Are Crop Yields Normally Distributed? A Reexamination
发现对短期面板数据逐个体去势会严重降低正态性检验的效力并导致第二类错误，提出一种替代的误差分量隐含去势方法，并应用于大型数据集，结果显示产量残差的正态性被拒绝，且假设正态性会降低现有作物保险产品中大量生产者的潜在保费率。
This article demonstrates that normality test procedures that include individual detrending of short‐term panel data can severely reduce the power of normality tests and strongly bias normality tests in a Type II direction. An alternative error component implicit detrending procedure is suggested that demonstrates higher power for the distributions examined. Both procedures are applied to a large data set with normality of yield residuals being rejected. Assuming normality is shown to reduce potential premium rates for a large number of producers in an existing crop insurance product.