Researches on the Efficient Frontier of Mean-CVaR under Normal Distribution Condition

Abstract

In this paper, we present the mean- CVaR model under normal distribution condition on the basis of mean-variance model and obtain the calculation method of CVaR. Through comparing mean-CVaR model with mean-variance model, we find out that for any given confidence level, if the minimum CVaR portfolio exists, it lies above the minimum variance portfolio on the mean-variance efficient frontier and mean-CVaR method is more effective than mean-variance method as a risk management tool. An in-depth research on the efficient frontier of CVaR show that when the confidence level at which investors compute CVaR is not large enough, there would exist no mean-CVaR boundary and efficient frontier. Finally a calculation example show the result from using mean-CVaR portfolio model is more efficient than from using mean-variance portfolio model.