Approximating Risk Aversion in Decision Analysis Applications

This paper investigates the impact of risk aversion in decision analyses under uncertainty with a single evaluation measure and presents a simple procedure for approximately addressing risk aversion in a way that is defensible for many decisions. Specifically, a simulation study is presented that leads to guidelines for determining when an expected utility analysis should be conducted for a decision, rather than simply an expected value analysis, and what form of utility function should be used for this expected utility analysis. The simulation study shows that a sensitivity analysis using an exponential utility function should be conducted for most decision analyses, but that this sensitivity analysis can often establish, without requiring utility information from the decision maker , that no further utility analysis is required. In addition, when further utility analysis is required, the simulation study shows that this can be done in a simple way using an exponential utility function that will be accurate for many decision analyses. However, in situations where there is equal or greater downside risk than upside potential, a more detailed study of the decision maker's utility function may be necessary.