Abstract
Risk measures for tail risk have an important application in the dynamic portfolio insurance strategies. We propose a new risk measure called SlideVaR which overcome the limitation of traditional measures like VaR and ES, and can sufficiently reflect the market changes. Several important properties of SlideVaR and its generalized risk measure have been investigated. Then, we further apply SlideVaR into constructing dynamic portfolio insurance strategy. Our numerical analysis shows that SlideVaR-based portfolio insurance strategy has advantage especially in markets where the state changes frequently.
Similar content being viewed by others
Notes
Distortion function \(g:[0,1] \rightarrow [0,1]\) is the non-decreasing function such that: \(g(0) = 0\), \(g(1) = 1\). Given a distortion function g, the distortion risk measure \(\rho _{g}(X)\) is:
$$\begin{aligned} \rho _{g}(X) = \int _{-\infty }^{0}[g(P[X> x])-1]dx + \int _{0}^{+\infty }g(P[X > x])dx. \end{aligned}$$(6)More information about distortion risk measure can be found in Appendix 3.
Cotter and Dowd (2006) suggest that spectral risk measures can be used by futures clearinghouses to set margin requirements that reflect their corporate risk aversion; Sriboonchitta et al. (2010) and Wächter and Mazzoni (2013) study the relationship between utility function and spectral risk measures; Dowd and Cotter (2007) discuss the relationship between the choice of risk aversion function and spectral risk measures.
It is difficult to get the analytical closed-form expressions of \(U_{\beta }^{\phi }(X)\) under continuous functions such as exponential function and power function. From Remark 2 we know that, with the appropriate value of N, the tail thickness \(U_{\beta ,N}^{\phi }(X)\) under \(\phi ^s_{\beta ,N}(p)\) can be regarded as an approximation.
See e.g. Norton et al. (2021) for the detailed calculation.
The market tendency is mainly controlled by the AR model in process (35). For bull state with a mean return of \(10\%\), the parameters are \(\phi _{0}= 3.57\times 10^{-4}\) and \(\phi _{1} = 0.10\); and for bear state with a mean return of \(3\%\) the parameters are \(\phi _{0}= 1.07\times 10^{-4}\) and \(\phi _{1} = 0.10\). For the process (36), the degree of volatility cluster is decided by \(\gamma _1 = 0.80\) and \(\gamma _2 = 0.10\). The parameter \(\gamma _0\) of three scenarios is \(\gamma _0 = 3\times 10^{-5}\) for the high volatility scenario, \(\gamma _0 = 1.50\times 10^{-5}\) for the middle volatility scenario, and \(\gamma _0 = 0.50\times 10^{-5}\) for the low volatility scenario.
As introduced, the parameters for bull state are \(\phi _{0}= 3.57\times 10^{-4}\) and \(\phi _{1} = 0.10\); and those for bear state are \(\phi _{0}= 1.07\times 10^{-4}\), and \(\phi _{1} = 0.10\). For the volatility parameter, the parameters are \(\sigma ^2=3.36\times 10^{-4}\) for the high volatility scenario, \(\sigma ^2=1.57\times 10^{-4}\) for the middle volatility scenario, and \(\sigma ^2=3.93\times 10^{-5}\) for the low volatility scenario.
Specifically, we calculate the Protection Ratio as the ratio of the number of simulated paths with the final portfolio value greater than or equal to \(99.99\%\) of the initial value in the total number of simulations.
The results reported later are denoted with * for the probability level 0.1, ** for level 0.05, and *** for level 0.01.
In these two tables, SlideVaR-, VaR-, ES- and GlueVaR-based DPPI strategies are denoted as SlideVaR-PI, VaR-PI, ES-PI and GlueVaR-PI.
References
Acerbi C (2002) Spectral measures of risk: A coherent representation of subjective risk aversion. J Bank Finan 26(7):1505–1518
Adam A, Houkari M, Laurent JP (2008) Spectral risk measures and portfolio selection. J Bank Finan 32(9):1870–1882
Alexander C, Sarabia JM (2012) Quantile uncertainty and value-at-risk model risk. Risk Analysis: An Int J 32(8):1293–1308
Ameur HB, Prigent J (2007) Portfolio insurance: determination of a dynamic cppi multiple as function of state variables. Thema University of Cergy, Working paper
Ameur HB, Prigent JL (2014) Portfolio insurance: Gap risk under conditional multiples. Eur J Oper Res 236(1):238–253
Artzner P, Delbaen F, Eber JM, Heath D (1999) Coherent measures of risk. Math Financ 9(3):203–228
Balbás A, Garrido J, Mayoral S (2009) Properties of Distortion Risk Measures. Methodol Comput Appl Probab 11(3):385–399
Balder S, Brandl M, Mahayni A (2009) Effectiveness of cppi strategies under discrete-time trading. J Econ Dyn Control 33(1):204–220
Belles-Sampera J, Merigó JM, Guillén M, Santolino M (2013) The connection between distortion risk measures and ordered weighted averaging operators. Insurance: Mathematics and Economics 52(2):411–420
Belles-Sampera J, Guillén M, Santolino M (2014) Beyond Value-at-Risk: GlueVaR Distortion Risk Measures. Risk Anal 34(1):121–134
Benartzi S, Thaler RH (1995) Myopic loss aversion and the equity premium puzzle. Q J Econ 110(1):73–92
Benninga S (1990) Comparing portfolio insurance strategies. Finanzmarkt und Portfolio Management 4(1):20–30
Bird R, Dennis D, Tippett M (1988) A stop loss approach to portfolio insurance. J Portf Manag 15(1):35
Black F, Jones RW (1987) Simplifying portfolio insurance. J Port Manag 14(1):48–51
Black F, Perold A (1992) Theory of constant proportion portfolio insurance. J Econ Dyn Control 16(3–4):403–426
Cai J, Wang Y, Mao T (2017) Tail subadditivity of distortion risk measures and multivariate tail distortion risk measures. Insurance: Mathematics and Economics 75:105–116
Cerreia-Vioglio S, Maccheroni F, Marinacci M, Montrucchio L (2011) Risk measures: rationality and diversification. Math Financ 21(4):743–774
