ESTIMATION OF IBNR CLAIMS RESERVES USING CHAIN LADDER METHOD AND GENERALIZED LINEAR MODEL (GLM) WITH OVER-DISPERSED POISSON (ODP) DISTRIBUTION

Abstract

Within the insurance industry, the calculation of reserves for predicting fund allocation becomes highly prominent, as companies must do it accurately to settle future claims. Consequently, insurance firms enhance financial security, ensuring sufficient resources are available to meet obligations, particularly during periods of increased claims activity. This research discusses the calculation of claim reserve estimates using Chain Ladder (CL) method and the Generalized Linear Model (GLM) with Over-Dispersed Poisson (ODP) distribution. Leveraging the simplicity of the CL for baseline projections and the flexibility of GLMs for refining estimates based on additional predictors. Both models yield estimation results for the run-off triangle, illustrating the expected development of claims over a specific period. The case study used in this research involves Auto data from 2010 to 2019 at Zurich Insurance Company in the North America region. The results of calculations using the CL method and GLM with the ODP approach give the same results. Estimates using the ODP approach provide a confidence interval between USD 1,479,610 and USD 1,480,253. The prediction error with MAPE results is the same for both CL and GLM method which is 42,81%. Based on these results, it can be concluded that insurance companies can use the GLM method as a method that provides confidence intervals for making insurance decisions in providing claims reserves.