Multi-confidence-level Worst-case CVaR Based on the Mixture Distribution and Its Application.docVIP

Multi-confidence-level Worst-case CVaR Based on the Mixture Distribution and Its Application 　　Abstract. The recent financial crisis has emphasised the significance for risk management. In this paper we consider the robust risk management in multi-confidence levels. Under the assumption of mixture distribution of random variables, this paper establishes optimization model of multi-confidence-level worst-case Conditional Value-at-Risk and uses multi-objective decision to transform it into linear program. As an application, Monte Carlo Simulation is used to illustrate the proposed approach. 　　Key words: Multi-confidence-level, Mixture distribution, Monte Carlo Simulation. 　　1. Introduction 　　The expected return and risk are the two core issues of risk assets investment. How to measure the benefits and risks of portfolio investment and how to balance the two indicators of asset allocation are the two problems investors should solve in the first place. Based on these issues, Markowitz put forward to the Mean-Variance Model in 1952, which is groundbreaking of the modern portfolio theory. 　　As extension of Mean-Variance Model and risk management, Value-at-Risk (VaR) is put forward by J.P. Morgan in the early 1990s. Following the recommendations of the Basel Committee on Banking Supervision (2009)[1], many financial institutions have flexibility over their choice of model for estimating VaR (Degiannakis et al., 2013) [2]. However, VaR has been criticized in the later researches. Classical formulations of the portfolio optimization problem with VaR approaches could result in a portfolio extremely sensitive to errors in the data, such as mean and covariance matrix of the returns (Ghaoui et al., 2003) [3]. Moreover, VaR fails to be coherent (Artzner et al., 1999)[4] and suffers from being unstable and difficult to work with numerically when losses are not normally distributed (Rockafellar and Uryasev, 2000) [5]. 　　The development of coherent risk measuring gives ris