Comparative analysis of neural networks and traditional actuarial methods for estimating casualty insurance reserve liability

The primary purpose of this study is to explore the use of artificial neural network methods for estimating insurance company claim reserve liability. This is accomplished by comparing the performance of artificial neural network methods to several traditional actuarial loss development methods using both industry data and simulated data. The second objective is to investigate differences in results from testing methods over industry data vs. simulated data. The third objective is to identify specific data attributes that are associated with estimation error for various methods. The goal is to determine if particular methods perform better when specific data characteristics exist. This is accomplished by correlating data attribute measures with method error across data sets and methods.