Estimating State-Dependent Volatility of Investment Projects: A Simulation Approach

Project volatility is an essential parameter for real options analysis, and it may also be useful for risk analysis. Many volatility estimation procedures only consider the volatility in the first year of the project. Others consider that different years may have different values of the project volatility. In this paper I show that volatility may change not only with time but also with the state of the project. I consider two possible definitions for the project volatility, the log-variance and the variance of the project value, and I propose three procedures for estimating state-dependent volatility: two-level simulation, one and a half level simulation and a regression procedure. Computational experiments show that the one and a half level simulation procedure and the regression procedure lead to the most accurate estimations of project volatility.