Analysis of Robust Chain Ladder Method in Estimating Australian Motor Insurance Reserves With Outlying Dataset

Abstract

Reserves are one of the most crucial components for an insurance company to make sure it has enough money to pay off all the incurred claims. The presence of outliers in the incurred claims data harbors risk on inaccurately predicting reserves to cover claim amounts, usually achieved by the standard chain ladder reserving method. To remedy the effect of the outliers, the robust chain ladder reserving method is used by setting the median value to predict estimated reserve. On this research, we utilized both methods on various datasets. The purpose of this paper is to determine the best method that can be utilized by insurance company in various scenario to obtain the most optimized reserved estimate that can minimize the risk of being unable to pay the insurance claim or even the risk of over allocating reserves that could pose profitability issue. The primary data used are the Australian domestic motor insurance claims from 2012 to 2017, obtained from Australian Prudential Regulation Authority (APRA). The dataset is then manipulated to have outliers. After calculating the estimation, the result is compared to assess the strength of the methods using Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) calculation. In conclusion, we found that the robust chain ladder reserving method works better in an outlying dataset. We also identify cases in which robust chain ladder are not appropriately used.