Message Board

Dear readers, authors and reviewers,you can add a message on this page. We will reply to you as soon as possible!

2023 Volume 48 Issue 5
Article Contents

WANG Zhihua, PENG Zuoxiang. Tail Dependence Function of the Dynamic Bivariate Skew-normal Distribution[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(5): 42-46. doi: 10.13718/j.cnki.xsxb.2023.05.006
Citation: WANG Zhihua, PENG Zuoxiang. Tail Dependence Function of the Dynamic Bivariate Skew-normal Distribution[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(5): 42-46. doi: 10.13718/j.cnki.xsxb.2023.05.006

Tail Dependence Function of the Dynamic Bivariate Skew-normal Distribution

More Information
  • Corresponding author: PENG Zuoxiang
  • Received Date: 04/09/2022
    Available Online: 20/05/2023
  • MSC: O211.3

  • Based on the study of tail dependence function of bivariate normal distribution under Hüsler-Reiss condition, combined with the study of the bivariate skew-normal correlation property and the derivation of the tail dependence coefficient, this paper gives the tail dependence function of the dynamic bivariate skew-normal distribution when the skewness parameters are greater than zero and less than zero.
  • 加载中
  • [1] KLVPPELBERG C, KUHN G, PENG L. Estimating the Tail Dependence Function of an Elliptical Distribution[J]. Bernoulli, 2007, 13(1): 229-251.

    Google Scholar

    [2] HULT H, LINDSKOG F. Multivariate Extremes, Aggregation and Dependence in Elliptical Distributions[J]. Advances in Applied Probability, 2002, 34(3): 587-608. doi: 10.1239/aap/1033662167

    CrossRef Google Scholar

    [3] JOE H, LI H, NIKOLOULOPOULOS A K. Tail Dependence Functions and Vine Copulas[J]. Journal of Multivariate Analysis, 2010, 101(1): 252-270. doi: 10.1016/j.jmva.2009.08.002

    CrossRef Google Scholar

    [4] BORTOT P. Tail Dependence In Bivariate Skew-Normal and Skew-t Distribution[EB/OL]. (2010-03-11)[2022-03-28]. http://www2.stat.unibo.it/bortot/ricerca/paper-sn-2.pdf.

    Google Scholar

    [5] FUNG T, SENETA E. Tail Asymptotics for the Bivariate Skew Normal[J]. Journal of Multivariate Analysis, 2016, 144: 129-138. doi: 10.1016/j.jmva.2015.11.002

    CrossRef Google Scholar

    [6] FRAHM G, JUNKER M, SCHMIDT R. Estimating the Tail-Dependence Coefficient: Properties and Pitfalls[J]. Mathematics and Economics, 2005, 37(1): 80-100. doi: 10.1016/j.insmatheco.2005.05.008

    CrossRef Google Scholar

    [7] AZZALINI A, DALLA VALLE A. The Multivariate Skew-Normal Distribution[J]. Biometrika, 1996, 83(4): 715-726. doi: 10.1093/biomet/83.4.715

    CrossRef Google Scholar

    [8] AZZALINI A A. A Class of Distributions Which Include the Normal Scandinavian Journal of Statistics, 1985, 12: 171-178.

    Google Scholar

    [9] BALAKRISHNAN N, SCARPA B. Multivariate Measures of Skewness for the Skew-Normal Distribution[J]. Journal of Multivariate Analysis, 2012, 104(1): 73-87. doi: 10.1016/j.jmva.2011.06.017

    CrossRef Google Scholar

    [10] 陈超, 田芫, 宗序平. 偏对称正态分布的若干性质[J]. 淮阴师范学院学报(自然科学版), 2019, 18(1): 12-18.

    Google Scholar

    [11] CAPITANIO A. On the Approximation of the Tail Probability of the Scalar Skew-Normal Distribution[J]. International Journal of Statistics, 2010, 68: 299-308.

    Google Scholar

    [12] AZZALINI A, CAPITANIO A. The Skew-Normal and Related Families[M]. Cambridge: Cambridge University Press, 2014.

    Google Scholar

    [13] LIAO X, PENG Z X, NADARAJAH S, et al. Rates of Convergence of Extremes from Skew-Normal Samples[J]. Statistics & Probability Letters, 2014, 84: 40-47.

    Google Scholar

    [14] HU S, PENG Z X, NADARAJAH S. Tail Dependence Functions of the Bivariate Hüsler-Reiss Model[J]. Statistics & Probability Letters, 2022, 180: 109-235.

    Google Scholar

    [15] HVSLER J, REISS R D. Maxima of Normal Random Vectors: Between Independence and Complete Dependence[J]. Statistics & Probability Letters, 1989, 7(4): 283-286.

