THE INFLUENCE OF DEPENDENCE ON DATA NETWORK MODELS OF BURSTINESS

V. Kulkarni and Sidney Resnick

A company wishes to estimate or predict its financial exposure in a reporting period of length T (typically one quarter) due to warranty claims. We propose a fairly general random measure model which allows computation of the Laplace transform of the total claim made against the company in the reporting interval due to warranty claims. When specialized to a Poisson process of both sales and warranty claims, statistical estimation of relevent quantities is possible. The methodology is illustrated by analyzing a year's worth of data on automobile sales and claims from Company X.