Chen JS, Chang CL, Hou JL, Lin YT (2008) Dynamic proportion portfolio insurance using genetic programming with principal component analysis. Expert Syst Appl 35(1):273–278
Chen Z, Chen B (2020) Dhaene J (2020) Fair dynamic valuation of insurance liabilities: a loss averse convex hedging approach. Scand Actuar J 9:792–818
Chen Z, Chen B, Dhaene J, Yang T (2021a) Fair dynamic valuation of insurance liabilities via convex hedging. Insurance: Mathematics and Economics 98:1–13
Chen Z, Chen B, Hu Y, Zhang H (2021b) Hedge inflation risk of specific purpose guarantee funds. In: European Financial Management, forthcoming
Choquet G (1954) Theory of capacities. Annales de l’Institut Fourier 5:131–295
Coleman T, Kim Y, Li Y, Patron M (2007) Robustly hedging variable annuities with guarantees under jump and volatility risks. J Risk Insurance 74(2):347–376
Cotter J, Dowd K (2006) Extreme spectral risk measures: an application to futures clearinghouse margin requirements. J Bank Finan 30(12):3469–3485
Cox LA Jr (2012) Confronting deep uncertainties in risk analysis. Risk Analysis: An Int J 32(10):1607–1629
Darkiewicz G, Dhaene J, Goovaerts M (2003) Coherent distortion risk measure: a pitfall. In: Working paper presented to 2003 IME conference
Denuit M, Dhaene J, Goovaerts M, Kaas R (2006) Actuarial theory for dependent risks: measures, orders and models. John Wiley & Sons
Dhaene J, Denuit M, Goovaerts MJ, Kaas R, Vyncke D (2002a) The concept of comonotonicity in actuarial science and finance: applications. Insurance: Mathematics and Economics 31(2):133–161
Dhaene J, Denuit M, Goovaerts MJ, Kaas R, Vyncke D (2002b) The concept of comonotonicity in actuarial science and finance: theory. Insurance: Mathematics and Economics 31(1):3–33
Dhaene J, Vanduffel S, Goovaerts MJ, Kaas R, Tang Q, Vyncke D (2006) Risk measures and comonotonicity: a review. Stoch Model 22(4):573–606
Dhaene J, Laeven RJA, Vanduffel S, Darkiewicz G, Goovaerts MJ (2008) Can a coherent risk measure be too subadditive? J Risk Insurance 75(2):365–386
Dhaene J, Kukush A, Linders D, Tang Q (2012) Remarks on quantiles and distortion risk measures. Eur Actuar J 2(2):319–328
Dhaene J, Kukush A, Linders D (2020) Comonotonic asset prices in arbitrage-free markets. J Comput Appl Math 364:112310
Dichtl H, Drobetz W (2011) Portfolio insurance and prospect theory investors: Popularity and optimal design of capital protected financial products. Journal of Banking & Finance 35(7):1683–1697
Dimson E, Marsh P, Staunton M (2008) The worldwide equity premium: a smaller puzzle. In: Handbook of the equity risk premium, Elsevier, pp 467–514
Dowd K, Cotter J (2007) Spectral risk measures and the choice of risk aversion function. Tech Rep. Working Paper
Drapeau S, Kupper M (2013) Risk preferences and their robust representation. Math Oper Res 38(1):28–62
Embrechts P, Nešlehová J, Wüthrich MV (2009) Additivity properties for value-at-risk under archimedean dependence and heavy-tailedness. Insurance: Mathematics and Economics 44(2):164–169
Estep T, Kritzman M (1988) Tipp: Insurance without complexity. J Port Manag 14(4):38–42
Feng R (2018) An introduction to computational risk management of equity-linked insurance
Feng R, Shimizu Y (2016) Applications of central limit theorems for equity-linked insurance. Insurance: Mathematics and Economics 69:138–148
Feng R, Vecer J (2017) Risk based capital for guaranteed minimum withdrawal benefit. Quantitative Finance 17(3):471–478
Feng R, Volkmer HW (2012) Analytical calculation of risk measures for variable annuity guaranteed benefits. Insurance: Mathematics and Economics 51(3):636–648
Feng R, Volkmer HW (2016) An identity of hitting times and its application to the valuation of guaranteed minimum withdrawal benefit. Math Financ Econ 10(2):127–149
Feng R, Volkmer HW et al (2014) Spectral methods for the calculation of risk measures for variable annuity guaranteed benefits. Astin Bull 44(3):653–681
Figlewski S, Chidambaran N, Kaplan S (1993) Evaluating the performance of the protective put strategy. Financial Analysts Journal pp 46–69
Föllmer H, Schied A (2002) Convex measures of risk and trading constraints. Finance Stochast 6(4):429–447
Frittelli M, Gianin ER (2002) Putting order in risk measures. J Bank Fin 26(7):1473–1486
Frittelli M, Maggis M, Peri I (2014) Risk Measures on \(\cal{P}\left(\mathbb{R}\right)\)) and Value-at-Risk with Probability/Loss Function. Math Financ 24(3):442–463
Grabisch M, Marichal JL, Mesiar R, Pap E (2011) Aggregation functions: Construction methods, conjunctive, disjunctive and mixed classes. Inf Sci 181(1):23–43
Hamidi B, Maillet B, Prigent J (2008) A time-varying proportion portfolio insurance strategy based on a caviar approach. Tech. rep., Working paper, University of Cergy-THEMA
Hamidi B, Maillet BB, Prigent JL (2009) A risk management approach for portfolio insurance strategies. In: Proceedings of the 1st EIF International Financial Research Forum, Economica
Hamidi B, Maillet B, Prigent JL (2014) A dynamic autoregressive expectile for time-invariant portfolio protection strategies. J Econ Dyn Control 46:1–29
Hanbali H, Linders D, Dhaene J (2021) Value-at-risk, tail value-at-risk and upper tail transform of the sum of two counter-monotonic random variables
Happersberger D, Lohre H, Nolte I (2020) Estimating portfolio risk for tail risk protection strategies. Euro Fin Manag. https://doi.org/10.1111/eufm.12256
Hardy M (2003) Investment guarantees: modeling and risk management for equity-linked life insurance, vol 215. John Wiley & Sons
