Research on Agricultural Drought Risk Measurement Based on Bayes Hybrid Model

Abstract

The development of agricultural economy depends to a large extent on the drought. It is necessary to accurately analyze the current drought risk in order to formulate a more reliable drought risk management strategy and reduce the impact of disasters on the development of the agricultural economy. In order to improve the level of drought risk measurement, this paper selects VaR as the measurement tool, and proposes a mixed distribution model research. Use this model to fit the distribution of drought loss rate, and measure the drought risk by estimating VaR. Among them, the mixed distribution model is mainly composed of two parts, namely GPD and conventional distribution. The former is used to characterize the risk tail. Considering the difficulty of selecting the GPD distribution threshold, this paper introduces the Bayes calculation method to optimize, forming a Bayes hybrid model, including Norm-GDP model and Gamma-GPD model. The application results show that the fitting results generated by the Norm-GDP model application have a better distribution of drought loss rates, and the VaR estimation results are more reliable. Taking 10-year, 20-year, and 100-year disasters as examples, the estimated drought loss rate is 9.46%, 11.05%, and 30.22%. The generation of these metric values can provide a reference for my country's agricultural drought risk management.