Simulation-based valuation of project finance - Does model complexity really matter?

paper analyzes the impact of model complexity on the valuation of project financed investments along two dimensions: The NPV distribution and the expected default probability of the project. Model complexity is separated in forecast and simulation complexity. Thereby, we aim to identify elements of model complexity that are crucial for the model accuracy in a project finance context.For this purpose we develop a project finance valuation tool, which generates cash flow distributions on the basis of a Monte Carlo simulation. The analysis reveals, among others, the following results: First, quasi-market valuation should be used for determining the leverage adjusted cost of equity. Second, we demonstrate that 100,000 iterations in the Monte Carlo simulation are sufficient to obtain stable results. Third, time resolution in the simulation procedure has a significant impact on the obtained default probability as well as on the NPV distribution. Fourth, the choice of the volatility forecasting model significantly affects the simulation results. Fifth, we do not find a strong impact of sophisticated correlation forecasting techniques. Interestingly, cash flow distributions are quite similar regardless whether we use historical correlation or forecast correlation based on a dynamic conditional correlation (DCC) model.