Hull JC (2003) Options futures and other derivatives. Pearson Education India
Jiang C, Ma Y, An Y (2009) The effectiveness of the var-based portfolio insurance strategy: An empirical analysis. Int Rev Financ Anal 18(4):185–197
Kato T (2015) Vwap execution as an optimal strategy. JSIAM Letters 7:33–36
Konishi H (2002) Optimal slice of a vwap trade. J Fin Markets 5(2):197–221
Lee HI, Chiang MH, Hsu H (2008) A new choice of dynamic asset management: the variable proportion portfolio insurance. Appl Econ 40(16):2135–2146
Mitchell D, Białkowski J, Tompaidis S (2020) Volume-weighted average price tracking: A theoretical and empirical study. IISE Transactions 52(8):864–889
Norton M, Khokhlov V, Uryasev S (2021) Calculating cvar and bpoe for common probability distributions with application to portfolio optimization and density estimation. Ann Oper Res 299(1):1281–1315
Pérignon C, Smith DR (2010) Diversification and Value-at-Risk. J Bank Fin 34(1):55–66
Qian PY, Wang ZZ, Wen ZW (2019) A composite risk measure framework for decision making under uncertainty. J Oper Res Soc China 7(1):43–68
Righi MB (2018) A composition between risk and deviation measures. Ann Oper Res pp 1–15
Rubinstein M, Leland HE (1981) Replicating options with positions in stock and cash. Financ Anal J 37(4):63–72
Song Y, Yan JA (2009) Risk measures with comonotonic subadditivity or convexity and respecting stochastic orders. Insurance: Mathematics and Economics 45(3):459–465
Sriboonchitta S, Nguyen HT, Kreinovich V (2010) How to relate spectral risk measures and utilities. Int J Intell Technol Appl Stat 3(2):141–158
Wächter HP, Mazzoni T (2013) Consistent modeling of risk averse behavior with spectral risk measures. Eur J Oper Res 229(2):487–495
Wang S (1995) Insurance Pricing and Increased Limits Ratemaking by Proportional Hazards Transforms. Insurance: Mathematics and Economics 17(1):43–54
Wang S (1998) An actuarial index of the right-tail risk. North American Actuarial J 2(2):88–101
Wang SS, Young VR, Panjer HH (1997) Axiomatic characterization of insurance prices. Insurance: Mathematics and Economics 21(2):173–183
Wirch JL, Hardy MR (2001) Distortion risk measures: Coherence and stochastic dominance. In: International congress on insurance: Mathematics and economics, pp 15–17
Yin C, Zhu D (2015) New class of distortion risk measures and their tail asymptotics with emphasis on var. arXiv preprint arXiv:150308586
Acknowledgements
This work is financially supported by National Natural Science Foundation of China (Grant No. 72101256), MOE (Ministry of Education in China) Project of Humanities and Social Sciences (Project No.21YJC790016), National Key R&D Program of China (Grant No. 2018YFA0703900), National Natural Science Foundation of China (Grant Nos. 11871309 and 11371226), and National Social Science Fund of China (Grant No. 21AZD028). This work is also supported by Public Computing Cloud Platform, Renmin University of China.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Preliminaries of Risk Measures
1.1 Coherent Risk Measure
Coherent risk measure was proposed by Artzner et al. (1999). Define a probability space \((\Omega , \mathcal {F},\mathbb {P})\), denote \(\mathcal {X}\) as the set of loss random variables defined on \((\Omega , \mathcal {F},\mathbb {P})\), then
Definition 3
(Coherent) A risk measure satisfying the four axioms of translation invariance, sub-additivity, positive homogeneity and monotonicity, is called coherent risk measure.
-
(a)
Translation invariance:
$$\begin{aligned} \rho (X + a)=\rho (X)+a, \,\, \forall X \in \mathcal {X}, \,\, \forall a \in \mathbb {R}. \end{aligned}$$(37) -
(b)
Sub-additivity:
$$\begin{aligned} \rho (X_1+X_2) \leqslant \rho (X_1)+\rho (X_2), \,\, \forall X_1, \,\, X_2 \in \mathcal {X}. \end{aligned}$$(38) -
(c)
Positive homogeneity:
$$\begin{aligned} \rho (\lambda X)=\lambda \rho (X), \,\, \forall \lambda \geqslant 0, \,\, \forall X \in \mathcal {X}. \end{aligned}$$(39) -
(d)
Monotonicity:
$$\begin{aligned} \text{ if } X \leqslant Y, \text{ then } \rho (X) \leqslant \rho (Y), \,\, \forall X, \,\, Y \in \mathcal {X}. \end{aligned}$$(40)
1.2 Spectral Risk Measure
As we mentioned in Sect. 3.1, \(U_{\beta }^{\phi }\) is a spectral risk measure and \(\phi _{\beta }\) is the corresponding risk aversion function. Spectral risk measure was proposed by Acerbi (2002). First, Acerbi (2002) define the risk aversion function \(\phi\) in the normed space \(L^1([0,1])\) where every element is represented by a class of functions which differ at most on a subset of [0, 1] of zero measure. The norm in given by
Second, let \(\phi\) satisfies
-
Monotonicity: \(\phi\) is decreasing if \(\forall q \in (a,b)\) and \(\forall \varepsilon >0\) such that \([q-\varepsilon ,q+\varepsilon ] \subset [a,b]\),
$$\begin{aligned} \int _{q-\varepsilon }^{q}\phi (p)dp \geqslant \int _{q}^{q+\varepsilon }\phi (p)dp. \end{aligned}$$(42)\(\phi\) is increasing if \(\forall q \in (a,b)\) and \(\forall \varepsilon >0\) such that \([q-\varepsilon ,q+\varepsilon ] \subset [a,b]\),
$$\begin{aligned} \int _{q-\varepsilon }^{q}\phi (p)dp \leqslant \int _{q}^{q+\varepsilon }\phi (p)dp. \end{aligned}$$(43) -
Positivity: \(\phi\) is positive if \(\forall I \subset [a,b]\),
$$\begin{aligned} \int _I\phi (p)dp \geqslant 0. \end{aligned}$$(44)
Then, Acerbi (2002) give the following,
Definition 4
(Spectral risk measure) If a risk aversion function \(\phi \in L^1([0,1])\) satisfies: (1) \(\phi\) is positive, (2) \(\phi\) is increasing, (3) \(|| \phi ||=1\). Then we call the function
is the spectral risk measure generated by \(\phi\).
