Vector risk functions (Q1762365)

This paper introduces a general framework for vector-valued risk functions using Banach lattices and Bochner integrable vector-valued random variables. Based on the properties of the risk function, various risk metrics such as deviations and expectation bounded coherent risk measures are introduced. The vector-valued risk function is different from the existing vector risk measures in that the former is not a set-valued function. The authors also investigate the relationships between scalar and vector risk functions and illustrate that the proposed general framework provides a way to integrate several scalar and vector risk functions. The representation theorems for deviations and expectation bounded coherent risk measures are established. These theorems are developed based on the duality of two functional spaces and some kind of envelope generated by the sub-gradients. Some examples of applications of the general results to dynamic risk measures and risk optimization are presented. The results may have potential applications to pricing and hedging as well as portfolio selection.