Neyman Orthogonalized Specification Tests for Conditional Tail Index Models
Abstract
We develop specification tests for parametric conditional tail index models of heavy-tailed distributions. The tests are based on a projected empirical process that removes the first-order effect of estimating the null parameter. Within the conditional Hall class, we identify conditions under which projection also eliminates the leading Pareto approximation bias in the test process, permitting lower thresholds while preserving asymptotically correct size. We establish the validity of the tests via multiplier bootstrap. For log-linear models, bootstrap validity extends to settings in which covariates among observations above a common threshold concentrate and the smallest eigenvalue of the information matrix tends to zero. We further propose a data-driven threshold selection rule based on an estimate of the Pareto approximation bias after projection. The rule allows lower thresholds to retain more observations for testing while limiting size distortions, with potential gains in power. The finite sample performance of the tests is examined through simulations and a real data application.
Related
- Sequential versus fully conditional specification for tree-based synthesis of health data
- Embedded Conditional Independence Tests for Large Language Model Generated Text with an Application to German Parliament Speeches
- Grouped Semi-Supervised Estimation of Two-Sample Functionals via Single-Index Conditional Imputation
- Higher-Order Neyman Orthogonality in Moment-Condition Models
- A Partial-Matching Proof of the Neyman--Pearson Lemma
BibTeX
@misc{peng2026neyman,
title = {{Neyman Orthogonalized Specification Tests for Conditional Tail Index Models}},
author = {Liuhua Peng and Shuo Li and Xiaojun Song},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.07219},
eprint = {2611.07219},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.07219},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Neyman Orthogonalized Specification Tests for Conditional Tail Index Models
AU - Liuhua Peng
AU - Shuo Li
AU - Xiaojun Song
PY - 2026
DA - 2026/11/04
PB - arXiv
DO - 10.48550/arxiv.2611.07219
UR - https://arxiv.org/abs/2611.07219
AB - We develop specification tests for parametric conditional tail index models of heavy-tailed distributions. The tests are based on a projected empirical process that removes the first-order effect of estimating the null parameter. Within the conditional Hall class, we identify conditions under which projection also eliminates the leading Pareto approximation bias in the test process, permitting lower thresholds while preserving asymptotically correct size. We establish the validity of the tests via multiplier bootstrap. For log-linear models, bootstrap validity extends to settings in which covariates among observations above a common threshold concentrate and the smallest eigenvalue of the information matrix tends to zero. We further propose a data-driven threshold selection rule based on an estimate of the Pareto approximation bias after projection. The rule allows lower thresholds to retain more observations for testing while limiting size distortions, with potential gains in power. The finite sample performance of the tests is examined through simulations and a real data application.
ER -