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 6
Article Contents

YU Lele, PENG Zuoxiang. Asymptotic Properties of a Class of Semi-parametric Tail Exponential Estimators[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(6): 54-58. doi: 10.13718/j.cnki.xsxb.2023.06.008
Citation: YU Lele, PENG Zuoxiang. Asymptotic Properties of a Class of Semi-parametric Tail Exponential Estimators[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(6): 54-58. doi: 10.13718/j.cnki.xsxb.2023.06.008

Asymptotic Properties of a Class of Semi-parametric Tail Exponential Estimators

More Information
  • Corresponding author: PENG Zuoxiang
  • Received Date: 05/09/2022
    Available Online: 20/06/2023
  • MSC: O211.4

  • Based on the statistics constructed by logarithmic function and power function, this paper proposes a class of semi parametric tail exponential estimators, and proves their consistency and asymptotic normality.
  • 加载中
  • [1] DE HAAN L, FERREIRA A. Extreme Value Theory [M]. New York: Springer, 2006.

    Google Scholar

    [2] RESNICK S I. Point Processes [M]//Extreme Values, Regular Variation and Point Processes. New York: Springer, 1987: 123-161.

    Google Scholar

    [3] HILL B M. A Simple General Approach to Inference about the Tail of a Distribution [J]. The Annals of Statistics, 1975, 3(5): 1163-1174.

    Google Scholar

    [4] DANIELSSON J, JANSEN D W, DE VRIES C G. The Method of Moments Ratio Estimator for the Tail Shape Parameter [J]. Communications in Statistics-Theory and Methods, 1996, 25(4): 711-720. doi: 10.1080/03610929608831727

    CrossRef Google Scholar

    [5] PAULAUSKAS V, VAI CČIULIS M. A Class of New Tail Index Estimators [J]. Annals of the Institute of Statistical Mathematics, 2017, 69(2): 461-487. doi: 10.1007/s10463-015-0548-3

    CrossRef Google Scholar

    [6] QI Y C. On the Tail Index of a Heavy Tailed Distribution [J]. Annals of the Institute of Statistical Mathematics, 2010, 62(2): 277-298. doi: 10.1007/s10463-008-0176-2

    CrossRef Google Scholar

    [7] PAULAUSKAS V, VAI CČIULIS M. Comparison of the Several Parameterized Estimators for the Positive Extreme Value Index [J]. Journal of Statistical Computation and Simulation, 2017, 87(7): 1342-1362. doi: 10.1080/00949655.2016.1263303

    CrossRef Google Scholar

    [8] VAI CČIULIS M, MARKOVICH N M. A Class of Semiparametric Tail Index Estimators and Its Applications [J]. Automation and Remote Control, 2019, 80(10): 1803-1816. doi: 10.1134/S0005117919100035

    CrossRef Google Scholar

    [9] BERAN J, SCHELL D, STEHLÍK M. The Harmonic Moment Tail Index Estimator: Asymptotic Distribution and Robustness [J]. Annals of the Institute of Statistical Mathematics, 2014, 66(1): 193-220. doi: 10.1007/s10463-013-0412-2

    CrossRef Google Scholar

    [10] CAEIRO F, GOMES M I, BEIRLANT J, et al. Mean-of-Order p Reduced-Bias Extreme Value Index Estimation under a Third-Order Framework [J]. Extremes, 2016, 19(4): 561-589. doi: 10.1007/s10687-016-0261-5

    CrossRef Google Scholar

    [11] BEIRLANT J, VYNCKIER P, TEUGELS J L. Tail Index Estimation, Pareto Quantile Plots Regression Diagnostics [J]. Journal of the American Statistical Association, 1996, 91(436): 1659-1667.

    Google Scholar

    [12] RESINICK S I. Heavy-tail Phenomena [M]. New York: Springer, 2007.

    Google Scholar

    [13] CASELLA G, BERGER R L. Statical Inference [M]. New Jersey: Prentice Hall, 2002.

