IBNR Reserve Estimation Using Chain-Ladder and Bootstrap Chain-Ladder Methods

Abstract

The Incurred But Not Reported (IBNR) claims reserve is a crucial component in maintaining the financial stability of insurance companies. However, traditional deterministic approaches such as the Chain-Ladder method provide point estimates based on historical claim development patterns but do not capture the uncertainty inherent in future claim outcomes. As claim data often exhibit variability across development periods, stochastic approaches are required to quantify this uncertainty and improve the reliability of reserve estimates. Among various stochastic reserving methods, the bootstrap approach is particularly suitable due to its simplicity and direct compatibility with the traditional Chain-Ladder framework. However, empirical applications of bootstrap-based reserving methods in Indonesian motor vehicle insurance remain limited. This study addresses this gap by applying the Bootstrap Chain-Ladder method as a stochastic extension and comparing its performance with the conventional deterministic approach. Using semiannual motor vehicle claim data from PT Asuransi XYZ over the period 2019–2023, the deterministic Chain-Ladder method produced an IBNR estimate of Rp 40.17 billion, while the Bootstrap Chain-Ladder resulted in an estimate of Rp 40.94 billion, reflecting the incorporation of variability in the estimation process. Although the deterministic approach achieved slightly lower prediction errors, as measured by Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE), the bootstrap method generated a more conservative estimate and provided additional insight into reserve uncertainty. These findings suggest that integrating stochastic elements into traditional reserving frameworks enhances financial prudence while maintaining practical applicability in the general insurance industry.