Spatial Pattern of Yield Distributions: Implications for Crop Insurance
研究了作物保险中因天气空间相关导致的风险非独立性问题，提出了一种跨区域风险池的统计检验方法，并用美国县级玉米产量数据验证了该方法能提升保险定价的准确性。
Abstract Crop insurance is similar to flood and hurricane insurance in that spatially correlated weather tends to cause violations of the independence assumption. Ideally, one would seek to pool uncorrelated risk drawn from the same distribution in crop insurance. This article proposes a testing procedure for the cross‐sectional pooling of group units, and empirically analyzes whether the proposed test improves out‐of‐sample rating performance. We utilize a balanced panel of U.S. county‐level corn yields for 510 counties, and the results of an out‐of‐sample crop insurance rating performance exercise provide economic significance to the proposed pooling methodology and results.