By some examples, Acerbi (2002) and Adam et al. (2008) show that, in a real-world risk management application the integral 45 will always be well defined and finite. For instance, if \(\phi (p)\) is bounded and X is a finite integrable random variable, then \(U_{\beta }^{\phi }\) is well-defined and finite.
1.3 Distortion Risk Measures
Distortion risk measures was proposed by Wang et al. (1997). Distortion function \(g:[0,1] \rightarrow [0,1]\) is the non-decreasing function such that: \(g(0) = 0\), \(g(1) = 1\).
Definition 5
Given a distortion function g, the distortion risk measure \(\rho _{g}(X)\) is:
Distortion risk measure is a general framework that can be expressed as a special Chouqet Integral (Choquet 1954), and a distortion risk measure is coherent if and only if the distortion function is concave (Balbás et al. 2009; Wirch and Hardy 2001; Dhaene et al. 2012). Several widely used risk measures are special cases of \(\rho _{g}(X)\). For example, the distortion function of \(VaR_{\alpha }\) is
The distortion function of \(ES_{\alpha }\) is:
Proof of Propositions and Corollaries
1.1 Proof of Proposition 1
Proof
Due to that \(U_{\beta }^{\phi }(X)\) is a spectral risk measure and S(x) is a non-decreasing function, if \(X \leqslant _{st} Y\), we have
Then we can have
Due to that \(ES_{\alpha }(Y) \geqslant VaR_{\beta }(Y)\) and \(VaR_{\beta }(Y) \geqslant VaR_{\beta }(X)\), we can have
1.2 Proof of Proposition 2
Proof
\(\forall X \in \tilde{\mathcal {X}}\), we have \(SlideVaR_{\alpha ,\beta }^{\phi }(X) = ES_{\alpha }(X)\). Therefore, \(\forall X,Y \in \tilde{\mathcal {X}}\), we have
Then (a) is proved. \(\forall X^{*}\in \mathcal {X}^{*}\), if \(X^{*} \leqslant _{st} X\) (and Y), we have
that is
According to (a), (b) can be proved.
1.3 Proof of Corollary 2
Proof
\(\forall X, Y\) which satisfy the conditions in Proposition 2, we have \(SlideVaR_{\alpha ,\beta }^{\phi }(X) = ES_{\alpha }(X)\), \(SlideVaR_{\alpha ,\beta }^{\phi }(Y) = ES_{\alpha }(Y)\). Then \(X \leqslant _{icx} Y\) implies that \(SlideVaR_{\alpha ,\beta }^{\phi }(X) \leqslant SlideVaR_{\alpha ,\beta }^{\phi }(Y)\).
1.4 Proof of Corollary 1
Proof
\(\forall X\in \tilde{\mathcal {X}}\), if \(X \leqslant _{st} Y\), we have
which implies that \(S\circ U_{\beta }^{\phi }(Y)= 1\) and then \(Y \in \tilde{\mathcal {X}}\).
1.5 Proof of Proposition 3
Proof
\(\forall a \in \mathcal {R}\), We have
If \(a \geqslant 0\), we have \(S\circ (U_{\beta }^{\phi }(X)+a) \geqslant S\circ (U_{\beta }^{\phi }(X))\), then
The equal sign holds when \(a=0\). Similarly, if \(a \leqslant 0\), we have \(S\circ (U_{\beta }^{\phi }(X)+a) \leqslant S\circ U_{\beta }^{\phi }(X)\), then
The equal sign holds when \(a=0\).
1.6 Proof of Proposition 4
Proof
Due to that
we can have
From \(U_{\beta }^{\phi }(X)>0\), we know that, if \(\lambda \geqslant 1\), we have \(S\circ ( \lambda U_{\beta }^{\phi }( X))- S\circ U_{\beta }^{\phi }( X) \geqslant 0\) which implies that \(SlideVaR_{\alpha ,\beta }^{\phi }(\lambda X) - \lambda SlideVaR_{\alpha ,\beta }^{\phi }( X) \geqslant 0\). The equal sign holds when \(\lambda =1\). Similarly, if \(\lambda \leqslant 1\), we have \(S\circ ( \lambda U_{\beta }^{\phi }( X))- S\circ U_{\beta }^{\phi }( X) \leqslant 0\) which implies that \(SlideVaR_{\alpha ,\beta }^{\phi }(\lambda X) - \lambda SlideVaR_{\alpha ,\beta }^{\phi }( X) \leqslant 0\). The equal sign holds when \(\lambda =1\).
1.7 Proof of Proposition 5
Proof
\(\forall X \in \tilde{\mathcal {X}}\), we have \(SlideVaR_{\alpha ,\beta }^{\phi }(X) = ES_{\alpha }(X)\). Therefore, \(\forall X,Y \in \tilde{\mathcal {X}}\), we have
Then (a) is proved. \(\forall X^{*}\in \mathcal {X}^{*}\), if \(X^{*} \leqslant _{st} X\) (and Y) (or \(X^{*} \leqslant _{lr} X\) (and Y) or \(X^{*} \leqslant _{hr} X\) (and Y)), we have
which means that \(X,Y \in \tilde{\mathcal {X}}\). According to (a), (b) can be proved.
