上财系列 金融风险控制与管理 QRM - Aggregate Risk.pptVIP

@Ri$kLab.CNTM Slide* VaR for a portfolio of defaultable bonds Assumption d = 100 defaultable corporate bonds, i.i.d. probability(Loss)=2%, pricet(bond)=100. Loss presentation Two Portfolios (current value with 10,000) Portfolio A – fully concentrated, 100 units of bond one Portfolio B – completely diversified, one unite of each of the bonds Question, which portfolio is more risky? @Ri$kLab.CNTM Slide* VaR for a portfolio of defaultable bonds - cnt VaR Analysis portfolio A Portfolio B Risk capital required for portfolio B is higher than for portfolio A @Ri$kLab.CNTM Slide* VaR for a portfolio of defaultable bonds - cnt Nonsensical results Portfolio A – even after a withdrawal of a risk capital 500 the profolio is still acceptable to a risk controller working with VaR at the 95% level Portfolio B – a bank would need an additional risk capita of 25 to satisfy a regulator working wth VaR at the 95% level. Cmp – the measuring risk with VaR can lead to nonsensical results, which show that VaR is not subadditive in general. the reason that the assets making up portfolio have very skewed distrbutions whatever there exists dependence structure. @Ri$kLab.CNTM Slide* VaR for Elliptical Risk Factors Theorem Suppose that X~Ed(μ,∑,Ψ)(椭圆分布),.i.e., and define the set M of linearized portfolio losses of the form then for any two losses L1, L2∈M and 0.5≤ a≤1, @Ri$kLab.CNTM Slide* Coherent Risk Measures based on Loss Distributions - Expected Shortfall Expected Shortfall (ES) is a coherent risk measure. Note that ES satisfies axioms Translation invariance Subadditivity Positive homogeneity Monotonicity @Ri$kLab.CNTM Slide* Coherent Risk Measures based on Loss Distributions – Coherent Premium Principle A Coherent Premium Principle, by Sischer, , “Risk Capital Allocation by Coherent Risk Measures Based on One-Sided Moments”, Insurance: Mathematics and Economics, 32:. 135-146 @Ri$kLab.CNTM Slide* Coherent Risk Measures as General