An R Package For Crossed Classification Credibility Model: Application Regarding Non-Performing Loan

Abstract

Credibility theory, which is used to determine the premium in the non-life branches of insurance, is a calculation method which is used for making weighted estimation of balanced allocation between past and recent period data. The procedure of weighting is done with the Z credibility factor. There are miscellaneous methods which are named as credibility models to determine Z value. One of these models is Crossed Classification Credibility Model, which is introduced by Dannenburg (1995). In this model, an insurance portfolio is subdivided by two qualitative risk factors, modeled in symmetrical way. Especially, this model offers an alternative method when data are unclassifiable hierarchically. Simultaneously, this model considers the joint and separates the effects of risk factors. To predict the premiums in this model, variance components are obtained by solving the linear equation system must be calculated. However, this system cannot be solved explicitly. Also, too many parameters must be calculated for the premium estimation. Here, calculation errors can occur, and it is very difficult to find the correct results. Moreover, there is no tool that can easily perform these operations on a computer. In this study, the R package cccm has been developed to calculate the structural parameters easily, quickly, and accurately for Crossed Classification Credibility Model. Package cccm explained step by step for the users interested in to solve Crossed Classification Credibility problems.