1.8 Proof of Proposition 6
Proof
From Dhaene et al. (2006) we know that VaR and distortion risk measure are additive of for sums of comonotonic risks. Then we can have
Due to that \(U_{\beta }^{\phi }( X)>0\), \(U_{\beta }^{\phi }(Y)>0\), we can have
1.9 Proof of Proposition 7
Proof
-
1.
Due to that \(\forall u \in [0,1]\), \(\sum \limits _{i=1}^N S_i(x)\cdot g_i(u)\) is non-decreasing with respect to \(x \in \mathcal {R}\), we can have that if \(X \leqslant _{st} Y\) (or \(X \leqslant _{lr} Y\), or \(X \leqslant _{hr} Y\)),
$$\begin{aligned} \kappa _{X}(u) = \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot g_i(u) \leqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(Y)\cdot g_i(u) = \kappa _{Y}(u). \end{aligned}$$From the definition, we know that
$$\begin{aligned} SlideM(X)&= \int _{-\infty }^{0}\left[ \kappa _{X}\left( P\left[ X> x\right] \right) -1\right] dx + \int _{0}^{+\infty }\kappa _{X}\left( P\left[ X> x\right] \right) dx \\&\leqslant \int _{-\infty }^{0}[\kappa _{Y}(P[Y> y])-1]dy + \int _{0}^{+\infty }\kappa _{Y}(P[Y > y])dy \\&= SlideM(Y). \end{aligned}$$ -
2.
\(\forall X \in \tilde{\mathcal {X}}_G\), we have \(SlideM(X) = \rho _{g_1}(X)\). If \(g_1(u)\) is concave, we have \(\rho _{g_1}(X+Y)\leqslant \rho _{g_1}(X) +\rho _{g_1}(Y)\) (Dhaene et al. 2012). Therefore, \(\forall X,Y \in \tilde{\mathcal {X}}_G\), we have
$$\begin{aligned} SlideM(X+Y) \leqslant \rho _{g_1}(X+Y)\leqslant \rho _{g_1}(X)+\rho _{g_1}(Y) =SlideM(X)+SlideM(Y). \end{aligned}$$Then (a) is proved. Due to that \(g_1(u) \geqslant g_2(u) \geqslant \cdots \geqslant g_N(u)\), and \(\sum \limits _{i=1}^N S_i(x)\cdot g_i(u)\) is non-decreasing with respect to x, \(\forall X^{*}\in \mathcal {X}^{*}\), if \(X^{*} \leqslant _{st} X\) (and Y) (or \(X^{*} \leqslant _{lr} X\) (and Y) or \(X^{*} \leqslant _{hr} X\) (and Y)), we have
$$\begin{aligned} S_1 \circ U_{\beta }^{\phi }(Y) = S_1 \circ U_{\beta }^{\phi }(X) = S_1 \circ U_{\beta }^{\phi }(X^{*}) = 1, \end{aligned}$$According to (a), (b) can be proved.
-
3.
\(\forall X, Y\) which satisfy (a) \(\forall\) \(X,Y\in \tilde{\mathcal {X}}_G\), or (b) if \({\exists }X^{*}\in \mathcal {X}^{*}_G\) s.t. \(X^{*} \leqslant _{st} X\) (and Y), we have \(SlideM(X) = \rho _{g_1}(X)\), \(SlideM(Y) = \rho _{g_1}(Y)\). Then \(X \leqslant _{icx} Y\) implies that \(SlideM(X) \leqslant SlideM(Y)\).
-
4.
\(\forall X\in \tilde{\mathcal {X}}_G\), we have \(S_1\circ U_{\beta }^{\phi }(X)= 1\). If \(X \leqslant _{st} Y\) (or \(X \leqslant _{lr} Y\) or \(X \leqslant _{hr} Y\)), due to that \(g_1(u) \geqslant g_2(u) \geqslant \cdots \geqslant g_N(u)\), and \(\sum \limits _{i=1}^N S_i(x)\cdot g_i(u)\) is non-decreasing with respect to x, we have \(S\circ U_{\beta }^{\phi }(Y) = S\circ U_{\beta }^{\phi }(X)= 1\), which implies that \(Y \in \tilde{\mathcal {X}}_G\).
-
5.
\(\forall a \in \mathcal {R}\), we have
$$\begin{aligned} SlideM(X+a) = \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X+a)\cdot \rho _{g_i}(X+a) = \sum \limits _{i=1}^N S_i \circ \left( U_{\beta }^{\phi }(X)+a \right) \cdot \rho _{g_i}(X) +a. \end{aligned}$$If \(a \geqslant 0\), we know that \(\sum \limits _{i=1}^N S_i \circ \left( U_{\beta }^{\phi }(X)+a \right) \geqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\) which implies that
$$\begin{aligned} SlideM(X+a) \geqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X)+a = SlideM(X)+a. \end{aligned}$$The equal sign holds when \(a=0\). Similarly, if \(a \leqslant 0\), we know that \(\sum \limits _{i=1}^N S_i \circ \left( U_{\beta }^{\phi }(X)+a \right) \leqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\) which implies that
$$\begin{aligned} SlideM(X+a) \leqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X)+a = SlideM(X)+a. \end{aligned}$$The equal sign holds when \(a=0\).
-
6.