    Google Scholar

    [16] FUNG T, SENETA E. Quantile Function Expansion Using Regularly Varying Functions[J]. Methodology and Computing in Applied Probability, 2018, 20(4): 1091-1103. doi: 10.1007/s11009-017-9593-0

    CrossRef Google Scholar

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(1407) PDF downloads(179) Cited by(0)

Access History

Other Articles By Authors

Tail Dependence Function of the Dynamic Bivariate Skew-normal Distribution

    Corresponding author: PENG Zuoxiang

Abstract: Based on the study of tail dependence function of bivariate normal distribution under Hüsler-Reiss condition, combined with the study of the bivariate skew-normal correlation property and the derivation of the tail dependence coefficient, this paper gives the tail dependence function of the dynamic bivariate skew-normal distribution when the skewness parameters are greater than zero and less than zero.

  • 设(XY)是独立同分布的随机变量,具有连续的边缘分布函数FXFY. 对xy≥0,定义尾相依函数为

    D(1,1)为上尾相依系数[1-2]. 有关尾相依函数的性质见文献[3];有关尾相依系数的推导见文献[4-6]. 尾相依的概念描述了两个随机变量的尾相关结构,尾相关性的一个常用度量方法是所谓的尾相依系数. 尾相依系数在现代风险管理中越来越受到重视,因此估计尾部依赖性是非常重要的. 尾相依系数是尾相依函数的一种特殊情况,它度量两个变量之间的极值相关性. 因此本文考虑动态二元偏正态分布的尾相依函数.

    二元偏正态分布的联合密度函数为

    其中:$\phi_2(\cdot ; \rho)$表示相关系数为$\rho \in(-1, 1)$的二元正态随机变量的密度函数;$\varPhi(\cdot)$是标准正态分布函数[7-9]. 偏正态分布相关性质的研究见文献[10-12]. 对于二元偏正态分布的尾部独立性和尾相依系数的收敛速度的研究见文献[4-5, 13]. 假设($\xi, \eta$)服从等偏动态二元偏正态分布,即$\rho=\rho_n, \alpha_1=\alpha_2$,由(2)式可知其联合密度函数为$f(x, y)=2 \phi_2\left(x, y ; \rho_n\right) \varPhi(\alpha x+\alpha y), \alpha \in \mathbb{R}$,记连续的边缘分布函数为$F(\cdot)$,边缘密度函数为$f(x)=2 \phi(x) \varPhi\left(\beta_n x\right)$,偏度系数$\beta_n=\frac{\alpha\left(1+\rho_n\right)}{\sqrt{1+\alpha^2\left(1-\rho_n^2\right)}}$,依据文献[5]假设当$n \rightarrow \infty$时,$\rho_n$满足

    α=0时,文献[14]给出了二元正态分布在Hüsler-Reiss条件[15]即(3)式成立情况下的尾相依函数. 对xy>0,根据(1)式再结合文献[1, 14],定义(ξη)的下尾相依函数为

    其中

    本文分别考虑偏度参数α < 0和α>0时的动态二元偏正态分布的尾相依函数.

1.   主要结论
  • 对参数为$\alpha \in \mathbb{R}$的偏正态分布函数F,定义

    $F^{\leftarrow}(x)$表示$F(x)$的逆函数. 我们有如下两个引理.

    引理1    设函数$g_n(v)$(v)由(6)式给出,则

       由文献[16]易得.

    引理2    假设(3)式成立,当$n \rightarrow \infty$时,有

       结合引理1和(3)式得证.

    定理1    假设(3)式成立,则对xy>0,有

       根据(5),(6)式以及条件概率积分公式有

    其中

    $\alpha <0$,则$\beta_n <0$,当$n \rightarrow \infty$时,$\frac{x}{n} \rightarrow 0^{+}, g_n(x)=F^{\leftarrow}\left(\frac{x}{n}\right) \rightarrow-\infty$,所以$\varPhi\left(\beta_n g_n(x)\right) \rightarrow 1$. 由

    及(8)式可得

    由于$n \rightarrow \infty$时,$\varPhi\left(\alpha\left(g_n(x)+g_n(v)\right)\right) \rightarrow 1, \varPhi\left(\beta_n g_n(x)\right) \rightarrow 1$,再结合引理2,(8)式取极限为

    由(7)和(9)式,根据Φ(·)的连续性和控制收敛定理,有

    ε任意性,α < 0情形得证.

    $\alpha>0$,当$n \rightarrow \infty$时,$g_n(v) \rightarrow-\infty$,并且$-\alpha \sqrt{1-\rho_n^2} z-\alpha\left(1+\rho_n\right) g_n(v)>0$. 当$\beta_n>0$时,$-\beta_n g_n(v)>0$. 由$\varPhi(x)=1-\varPhi(-x)$可得

    根据Mills不等式

    $n \rightarrow \infty$时,有

    同样地

    再结合引理2时(8)式取极限,有

    由(7)和(11)式,根据Φ(·)的连续性和控制收敛定理,有

    ε的任意性,定理证毕.

Reference (16)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return