Research On Credit Portfolio Optimization Model Based On CVaR

Commercial bank credit risk is the main risk of commercial banks,the essence of banking crisis is commercial bank asset allocation errors,therefore the study of loan portfolio optimization is critical to the survival and development of commercial bank.In the theory of loan porfolio,the choice of risk quantification has been one problem should be solved,domestic and foreign scholars put forward a number of risk measurement methods for this problem.VaR has been widespread concern in recent years.While researchers found that VaR does not satisfy times additive,not necessarily satisfy convexity and it's under the assumption of normal distribution.VaR would be useless in some extreme cases.Therefore,this thesis choose CVaR as risk measurement,which makes up for the deficiencies of VaR.This thesis use minimum CVaR of loan portfolio as a object function,considering three friction factors including transaction costs,lower and upper bounded limitation and entropy constraint,different mean-CVaR loan portfolio optimization models are set up.The model is a convex programming problem will be solved by using the method of sequence planning and rotation algorithm.Finally,specific examples demonstrate the effectiveness of the model and algorithm.Three different explanations explain the different effects of different friction factor on the loan yield,it helps commercial bank consider different friction factors and put forward effective suggestions when choosing fortfolio.