Due to that
$$\begin{aligned} SlideM(\lambda X)&= \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(\lambda X)\cdot \rho _{g_i}(\lambda X) = \lambda \sum \limits _{i=1}^N S_i \circ (\lambda U_{\beta }^{\phi }( X))\cdot \rho _{g_i}(X), \\ \lambda SlideM(X)&= \lambda \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X), \end{aligned}$$we can have
$$\begin{aligned} SlideM(\lambda X) - \lambda SlideM(X) = \lambda \left[ \sum \limits _{i=1}^N S_i \circ (\lambda U_{\beta }^{\phi }( X)) \cdot \rho _{g_i}(X) - \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X) \right] . \end{aligned}$$From \(U_{\beta }^{\phi }(X)>0\), we know that, if \(\lambda \geqslant 1\), we have \(\sum \limits _{i=1}^N S_i \circ (\lambda U_{\beta }^{\phi }( X)) \cdot \rho _{g_i}(X) - \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X) \geqslant 0\) which implies that \(SlideM(\lambda X) - \lambda SlideM(X) \geqslant 0\). The equal sign holds when \(\lambda =1\). Similarly, if \(\lambda \leqslant 1\), we have \(\sum \limits _{i=1}^N S_i \circ (\lambda U_{\beta }^{\phi }( X)) \cdot \rho _{g_i}(X) - \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X) \leqslant 0\) which implies that \(SlideM(\lambda X) - \lambda SlideM(X) \leqslant 0\). The equal sign holds when \(\lambda =1\).
-
7.
\(\forall X \in \tilde{\mathcal {X}}_G\), we have \(SlideM(X) = \rho _{g_1}(X)\). Therefore, \(\forall X,Y \in \tilde{\mathcal {X}}_G\), we have
$$\begin{aligned}&SlideM(\lambda X+(1-\lambda ) Y) \leqslant \rho _{g_1}(\lambda X+(1-\lambda ) Y)\leqslant \rho _{g_1}(\lambda X)+\rho _{g_1}((1-\lambda ) Y) \\ =&\lambda \rho _{g_1}(X)+(1-\lambda )\rho _{g_1}( Y) = \lambda SlideM(X)+(1-\lambda ) SlideM(Y). \end{aligned}$$Then (a) is proved. \(\forall X^{*}\in \mathcal {X}^{*}_G\), if \(X^{*} \leqslant _{st} X\) (and Y) (or \(X^{*} \leqslant _{lr} X\) (and Y) or \(X^{*} \leqslant _{hr} X\) (and Y)), we have
$$\begin{aligned} S_1\circ U_{\beta }^{\phi }(Y) = S_1\circ U_{\beta }^{\phi }(X) = S_1\circ U_{\beta }^{\phi }(X^{*}) = 1, \end{aligned}$$which means that \(X,Y \in \tilde{\mathcal {X}}_G\). According to (a), (b) can be proved.
-
8.
From Dhaene et al. (2006) we know that distortion risk measure is additive of for sums of comonotonic risks. Then we can have
$$\begin{aligned} SlideM(X+Y)&= \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X+Y)\cdot \rho _{g_i}(X+Y) \\&= \sum \limits _{i=1}^N S_i \circ (U_{\beta }^{\phi }(X)+U_{\beta }^{\phi }(Y))\cdot \rho _{g_i}(X)+\sum \limits _{i=1}^N S_i \circ (U_{\beta }^{\phi }(X)+U_{\beta }^{\phi }(Y))\cdot \rho _{g_i}(Y). \end{aligned}$$Due to that \(U_{\beta }^{\phi }( X)>0\), \(U_{\beta }^{\phi }(Y)>0\), we can have
$$\begin{aligned} SlideM(X+Y)&\geqslant \sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(X)\cdot \rho _{g_i}(X)+\sum \limits _{i=1}^N S_i \circ U_{\beta }^{\phi }(Y)\cdot \rho _{g_i}(Y) \\&= SlideM(X)+SlideM(Y). \end{aligned}$$
Other Examples of Analytical SlideVaR Expressions
1.1 Exponential Distribution
Let X be an exponential distributed random variable with parameters \(\lambda\), i.e. \(X \sim Exp(\lambda )\), then we know that \(E[X]=SD[X]=1/\lambda\). Because the exponential is
we can have
Thanks to integration by parts, we can get
For the function \(\phi ^e_{\beta }(p)\) in Eq. (11), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) = \dfrac{-\ln {(1-\beta )}}{\lambda }\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^e}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \frac{1}{\lambda (b-a)} \left( \ln \frac{1-\beta }{1-\alpha }+1\right) \left[ \frac{1}{\gamma (1-e^{\frac{\beta -1}{\gamma }})}\sum _{i=2}^{N} \left( e^{\frac{\beta _i-1}{\gamma }}-e^{\frac{\beta _{i-1}-1}{\gamma }}\right) (1-\beta _i) \right. \\&\left. \frac{-\ln {(1-\beta _i)}+1}{\lambda } + \frac{e^{\frac{\beta -1}{\gamma }}(1-\beta )}{\gamma (1-e^{\frac{\beta -1}{\gamma }})} \cdot \frac{-\ln {(1-\beta )}+1}{\lambda }-a \right] + \frac{-\ln {(1-\beta )}}{\lambda }; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X) = ES_{\alpha }(X)=\dfrac{-\ln {(1-\alpha )}+1}{\lambda }\).
For the function \(\phi ^p_{\beta }(p)\) in Eq. (12), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^p}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) = \dfrac{-\ln {(1-\beta )}}{\lambda }\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^p}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \frac{1}{\lambda (b-a)} \left( \ln \frac{1-\beta }{1-\alpha }+1\right) \left[ \frac{1-\gamma }{(1-\beta )^{1-\gamma }}\sum _{i=2}^{N} \left[ (1-\beta _{i})^{-\gamma }-(1-\beta _{i-1})^{-\gamma }\right] \right. \\&\left. (1-\beta _i)\frac{-\ln {(1-\beta _i)}+1}{\lambda } + (1-\gamma )\cdot \frac{-\ln {(1-\beta )}+1}{\lambda } - a \right] + \frac{-\ln {(1-\beta )}}{\lambda }; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^p}(X) \geqslant b\), \(SlideVaR(X) =ES_{\alpha }(X)= \dfrac{-\ln {(1-\alpha )}+1}{\lambda }\).
In addition, if \(U_{\beta ,N}^{\phi ^e}(X)\) or \(U_{\beta ,N}^{\phi ^p}(X) \equiv h(b-a)+a\) where h is a constant in [0, 1]. Then
which becomes \(GlueVaR_{\alpha ,\beta }^{h_1,h_2}(X)\) with \(h_1=h_2=h\) (Belles-Sampera et al. 2014).
