subscribe
 

Beating the unknown

7th May 2015

Posted By Paul Boughton


PERT distribution functions provide adequate uncertainty models set up from three input variables PERT distribution functions provide adequate uncertainty models set up from three input variables
 Investments with the highest-expected NPV may also provide the highest downside potential Investments with the highest-expected NPV may also provide the highest downside potential
Using the presented approach, an additional 30% in NPV could be achieved compared to the previously planned investment Using the presented approach, an additional 30% in NPV could be achieved compared to the previously planned investment

Evgeni Kochman, Fabian Sempf and Michael Kruse take a simplified approach for risk-based decision-making

Thanks to the turbulent business environment, making decisions is becoming both quicker and more complex. Nevertheless, most decisions are made based on single-numeric metrics (eg, net present value or internal rate of return). One reason for the widespread use of such metrics is that they offer simple calculations, but more importantly, decision-makers are familiar with them. They are able to understand the evaluation even with limited knowledge about the underlying assumptions.

The factor of uncertainty is usually included by adding lump-sum contingencies or calculating multiple scenarios using different assumptions (eg, sensitivity analyses). But even with risk-adjusted metrics, the possibility of neglecting huge upside and downside potential is still there. This might crucially affect the final decision, especially in changing and complex environments such as the energy sector.

In this article a strong but pragmatic approach to the assessment of risks associated with business decisions is presented. The approach is illustrated using a real case example in which a business case for a large-scale power plant was developed.

Knowing the rules

Of the vast number of underlying assumptions, which are the main risks affecting the potential outcome? Three broad categories of risks can be identified that need to be dealt with in decision-making:

* Low-impact uncertainties are the large number of small, independent risks with relatively low impact on the business case (eg, late deliveries and cost overruns).

* Non-negligible uncertainties involve manageable amounts of uncertainty with significant impact and a realistic probability of occurrence. Often, these uncertainties are correlated additionally. An example would be a political election that influences the boundary conditions of the business case (e.g. regulations).

* Black swans/catastrophic events are surprising, rarely occurring events with major impact on the outcome of a scenario (eg a vapour cloud explosion or a plane crashing into the power plant).

The focus of the assessment should be on non-negligible uncertainties. Experience shows that low-impact uncertainty can be covered by reasonably adding a contingency based on project management knowledge or historical data. Black swans 'distort' the business case and have to be treated separately from the assessment model (eg, through expected maximum-loss calculations and transferring these to insurances).

Non-negligible uncertainties represent the main uncertainty drivers and can therefore be crucial to the outcome of a business case. Emphasis on these factors can add certainty to the results and enhance the basis of the decision-making process – especially if complemented by ‘stress tests’ of the model through the assessment of major risks.

One common example of a non-negligible uncertainty in the power plant business is the market price of electrical power. For such uncertainties, very sophisticated and complicated price prediction models exist. But if businesses aim for reduced complexity, the PERT distribution provides a robust but easy-to-use solution that allows a fair balance between accuracy and model complexity. Historic data of the price development can then be used to derive the required input variables for the distribution: minimum value expected, maximum value expected as well as most likely value to occur.

Accordingly, uncertainty models can easily be designed for other influential factors (eg engineering costs). The PERT distribution offers the possibility of asking an expert about how she thinks certain factors will develop (including most likely, minimum and maximum values). This information is sufficient to adequately model uncertainties without having to burden experts with complicated mathematical constructs. In addition, the approach is easy to communicate when explaining the basis of the business case.

Playing your hand

Most of today’s business cases are based on large and often unmanageable numbers of assumptions. Integrating uncertainties into each assumption and combining all factors into a final result is difficult. A proven method is the Monte Carlo analysis. Based on uncertain input data, the Monte Carlo method generates a large number of scenarios to approximate the entire range of potential outcomes. And even though it is an approximation, the result is precise enough to give the decision-maker a decent understanding of the uncertainty level associated with an investment. Furthermore, a better comparison of alternative investments is possible.

An example from a power plant evaluation provides a good example of the advantage of risk assessments. Based on a simple NPV analysis, ‘Investment C was considered the best investment opportunity. But after taking the associated risks into account, huge downside potential was revealed. In this example, 'Investment B' presented the most reasonable choice, having a moderate expected return but a very low spread of potential outcomes.

Building Monte Carlo analyses from scratch is demanding. But existing solutions provide reliable and manageable approaches to Monte Carlo assessments in business applications (eg, integration of the Monte Carlo method within existing Excel models). Still, the challenge remains to systematically structure the risk analysis around the requirements of the existing business case.

Conclusion

Using the described methodologies, possible outcomes of an investment will be assessed more realistically than is possible with simple static models. Executive decision-makers will have deeper understanding of the consequences of their decisions. The pragmatic approach will give them an easy-to-understand explanation of the numbers presented. This way, valuable time can be used for actual decision-making instead of alignment and communication of oversimplified assumptions.

In the asset evaluation example from the energy and utilities industry, the use of the method led to the following two outcomes:

* A switch from the originally planned investment decision to a more lucrative option, and

* A drastic reduction in the alignment effort between the supervisory and executive boards due to a common understanding of the business case.

The revision ultimately resulted in increased benefit of approximately 30% of the previously expected NPV.

Evgeni Kochman, business analyst, Fabian Sempf, consultant and Michael Kruse, partner, are with international management consultants Arthur D Little in Germany. 







Subscribe

Subscribe



https://www.rotork.com/?utm_source=engineer%20live&utm_medium=banner%20ad&utm_campaign=Engineer%20Live

Newsbrief

twitter facebook linkedin © Setform Limited