Improved index insurance design and yield estimation using a dynamic factor forecasting approach

Abstract

Accurate crop yield forecasting is central to effective risk management for many stakeholders, including farmers, insurers, and governments, in various practices, such as crop management, sales and marketing, insurance policy design, premium rate setting, and reserving. This paper first investigates an innovative approach of yield forecasting using a dynamic factor model. Based on the proposed approach, we then design an enhanced weather index-based insurance (IBI) policy. The dynamic factor approach is motivated by its ability to effectively summarize the information in a large set of explanatory variables with common factors of a much lower dimension. This makes it possible to use an extensive set of variables in crop yield prediction without worrying about identification issues. Using both county-level and state-level crop production data from the state of Illinois, U.S., the empirical results show that the dynamic factor approach produces more accurate in- and out-of-sample forecasting results compared to the classical statistical models. The empirical results also support that the proposed IBI policy based on the dynamic forecasting model has small basis risk. This, in turn, greatly improves the IBI's hedge effectiveness against agricultural production as well as increases its practicality as an insurance policy for agriculture.