Comparison between parametric and non-parametric methods of estimating comprehensive motor claim severity distributions

Abstract

Claim severity is the amount of loss associated with an insurance claim. Insurers compensate policyholders who have suffered a loss from the occurrence of an insured risk. Insurance companies have been estimating claim severity by using normal distribution meaning; they assume an average cost of motor claims to estimate the total claims amount. However, this method is not very efficient because not all motor claims follow a normal distribution. To deal with this, there has been an introduction to using other parametric distributions such as the gamma and log-normal distribution. Parametric distributions do not consider the outlier claims that do not follow any of the parametric distributions and this is what led to using non-parametric distributions. The data used in this research study consisted of an auto-insurance portfolio of a company operating in Sweden, which was compiled by the Swedish Committee on the Analysis of Risk Premium in Motor Insurance, (Hallin & Ingenbleek, 1983). The motor insurance data is cross sectional and it involves the third-party liability auto-insurance claims for the year 1977. The only variable I worked with were the claim amounts. The main aim of this research study was to employ both parametric and non-parametric models in estimating the claim size distribution. From the data analysis that was carried out, it can be concluded that the non-parametric method is the most suitable one for estimating motor claims severity distributions .