1.2 Pareto Distribution
Let X be a Pareto distributed random variable with parameters \((k, \sigma )\) where \(k>0\), \(\sigma >0\), i.e. \(X \sim Pareto(k, \sigma )\), then we know that
Because the Pareto CDF is
we can have
From Norton et al. (2021) we know that
For the function \(\phi ^e_{\beta }(p)\) in Eq. (11), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) =\dfrac{\sigma }{(1-\beta )^{\frac{1}{k}}}\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^e}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \frac{k (1-\beta )^{\frac{1}{k}} - (1-\alpha )^{\frac{1}{k}}(k-1)}{(1-\alpha )^{\frac{1}{k}}(1-\beta )^{\frac{1}{k}}(k-1)} \cdot \frac{\sigma }{b-a}\left[ \frac{k\sigma }{\gamma (1-e^{\frac{\beta -1}{\gamma }})(k-1)} \right. \\&\left. \sum _{i=2}^{N} \left( e^{\frac{\beta _i-1}{\gamma }}-e^{\frac{\beta _{i-1}-1}{\gamma }}\right) (1-\beta _i)^{1-\frac{1}{k}}+ \frac{k\sigma e^{\frac{\beta -1}{\gamma }}(1-\beta )^{1-\frac{1}{k}}}{\gamma (1-e^{\frac{\beta -1}{\gamma }})(k-1)}-a \right] \\&+ \dfrac{\sigma }{(1-\beta )^{\frac{1}{k}}}; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X) = ES_{\alpha }(X)=\dfrac{k\sigma }{(1-\alpha )^{\frac{1}{k}}(k-1)}\).
For the function \(\phi ^p_{\beta }(p)\) in Eq. (12), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^p}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) =\dfrac{\sigma }{(1-\beta )^{\frac{1}{k}}}\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^p}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \frac{k (1-\beta )^{\frac{1}{k}} - (1-\alpha )^{\frac{1}{k}}(k-1)}{(1-\alpha )^{\frac{1}{k}}(1-\beta )^{\frac{1}{k}}(k-1)} \cdot \frac{\sigma }{b-a}\left[ \frac{(1-\gamma )k\sigma }{(1-\beta )^{1-\gamma }(k-1)} \right. \\&\left. \sum _{i=2}^{N} \left[ (1-\beta _{i})^{-\gamma }-(1-\beta _{i-1})^{-\gamma }\right] (1-\beta _i)^{1-\frac{1}{k}}+ \frac{(1-\gamma )k\sigma }{(1-\beta )^{\frac{1}{k}}(k-1)} - a \right] \\&+ \dfrac{\sigma }{(1-\beta )^{\frac{1}{k}}}; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^p}(X) \geqslant b\), \(SlideVaR(X) = ES_{\alpha }(X)=\dfrac{k\sigma }{(1-\alpha )^{\frac{1}{k}}(k-1)}\).
In addition, if \(U_{\beta ,N}^{\phi ^e}(X)\) or \(U_{\beta ,N}^{\phi ^p}(X) \equiv h(b-a)+a\) where h is a constant in [0, 1]. Then
which equals to \(GlueVaR_{\alpha ,\beta }^{h_1,h_2}(X)\) with \(h_1=h_2=h\) (Belles-Sampera et al. 2014).
1.3 Generalized Pareto Distribution
Let X be a Generalized Pareto distributed random variable with parameters \((\mu , \sigma , \xi )\) where \(\sigma >0\), i.e. \(X \sim GP(\mu , \sigma , \xi )\), then we know that
Because the Pareto CDF is
for \(x \geqslant \mu\) when \(\xi \geqslant 0\) and \(\mu \leqslant x \leqslant \mu -\frac{\sigma }{\xi }\) when \(\xi <0\), we can have
From Norton et al. (2021) we know that, with \(-1< \xi <1\),
We know that, if \(\xi =0\), X would become an exponential distributed random variable which we have discussed in Sect. 1. If \(\xi <0\), X would become a Pareto distributed random variable which we have discussed in Sect. 2. Therefore, in this section, we only consider the case where \(\xi \in (0,1)\).
For the function \(\phi ^e_{\beta }(p)\) in Eq. (11), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) = \mu + \sigma \dfrac{(1-\beta )^{-\xi }-1}{\xi }\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^p}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \left[ \frac{(1-\alpha )^{-\xi }-(1-\beta )^{-\xi }}{\xi }+ \frac{(1-\alpha )^{-\xi }}{1-\xi } \right] \frac{\sigma }{b-a}\left[ \frac{1}{\gamma (1-e^{\frac{\beta -1}{\gamma }})} \right. \\&\left. \sum _{i=2}^{N} \left( e^{\frac{\beta _i-1}{\gamma }}-e^{\frac{\beta _{i-1}-1}{\gamma }}\right) (1-\beta _i)\left( \mu + \sigma \frac{(1-\beta _i)^{-\xi }-1}{\xi }+ \sigma \frac{(1-\beta _i)^{-\xi }}{1-\xi } \right) \right. \\&\left. + \frac{e^{\frac{\beta -1}{\gamma }}(1-\beta )}{\gamma (1-e^{\frac{\beta -1}{\gamma }})} \left( \mu + \sigma \frac{(1-\beta )^{-\xi }-1}{\xi }+ \sigma \frac{(1-\beta )^{-\xi }}{1-\xi } \right) -a \right] + \mu \\&+ \sigma \frac{(1-\beta )^{-\xi }-1}{\xi }; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X)=ES_{\alpha }(X)=\mu + \sigma \dfrac{(1-\alpha )^{-\xi }-1}{\xi }+ \sigma \dfrac{(1-\alpha )^{-\xi }}{1-\xi }\).
