Incorporating Asymmetric Distributional Information in Robust Value-at-Risk Optimization

Authors:
Karthik Natarajan;Dessislava Pachamanova;Melvyn Sim
Affiliations:
Singapore-MIT Alliance and Department of Mathematics, National University of Singapore, Singapore 117543;Mathematics and Sciences Division, Babson College, Babson Park, Massachusetts 02457;Singapore-MIT Alliance and NUS Business School, National University of Singapore, Singapore 117592
Venue:
Management Science
Year:
2008

Citing 0
Cited 9

Goal-Driven Optimization

Operations Research
α-Conservative approximation for probabilistically constrained convex programs

Computational Optimization and Applications
Robust Approximation to Multiperiod Inventory Management

Operations Research
Distributionally Robust Optimization and Its Tractable Approximations

Operations Research
Tight Bounds for Some Risk Measures, with Applications to Robust Portfolio Selection

Operations Research
A smoothing sample average approximation method for stochastic optimization problems with CVaR risk measure

Computational Optimization and Applications
Value-at-Risk optimization using the difference of convex algorithm

OR Spectrum
Worst-Case Value at Risk of Nonlinear Portfolios

Management Science
Robust investment decisions under supply disruption in petroleum markets

Computers and Operations Research

Quantified Score

Hi-index	0.01

Visualization

Abstract

Value-at-Risk (VaR) is one of the most widely accepted risk measures in the financial and insurance industries, yet efficient optimization of VaR remains a very difficult problem. We propose a computationally tractable approximation method for minimizing the VaR of a portfolio based on robust optimization techniques. The method results in the optimization of a modified VaR measure, Asymmetry-Robust VaR (ARVaR), that takes into consideration asymmetries in the distributions of returns and is coherent, which makes it desirable from a financial theory perspective. We show that ARVaR approximates the Conditional VaR of the portfolio as well. Numerical experiments with simulated and real market data indicate that the proposed approach results in lower realized portfolio VaR, better efficient frontier, and lower maximum realized portfolio loss than alternative approaches for quantile-based portfolio risk minimization.