Loss Reserving

Loss reserving, the process of estimating future insurance claim payouts, aims to accurately predict outstanding liabilities and ensure financial stability. Recent research emphasizes improving prediction accuracy by incorporating granular claim-level data and leveraging advanced machine learning techniques, such as recurrent neural networks (including LSTMs) and Bayesian model averaging with LASSO, to better capture both temporal patterns and model uncertainty. These advancements address limitations of traditional aggregate methods, leading to more robust and reliable loss reserve estimations with better quantification of uncertainty. This improved accuracy has significant implications for insurance companies' solvency, pricing strategies, and regulatory compliance.

Papers

January 8, 2024

Enhancing Adaptive History Reserving by Spiking Convolutional Block Attention Module in Recurrent Neural Networks
Qi Xu, Yuyuan Gao, Jiangrong Shen, Yaxin Li, Xuming Ran, Huajin Tang, Gang Pan
Recurrent Neural Network Spiking Neural Network Spatio Temporal Temporal Feature Convolutional Block Attention Convolutional Structure Loss Reserving

September 30, 2022

Model error and its estimation, with particular application to loss reserving
G Taylor, G McGuire
Financial Application Estimation Task Model Error Bayesian Model Averaging Forecast Error Loss Reserving

January 27, 2022

Micro-level Reserving for General Insurance Claims using a Long Short-Term Memory Network
Ihsan Chaoubi, Camille Besse, Hélène Cossette, Marie-Pier Côté
Long Short Term Memory Real World Claim Insurance Data Loss Reserving

Loss Reserving

Papers

Enhancing Adaptive History Reserving by Spiking Convolutional Block Attention Module in Recurrent Neural Networks

Model error and its estimation, with particular application to loss reserving

Micro-level Reserving for General Insurance Claims using a Long Short-Term Memory Network