Risk preference modeling with conditional average: an application to portfolio optimization

Risk preference modeling with conditional average: an application to portfolio optimization

0.00 Avg rating0 Votes
Article ID: iaor200937793
Country: Germany
Volume: 165
Issue: 1
Start Page Number: 67
End Page Number: 95
Publication Date: Jan 2009
Journal: Annals of Operations Research
Authors:
Keywords: risk, programming: quadratic
Abstract:

The paper introduces a new risk measure called Conditional Average (CAVG), which was designed to cover typical attitudes towards risk for any type of distribution. It can be viewed as a generalization of Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR), two commonly used risk measures. The preference structure induced by CAVG has the interpretation in Yaari's dual theory of choice under risk and relates to Tversky and Kahneman's cumulative prospect theory. The measure is based on the new stochastic ordering called dual prospect stochastic dominance, which can be considered as a dual stochastic ordering to recently developed prospect stochastic dominance. In general, CAVG translates into a nonconvex quadratic programming problem, but in the case of a finite probability space it can also be expressed as a mixed-integer program. The paper also presents the results of computational studies designed to assess the preference modeling capabilities of the measure. The experimental analysis was performed on the asset allocation problem built on historical values of S&P 500 sub-industry indexes.

Reviews

Required fields are marked *. Your email address will not be published.