6.6  Sensitivity analysis

Sensitivity analysis should be performed on key cash flows and assumptions to determine the robustness of the PSC to potential changes in assumptions, risk components and the forecast operating environment over the term of the project. Sensitivity analysis can be used for the following purposes:

•  comparison with bids to identify the changes in base assumptions which would result in a different evaluation decision being reached; and

•  determine the relative robustness of the PSC to bids. This may be assessed as a qualitative factor if the PSC is close to the lowest bid.

Assumptions across a variety of key variables should be examined as part of a sensitivity analysis. The procurement team should also consider whether, for example, an increase in a base cost assumption (in the Raw PSC) would lower an associated risk. For example, if a sensitivity analysis considered the effect of an additional 10 per cent increase in the base capital costs of a project (reflected in the Raw PSC), the value of design and construction risk may be lower.

Sensitivities can be performed by varying individual assumptions, or by considering simultaneous changes in a number of variables. This allows both the impact of key factors to be considered, as well as examining a range of realistic scenarios where there is considerable interaction between variables. This may already be implicitly considered in the PSC where Monte Carlo simulation has been used to value risk. As a general rule, the amount of sensitivity analysis performed should reflect the materiality of key variables, the complexity of the PSC and the proximity of the PSC to the lowest cost bid.

Variables that are typically analysed using sensitivity analysis include:

•  length of the project (both the construction and concession periods);

•  periodic inflation rate;

•  construction costs, schedule and completion dates (both in the Raw PSC and the pricing of risk);

•  total service demand;

•  total operating costs;

•  third-party revenue; and

•  residual value.

Where possible, the financial model should be developed to allow different values for key variables over time.