Incorporating Risk and Uncertainty in Cost-Benefit Analysis

Cost-Benefit Analysis (CBA) is a tool for assessing the welfare effects of changes in regulatory and investment interventions. While in many ways an effective approach, a significant drawback of CBA, however, is that it relies on estimates for variables that cannot be predicted with complete accuracy. As such, expected outcomes generated by CBA, such as financial and economic net present values (NPVs), incorporate a degree of risk and uncertainty. It is therefore critical that CBA is based on transparent assumptions about the nature of risk and uncertainty affecting key variables: CBA cannot contribute to rational decision-making unless the distribution of outcomes is clear, and the effect on forecast reliability understood.Real-world risk and uncertainty generate numerous ex-ante outcomes at the point of appraisal. Correctly assessing risk and uncertainty is therefore one of the most difficult challenges decisionmakers face in applying the results of CBA. This report offers a systematic approach to the incorporation of risk and uncertainty in CBA. The primary objectives are to review the professional literature on risk and uncertainty; to provide a methodology for taking account of risk and uncertainty in CBA; and to suggest guidelines for the interpretation and application of CBA results in the decision-making process.The treatment of risk and uncertainty are clearly addressed in the CBA guidelines of most OECD countries, although approaches vary. The simplest procedures are based on sensitivity analysis, as applied to a deterministic base case. More comprehensive analysis is based on assumed probability distributions for the variables concerned. The CBA guidelines of multilateral financial institutions and a number of advanced economies (Australia, Canada, France, the UK, the US and the European Union) call for sensitivity analysis on a project-by-project basis, identifying specific long-term risks and uncertainties associated with the assumptions and values used in appraisal and evaluation.Still greater insight into the impact of risk and uncertainty on expected regulatory outcomes can be gained from a probabilistic modeling of variable distributions and their inter-dependencies. A Monte Carlo simulation is therefore recommended alongside sensitivity analysis, where data, time and budget permit