6.5.2 Advanced probability valuation techniques

Statistical techniques can be used to estimate the probability of risk by constructing probability distributions and interpreting the resulting outputs. These distributions are based on professional experience, supported where available by historical information and reliable assumptions for similar recent projects. Once these distributions have been calculated, a reliable estimate of probability can then be made to a given level of accuracy (known as the confidence interval).

Statistical risk measures have the advantage that they are based on rigorous economic principles, use a mix of professional experience and available information, and map a variety of possible outcomes. The accuracy and reliability of probability distribution estimates depends on the capability to provide reasonable forecasts of likely outcomes, supported by the quality of available information.

Instead of estimating each risk and its components separately, it may be possible to calculate a single risk measure through multivariable analysis and simulation. These techniques typically involve the use of computer-based simulation packages.

One accepted method of multivariable analysis is Monte Carlo simulation. This technique constructs an artificial probability distribution for total risk, or a subset of risks, based on assumed, or actual distributions for each of the individual risks. It then provides a single value for risk by simultaneously solving a number of different risk relationships.

In order for a meaningful Monte Carlo simulation to be performed, a sufficient data set should be available to allow assumptions to be made about the distribution of each risk variable. This may be possible when sufficient information exists to allow the construction of a multivariable equation, or through the engaging of a technical expert with experience in similar projects.

Where advanced probability valuation techniques and Monte Carlo, or other simulation techniques are used, it is generally helpful to employ technical experts, or external advisers with particular expertise to determine appropriate probability distributions, provide reliable probability estimates and perform the probability analysis and econometric assessment of the results.