For the function \(\phi ^p_{\beta }(p)\) in Eq. (12), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = VaR_{\beta }(X) = \mu + \sigma \dfrac{(1-\beta )^{-\xi }-1}{\xi }\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^p}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \left[ \frac{(1-\alpha )^{-\xi }-(1-\beta )^{-\xi }}{\xi }+ \frac{(1-\alpha )^{-\xi }}{1-\xi } \right] \frac{\sigma }{b-a}\left[ \frac{1-\gamma }{(1-\beta )^{1-\gamma }} \right. \\&\left. \sum _{i=2}^{N} \left[ (1-\beta _{i})^{-\gamma } -(1-\beta _{i-1})^{-\gamma }\right] (1-\beta _i)\left( \mu + \sigma \frac{(1-\beta _i)^{-\xi }-1}{\xi } \right. \right. \\&\left. \left. +\sigma \frac{(1-\beta _i)^{-\xi }}{1-\xi } \right) + (1-\gamma ) \left( \mu + \sigma \frac{(1-\beta )^{-\xi }-1}{\xi }+ \sigma \frac{(1-\beta )^{-\xi }}{1-\xi } \right) \right. \\&\left. -a \right] + \mu + \sigma \frac{(1-\beta )^{-\xi }-1}{\xi }; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X)=ES_{\alpha }(X)=\mu + \sigma \dfrac{(1-\alpha )^{-\xi }-1}{\xi }+ \sigma \dfrac{(1-\alpha )^{-\xi }}{1-\xi }\).
In addition, if \(U_{\beta ,N}^{\phi ^e}(X)\) or \(U_{\beta ,N}^{\phi ^p}(X) \equiv h(b-a)+a\) where h is a constant in [0, 1]. Then
which equals to \(GlueVaR_{\alpha ,\beta }^{h_1,h_2}(X)\) with \(h_1=h_2=h\) (Belles-Sampera et al. 2014).
1.4 Weibull Distribution
Let X be a Weibull distributed random variable with parameters \((\lambda , k)\), i.e. \(X \sim Weibull(\lambda , k)\), then we know that \(E[X]=\lambda \Gamma (1+\frac{1}{k})\), \(D[X]=\lambda ^2 \left[ \Gamma (1+\frac{2}{k})-\Gamma (1+\frac{1}{k})^2 \right]\) where \(\Gamma (a) = \int _0^\infty p^{a-1}e^{-p}dp\) is the gamma function. Because the Weibull CDF is
we can have
From Norton et al. (2021) we know that
where \(\Gamma _U(a,b) = \int _b^\infty p^{a-1}e^{-p}dp\) is the upper incomplete gamma function. For the function \(\phi ^e_{\beta }(p)\) in Eq. (11), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = \lambda (-\ln (1-\beta ))^{\frac{1}{k}}\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^e}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \left[ \frac{\Gamma _U\left( 1+\frac{1}{k},-\ln (1-\alpha ) \right) }{1-\alpha } - [-\ln (1-\beta )]^{\frac{1}{k}} \right] \frac{\lambda }{b-a}\left[ \frac{\lambda }{\gamma (1-e^{\frac{\beta -1}{\gamma }})} \right. \\&\left. \sum _{i=2}^{N} \left( e^{\frac{\beta _i-1}{\gamma }}-e^{\frac{\beta _{i-1}-1}{\gamma }}\right) \Gamma _U\left( 1+\frac{1}{k},-\ln (1-\beta _i) \right) + \frac{\lambda e^{\frac{\beta -1}{\gamma }}}{\gamma (1-e^{\frac{\beta -1}{\gamma }})} \right. \\&\left. \Gamma _U\left( 1+\frac{1}{k},-\ln (1-\beta ) \right) -a \right] +\lambda (-\ln (1-\beta ))^{\frac{1}{k}}; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X) = \dfrac{\lambda }{1-\alpha }\Gamma _U\left( 1+\frac{1}{k},-\ln (1-\alpha ) \right)\).
For the function \(\phi ^p_{\beta }(p)\) in Eq. (12), we have
Therefore, we have
-
When \(U_{\beta ,N}^{\phi ^e}(X) < a\), \(SlideVaR(X) = \lambda (-\ln (1-\beta ))^{\frac{1}{k}}\);
-
When \(a \leqslant U_{\beta ,N}^{\phi ^e}(X) < b\),
$$\begin{aligned} SlideVaR(X)&= \left[ \frac{\Gamma _U\left( 1+\frac{1}{k},-\ln (1-\alpha ) \right) }{1-\alpha } - [-\ln (1-\beta )]^{\frac{1}{k}} \right] \frac{\lambda }{b-a}\left[ \frac{\lambda (1-\gamma )}{(1-\beta )^{1-\gamma }} \right. \\&\left. \sum _{i=2}^{N} \left[ (1-\beta _{i})^{-\gamma }-(1-\beta _{i-1})^{-\gamma }\right] \Gamma _U\left( 1+\frac{1}{k},-\ln (1-\beta _i) \right) \right. \\&\left. + \frac{\lambda (1-\gamma )}{1-\beta }\Gamma _U\left( 1+\frac{1}{k},-\ln (1-\beta ) \right) -a \right] +\lambda (-\ln (1-\beta ))^{\frac{1}{k}}; \end{aligned}$$ -
When \(U_{\beta ,N}^{\phi ^e}(X) \geqslant b\), \(SlideVaR(X) = \dfrac{\lambda }{1-\alpha }\Gamma _U\left( 1+\frac{1}{k},-\ln (1-\alpha ) \right)\).
In addition, if \(U_{\beta ,N}^{\phi ^e}(X)\) or \(U_{\beta ,N}^{\phi ^p}(X) \equiv h(b-a)+a\) where h is a constant in [0, 1]. Then
which equals to \(GlueVaR_{\alpha ,\beta }^{h_1,h_2}(X)\) with \(h_1=h_2=h\).
Rights and permissions
About this article
Cite this article
Hu, W., Chen, C., Shi, Y. et al. A Tail Measure With Variable Risk Tolerance: Application in Dynamic Portfolio Insurance Strategy. Methodol Comput Appl Probab 24, 831–874 (2022). https://doi.org/10.1007/s11009-022-09951-4
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11009-022-09951-4