we know?• A broader issue is that there are likely

things we don't know at all.

• The classic dichotomy is between 'known unknowns' (things we know we don't know) and 'unknown unknowns' (things we don't even know we don't know).

• This would fall under the general category of uncertainty, i.e. events that cannot be specified at all and/or cannot be reliably assigned a probability.

'Climate Change' might be a general risk type which can be assigned a 99% probability (say) over some time period with specific probabilities for both specific impacts (e.g. high heat days) and intensity (10 high heat days per year versus 20). These are risks.

There also would be climate change impacts that we think might be possible but not for sure (e.g. snow in College Station – not as unlikely as it used to be!) and others we might have any idea of (e.g. the possibility that aliens might find our hotter planet to be hospitable – to be facetious!)

Classic 'engineering' approach to risk: overbuild

It is a stereotype of the 'engineering approach' to risk to solely focus on

• (1) construction and other physical project aspects and

• (2) to overbuild and overspecify to account for those risk in those aspects.

• Example – 'safety factors' of 10x etc.

This is an unfair stereotype in many ways but does point to the need for sophistication of approaching risk.

Real v financial assets (projects)

In a project investment context we also have to distinguish between risks and uncertainties associated with the 'real' asset and those associated with the corresponding 'financial' asset.

Real asset('fundamental)

Financial Asset ('derived')

Real risks are fundamental

The fundamental risks are those which arise from the real project and its operations and impacts.

One can't 'eat money', so to speak; a bad project will ultimately fail regardless of the financing.

So these real asset risks should be identified and predicted first.

A classic real risk: late delivery

Source: oliverwyman.com

A classic real risk: late delivery

Source: oliverwyman.com

Financial risk – 'secondary' but often critical

Although the real asset is fundamental, financial risk is often no less critical to project success.

A change in foreign exchange rates can sink a project even if it is sound from a real asset sense.

The interactions between real and financial should be as well understood as possible.

How to handle risks in project analysis

THE BLUNT INSTRUMENT APPROACH – Sensitivity analysis

This is the simplest approach in which one estimates a project return and then varies assumptions by some range (e.g. 10% above and 10% below the initial estimate) to see how the bottom line varies.

This is particularly useful in high uncertainty scenarios

Sensitivity Analysis example

Engineering Geology,Volumes 145–146, 7 September 2012, Pages 65–77. Application of risk, cost–benefit and acceptability analyses to identify the most appropriate geosynthetic solution to mitigate sinkhole damage on roads, Jorge P. Galvea, Corresponding author contact information, E-mail the corresponding author,, Francisco Gutiérrezb, Jesús Guerrerob., Juan Alonsoc, Ignacio Diegod

The classical probability approach

The classical risk approach is to define events and likelihoods and then probability weight them:

Expected outcome = Event 1 * risk probability 1 + .... Event N * risk probability N.

This is good when risk events and probabilities are well known.

Not good for uncertainty per se.

The real options approach Classical probability is static, i.e. events

are fixed with fixed probabilities. In actual fact one event can lead to a

sequence of other events and there can be multiple chains of events.

Real option models attempt to capture this fact.

These are best when such chains are well-known and there is high technical proficiency. They are subject to modeling sensitivity.

Source: juergendaum.com

The Black Swan One risk issue is that of the Black

Swan, i.e. a supposedly low probability event actually occurring more often than expected.

For most real project flows this is probably not an issue but it can be quite an issue with complex financial flows.

And with climate change we might need to take this into account more and more for 'pure' engineering.