Semiparametric Exponential Families for Heavy-Tailed Data
Preprint
Abstract
We propose a semiparametric method for estimating the mean of a heavy tailed population given a relatively small sample from that population and a larger sample from a related background population. We model the tail of the small sample as an exponential tilt of the better-observed large-sample tail, using a robust sufficient statistic motivated by extreme value theory. We show that our method can substantially outperform the sample mean and Winsorized mean. If the small and large samples are both drawn from regularly varying distributions with the same tail index, we show under mild additional conditions that our method achieves a better rate of convergence than the optimally Winsorized mean.Keywords
All Related Versions
This publication has 15 references indexed in Scilit:
- A Diagnostic for Selecting the Threshold in Extreme Value AnalysisJournal of the Royal Statistical Society Series B: Statistical Methodology, 2001
- Using a Bootstrap Method to Choose the Sample Fraction in Tail Index EstimationJournal of Multivariate Analysis, 2001
- An Introduction to Statistical Modeling of Extreme Values2001
- Using specially designed exponential families for density estimationThe Annals of Statistics, 1996
- Models for Exceedances Over High ThresholdsJournal of the Royal Statistical Society Series B: Statistical Methodology, 1990
- Using the bootstrap to estimate mean squared error and select smoothing parameter in nonparametric problemsJournal of Multivariate Analysis, 1990
- On Some Simple Estimates of an Exponent of Regular VariationJournal of the Royal Statistical Society Series B: Statistical Methodology, 1982
- A Simple General Approach to Inference About the Tail of a DistributionThe Annals of Statistics, 1975
- Residual Life Time at Great AgeThe Annals of Probability, 1974
- Robust Estimation of a Location ParameterThe Annals of Mathematical Statistics, 1964