Using neural networks as a support tool in the decision making for insurance industry

Abstract

Under the pressure of market competition, property and casualty insurance companies normally use price competition as a means to enlarge their market share. Traditionally, the fire insurance underwriters use the interpolation method to estimate in-between risks (the classification of exposure falling between two risk levels). This interpolation method gives the underwriter more discretion in pricing to compete in the market. However, problems such as cutthroat competition or adverse selections also play a role, because the interpolation method cannot provide the correct price reflecting these in-between risks. This paper proposes the back propagation neural network (BPNN) model as a tool for the underwriter to determine the proper premium rate of in-between risks. A detailed explanation of how the BPNN model solves problems caused by traditional interpolation method is provided.