This paper reviews the hierarchical credibility model as an extension of the Bühlmann-Straub model. It also describes credibility estimation for the premium using assumptions from the Bühlmann-Straub model at the particular levels of the hierarchical structure. The paper also includes an application of credibility theory and techniques for motor vehicle third-party liability insurance. (fragment of text)
Przedstawiono w sposób uporządkowany teorię zaufania, z uwzględnieniem najnowszych wyników oraz dokonano syntezy wyników zamieszczonych w artykułach własnych i innych autorów. Przedstawiono nowe bądź uzupełnione dowody znanych twierdzeń. Sformułowano nowe twierdzenia, tak aby przedstawiona teoria była kompletna. Teorię zilustrowano przykładami liczbowymi, wskazującymi zarówno na jej użyteczność, jak i ograniczenia.
EN
This article gives a structured overview of the development of credibility theory, which is one of the most important fields in insurance. In order to achieve this, a uniform approach and notation have been introduced and a synthesis of the results published by other authors carried out. Concepts and theorems related to linear Bayesian predictors of nett premiums are presented. A review is made regarding distribution pairs: the conditional distribution of a random variable describing a claim and the distribution of a random variable describing the structure of a non-homogeneous portfolio, whose Bayesian predictor is a predicator of credibility. Various practical models of credibility are presented, such as the Bühlmann model, together with it generalisation taking into account the importance of contracts, i.e. the Bühlmann-Straub model. Jewell's model enables us to consider a portfolio with a hierarchical structure. The case in which we observe a linear trend in the observations is considered in Hachemaister's model. Predictors of the credibility of nett premiums and estimators of parameters describing the structure of the population are derived using each of these models. In this article, for the first time statistical discrimination analysis is applied to the problem of classifying those insured in non-homogeneous portfolia. The theory presented is illustrated using simulated data. These examples show that this theory has great potential to be applied in insurance businesses. (original abstract)
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.