Valuing infrastructure investments as portfolios of interdependent real options

Abstract

The value of infrastructure investments is frequently influenced by enormous uncertainty surrounding both exogenous and endogenous factors. At the same time, however, their value is generally driven by much flexibility - i.e. options - with respect to design, financing, construction and operation. Real options analysis aims to pro-actively manage risks by valuing the flexibilities inherent in uncertain investments. Although real options generally occur within portfolios whose value is affected by both exogenous and endogenous uncertainty, most existing valuation approaches focus on single (i.e. individual) options and consider only exogenous uncertainty. In this thesis, we introduce an approach for modelling and approximating the value of portfolios of interdependent real options under exogenous uncertainty, using both influence diagrams and simulation-and-regression. The key features of this approach are that it translates the interdependencies between real options into linear constraints and then integrates these in a portfolio optimisation problem, formulated as a multi-stage stochastic integer programme. To approximate the value of this optimisation problem we present a transparent valuation algorithm based on simulation and parametric regression that explicitly takes into account the state variable's multidimensional resource component. We operationalise this approach using three numerical examples of increasing complexity: an American put option in a simple single-factor setting; a natural resource investment with a switching option in a one-factor setting; and the same investment in a three-factor setting. Subsequently, we demonstrate the ability of the proposed approach to evaluate a complex natural resource investment that features both a large portfolio of interdependent real options and four underlying uncertainties. We show how our approach can be used to investigate the way in which the value of that portfolio and its individual real options are affected by the underlying operating margin and the degrees of different uncertainties. Lastly, we extend this approach to include endogenous, decision- and state-dependent uncertainties. We present an efficient valuation algorithm that is more transparent than those used in existing approaches; by exploiting the problem structure it explicitly accounts for the path dependencies of the state variables. The applicability of the extended approach to complex investment projects is illustrated by valuing an urban infrastructure investment. We show the way in which the optimal value of the portfolio and its single, well-defined options are affected by the initial operating revenues, and by the degrees of exogenous and endogenous uncertainty.