Maximum weighted likelihood estimator for robust heavy-tail modelling of finite mixture models

TL;DR

This paper introduces the maximum weighted likelihood estimator (MWLE) for robustly estimating heavy-tail finite mixture models, especially in insurance claim severity data, demonstrating its consistency and robustness over traditional MLE.

Contribution

The paper develops MWLE for heavy-tail FMM, proving its consistency and robustness, and adapts the GEM algorithm for efficient estimation in complex mixture models.

Findings

MWLE outperforms MLE in tail estimation accuracy.

MWLE maintains robustness even under model misspecification.

Application to real insurance data confirms improved tail estimates.

Abstract

In this article, we present the maximum weighted likelihood estimator (MWLE) for robust estimations of heavy-tail finite mixture models (FMM). This is motivated by the complex distributional phenomena of insurance claim severity data, where flexible density estimation tools such as FMM are needed but MLE often produces unstable tail estimates under FMM. Under some regularity conditions, MWLE is proved to be consistent and asymptotically normal. We further prove that the tail index obtained by MWLE is consistent even if the model is misspecified, justifying the robustness of MWLE in estimating the tail part of FMM. With a probabilistic interpretation for MWLE, Generalized Expectation-Maximization (GEM) algorithm is still applicable for efficient parameter estimations. We therefore present and compare two distinctive constructions of complete data to implement the GEM algorithm. By exemplifying our approach on two simulation studies and a real motor insurance data set, we show that comparing to MLE, MWLE produces more appropriate estimations on the tail part of FMM, without much sacrificing the flexibility of FMM in capturing the body part.