Risk effects of GM corn: Evidence from crop insurance outcomes and high‐dimensional methods
利用作物保险损失成本比衡量产量风险，采用Cluster-Lasso高维方法控制天气变量后，发现抗根虫转基因玉米杂交种的产量风险更低。
Abstract This study evaluates whether genetically modified (GM) corn hybrids with rootworm resistant traits (GM‐RW) have lower yield risk. A crop insurance actuarial performance measure, the loss cost ratio (LCR), is used to represent yield risk. High‐dimensional methods are utilized in this study to maintain parsimony in the empirical specification, and facilitate estimation. Specifically, we employ the Cluster‐Lasso (cluster‐least absolute shrinkage and selection operator) procedure. This method produces uniformly valid inference on the main variable of interest (i.e., the GM‐RW variable) in a high‐dimensional panel data setting even in the presence of heteroskedastic, non‐Gaussian, and clustered error structures. After controlling for a large set of potential weather confounders using Cluster‐Lasso, we find consistent evidence that GM corn hybrids with rootworm resistant traits have lower yield risk.