    Google Scholar

    [14] CRAMÉR H, WOLD H. Some Theorems on Distribution Functions [J]. Journal of the London Mathematical Society, 1936, s1-11(4): 290-294. doi: 10.1112/jlms/s1-11.4.290

    CrossRef Google Scholar

    [15] PAULAUSKAS V, VAI CČIULIS M. On an Improvement of Hill and some other Estimators [J]. Lithuanian Mathematical Journal, 2013, 53(3): 336-355. doi: 10.1007/s10986-013-9212-x

    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(1405) PDF downloads(132) Cited by(0)

Access History

Other Articles By Authors

Asymptotic Properties of a Class of Semi-parametric Tail Exponential Estimators

    Corresponding author: PENG Zuoxiang

Abstract: Based on the statistics constructed by logarithmic function and power function, this paper proposes a class of semi parametric tail exponential estimators, and proves their consistency and asymptotic normality.

  • 设{Xn,n≥1}为独立同分布的随机变量序列,其分布函数为$ F(x) . X_{1, n} \leqslant \cdots \leqslant X_{n, n}$表示X1,…,Xn的次序统计量. 若存在规范化常数an>0和bn及非退化分布函数Gγ(x)使得

    由文献[1-2]可知

    则称F属于G的吸引场,记为FD(G),γ为极值指数. 令$U(t)=F^{\leftarrow}\left(1-\frac{1}{t}\right), t \geqslant 1 $. 当γ>0时,(1)式等价于

    文献[3]提出了著名的Hill估计量. 文献[4]为减小Hill估计量的偏差,构造了矩率估计量. 文献[5]利用函数$g_{r, u}(x)=x^r \ln ^u(x), x \geqslant 1 $构造出如下的统计量

    其中γr < 1,u>-1. 利用(3)式可以将Hill估计量、矩率估计量表示出来:

    极值指数估计的应用非常广泛,相关研究可参见文献[6-10].

    本文利用统计量Gn(kru)构造如下的尾指数估计量

    假定(2)式成立且存在序列k=k(n)满足当$ n \rightarrow \infty$时,

    考虑$ \hat{\gamma}_n(k, r)$的弱相合性. 此外,如果存在可测函数A(t)使得

    成立,则我们讨论$\hat{\gamma}_n(k, r) $的渐近分布,其中ρ < 0表示二阶参数.

1.   相合性和渐近正态性
  • 定理1   假定γr < 1成立,在(2)式和(4)式的条件下,$\hat{\gamma}_n(k, r) \stackrel{\mathrm{P}}{\longrightarrow} \gamma . $

    定理2    假定$ \gamma r<\frac{1}{2}$成立,在(4)式和(5)式的条件下,存在$\lambda \in \mathbb{R} $使得

    则当$ n \rightarrow \infty$时,

    其中

2.   定理的证明
  • 设{Ynn≥1}为独立同分布的标准Pareto序列,Y1,n,…,Ynn表示Y1,…,Yn的次序统计量,由文献[11]可得到$ \left\{X_i\right\}_{i=1}^n \stackrel{\mathrm{d}}{=}\left\{U\left(Y_i\right)\right\}_{i=1}^n, \left\{\frac{Y_{n-i, n}}{Y_{n-k, n}}\right\}_{i=0}^{k-1} \stackrel{\mathrm{d}}{=}\left\{Y_{k-i, k}\right\}_{i=0}^{k-1}$.

    定理1的证明    由文献[5]的定理1可知,当$n \rightarrow \infty $时,

    得到

    利用连续映射定理[12]和Slutsky定理[13],定理得证.

    对定理2的证明,我们需要下面的辅助引理.

    引理1    在定理2的条件下,当$n \rightarrow \infty $时,有

    其中

    (N1N2)是二维零均值高斯向量,满足

    其中

       由二阶正规变换条件(5)式知,对充分大的t

    利用文献[14]中的Cramer-Wold定理证明(7)式成立. 对任意$ (\varphi, \psi) \in \mathbb{R}^2$,有

    其中

    $ \gamma_r<\frac{1}{2}$时,

    由列为林德伯格中心极限定理可得

    与文献[15]引理1类似计算,有

    由(11)式,(12)式及Slutsky定理,知

    结合(10)式和(13)式,引理得证.

    定理2的证明   定义

    利用泰勒展式,(8)式和(9)式化简为

    得到

    由引理1知

    结合(6)式,定理2得证.

Reference (15)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return