Kernel estimators for the second order parameter in extreme value statistics

Abstract We develop and study in the framework of Pareto-type distributions a general class of kernel estimators for the second order parameter ρ , a parameter related to the rate of convergence of a sequence of linearly normalized maximum values towards its limit. Inspired by the kernel goodness-of-fit statistics introduced in Goegebeur et al. (2008) , for which the mean of the normal limiting distribution is a function of ρ , we construct estimators for ρ using ratios of ratios of differences of such goodness-of-fit statistics, involving different kernel functions as well as power transformations. The consistency of this class of ρ estimators is established under some mild regularity conditions on the kernel function, a second order condition on the tail function 1−F of the underlying model, and for suitably chosen intermediate order statistics. Asymptotic normality is achieved under a further condition on the tail function, the so-called third order condition. Two specific examples of kernel statistics are studied in greater depth, and their asymptotic behavior illustrated numerically. The finite sample properties are examined by means of a simulation study.

[1]  J. Teugels,et al.  Tail Index Estimation, Pareto Quantile Plots, and Regression Diagnostics , 1996 .

[2]  J. Geluk,et al.  Regular variation, extensions and Tauberian theorems , 1987 .

[3]  Liang Peng,et al.  Semi-parametric Estimation of the Second Order Parameter in Statistics of Extremes , 2002 .

[4]  A. Rényi.  On the theory of order statistics , 1953 .

[5]  P. Hall,et al.  Estimating a tail exponent by modelling departure from a Pareto distribution , 1999 .

[6]  Jan Beirlant,et al.  Tail Index Estimation and an Exponential Regression Model , 1999 .

[7]  L. Haan,et al.  A moment estimator for the index of an extreme-value distribution , 1989 .

[8]  Yuri Goegebeur,et al.  Burr regression and portfolio segmentation , 1998 .

[9]  Jan Beirlant,et al.  LINKING PARETO-TAIL KERNEL GOODNESS-OF-FIT STATISTICS WITH TAIL INDEX AT OPTIMAL THRESHOLD AND SECOND ORDER ESTIMATION , 2008 .

[10]  J. Teugels,et al.  Statistics of Extremes , 2004 .

[11]  B. Gnedenko.  Sur La Distribution Limite Du Terme Maximum D'Une Serie Aleatoire , 1943 .

[12]  M. J. Martins,et al.  “Asymptotically Unbiased” Estimators of the Tail Index Based on External Estimation of the Second Order Parameter , 2002 .

[13]  Cécile Mercadier,et al.  Semi-parametric estimation for heavy tailed distributions , 2010 .

[14]  Michel Loève,et al.  Probability Theory I , 1977 .

[15]  M. Ivette Gomes,et al.  A new class of semi-parametric estimators of the second order parameter. , 2003 .

[16]  Jan Beirlant,et al.  On Exponential Representations of Log-Spacings of Extreme Order Statistics , 2002 .

[17]  Armelle Guillou,et al.  A diagnostic for selecting the threshold in extreme value analysis , 2001 .

[18]  Edgar Kaufmann,et al.  Selecting the optimal sample fraction in univariate extreme value estimation , 1998 .

[19]  Alan H. Welsh,et al.  Adaptive Estimates of Parameters of Regular Variation , 1985 .

[20]  L. Haan,et al.  Extreme value theory : an introduction , 2006 .

[21]  Joseph L. Gastwirth,et al.  Asymptotic Distribution of Linear Combinations of Functions of Order Statistics with Applications to Estimation , 1967 .

[22]  M. Ivette Gomes,et al.  Tail index estimation for heavy‐tailed models: accommodation of bias in weighted log‐excesses , 2007 .