The Expected Shortfall or Conditional Value-at-Risk (CVaR) has been playing the role of main risk
measure in the recent years and paving the way for an enormous number of applications in risk management
due to its very intuitive form and important coherence properties. This work aims to explore this measure as a
probability-dependent utility functional, introducing an alternative view point for its Choquet Expected Utility
representation. Within this point of view, its main preference properties will be characterized and its utility
representation provided through local utilities with an explicit dependence on the assessed revenue’s
distribution (quantile) function. Then, an intuitive interpretation for the related probability dependence and the
piecewise form of such utility will be introduced on an investment pricing context, in which a CVaR maximizer
agent will behave in a relativistic way based on his previous estimates of the probability function. Finally, such
functional will be extended to incorporate a larger range of risk-averse attitudes and its main properties and
implications will be illustrated through examples, such as the so-called Allais Paradox.
|