In this talk, I will describe the conditional value at risk (CVaR) measure used in modelling risk aversion in decision making problems.

CVaR is a highly consistent risk measure for modelling risk aversion.

I will then present two applications of CVaR. The first application considers all problems that are representable by decision trees. In this application, I show that these problems under the CVaR criterion can be solved efficiently by solving a linear program. In the second application, I consider a basic problem in the area of production planning with random yield. For this problem, I present a risk aversion model. The model is nonconvex. I present an efficient locally optimal solution method and then provide a sufficient optimality condition.