Test for Regularly Varying Tail Against Rapidly/Slowly Varying Tail
We introduce a new non-parametric test for testing the tail behaviour of a random variable using a characterization based on ratio of extreme order statistics. The asymptotic null distribution of the proposed test statistic is obtained. A Monte Carlo simulation study is carried out to compute the empirical size and power under different alternatives. Finally, we implement the proposed test using two real data sets.
KeywordsPareto distribution Ratio of order statistics Regularly varying tailed distribution Tail of a distribution
The authors would like to thank the anonymous referee(s) and Editor-in-Chief for their suggestions and comments, which lead to improvement of the presentation.
There is no funding from any source for writing this article.
Compliance with ethical standards
Conflict of Interest
There is no conflict of interest among the author while publishing this article.
- Bettini C (1987) Forecasting populations of undiscovered oil fields with the log-Pareto distribution. Stanford University, StanfordGoogle Scholar
- Jung J, Berger AW, Balakrishnan H (2003) Modelling TTL-based Internet caches. In: INFOCOM 2003. Twenty-second annual joint conference of the IEEE computer and communications, vol 1. IEEE Societies, pp 417–426Google Scholar
- Neves C, Frages Alves I (2008) The ratio of maximum to the sum for testing super heavy tails. In: Arnold B, Balakrishnan N, Sarabia J, M’nguez R (eds) Advances in mathematical and statistical modelling. Birkhäuser, BostonGoogle Scholar
- Provost SB, Mabrouk I (2011) An extended inverse Gaussian model. Int J Stat Sci 11:135–144Google Scholar