Research on Ruin Probability of Risk Model Based on AR(1) Time Series

Abstract

The insurance industry typically exploits ruin theory on collected data to gain more profits. However, state-of-art approaches fail to consider the dependency of the intensity of claim numbers, resulting in the loss of accuracy. In this work, we establish a new risk model based on traditional AR(1) time series, and propose a fine-gained insurance model which has a dependent data structure. We leverage Newton iteration method to figure out the adjustment coefficient and evaluate the exponential upper bound of the ruin probability. We claim that our model significantly improves the precision of insurance model and explores an interesting direction for future research.