Sensitivity analysis refers to a method whereby the impact of changing one or more independent variables on the outcome variable in a model is determined. It involves varying the independent variables while keeping other variables unchanged to see how changes in the variables of interest affect the outcome. Sensitivity analysis is considered a “what-if” analysis and is very useful in determining the main variables that will influence project outcomes.
For mining financial models, sensitivity analysis helps to test the viability of projects by analyzing factors such as commodity prices, ore grades, recovery rates, OPEX, CAPEX, and discount rates. In a base case scenario for a mining project, one may have assumed a gold price of $2,000/oz that will give an NPV of $500 million. The model will be analysed for different levels of changes in the gold price, such as 20-30%, to determine the break-even point.
The first step is developing a deterministic base case model using production schedules, income statements, cost statements, and cash flow statements. The analysts choose 5-8 most crucial variables depending on their uncertainty and impact. They set realistic ranges for variables, such as ore grade ±10%, fuel costs +20-50%. Each variable is changed one by one to see the effects on NPV/IRR (one-way sensitivity). The results can be shown using Excel data tables or tornado charts that illustrate how the NPV/IRR varies depending on certain factors.
For stress testing, analysts develop extreme cases to assess how the model reacts to market crashes or operational problems. This analysis confirms whether the model works logically, such as lower CAPEX leading to lower cash flows. In mining operations, analysts may reduce recovery rates by 15% or increase OPEX in times of high inflation. These changes help determine whether the IRR is still above the hurdle rate, which could be 15%.
Further to univariate analysis, some complex applications apply scenario analysis involving multiple variables simultaneously or probabilistic approaches where there is application of probability distributions for Monte Carlo simulations, providing a measure of probability of failure of NPV (e.g., 20%). Tornado diagrams show order of sensitivities (price, grade is most sensitive) enabling stakeholders to concentrate on key areas.
In summary, sensitivity analysis helps improve mining project resiliency through testing the limits of assumptions and thus help in negotiations of contracts or investments. Shortcomings include the assumption of independence between variables and failure to consider correlation (e.g., correlation between price and grade) and thus is always complemented by scenario analysis. Sensitivity analysis is critical in the context of Africa, especially Cameroon, where geopolitical and foreign exchange risks need to be considered.
