Tianxing Yan

Title: Dependence modeling for heavy-tailed multi-peril insurance losses
Date: May 29th, 2024
Time: 10am
Location: HAL 123 & Zoom
Supervised by: Dr. Himchan Jeong and Dr. Yi Lu

Abstract: Assessing and managing risks is essential for insurance companies. We recognize the heavy-tailed behaviour and dependency among different coverages in insurance claim datasets. To capture claim dependency, we proposed a hierarchical model and several copula-based models. Composite models are applied to address the heavy-tailed behaviour of individual losses. To evaluate the performance of the proposed model from the insurance aspect, we approximate the risk measures using the Monte Carlo methodology. Finally, we demonstrate that the model considering dependency enhances model goodness-of-fit while providing more accurate risk measures of the aggregated losses for all types of coverage in total.