Estimating the Probability of Surrender Claims on Insurance Companies using the Point Process Model

Abstract

This study discusses the estimation of surrender claims in an insurance company. A surrender claim is a claim filed due to default by the insured (a person who terminates/exits the insurance before the maturity date or when an insured risk occurs). Surrender claims are also difficult to predict because they occur randomly. Therefore, this study uses a point process model to estimate the probability of surrender claims. Point process is a stochastic process that can explain random events in both space and time. Point process is characterized based on its conditional intensity. In this study, the conditional intensity of Point process is viewed as a renewal process called Hazard Rate. The estimation method used is the maximum likelihood temporal point process constructed with the Bernoulli process, where the time between events is divided into narrow intervals that are viewed as binomial trials. Thus, the Hazard Rate estimation results and the probability of future claims are obtained based on the Hazard Rate function of the time between claims arrivals, which is lognormally distributed. This study shows that the longer the time between claim arrivals, the greater the probability of a claim occurring in that time interval.