Damage Control Productivity: Why Econometrics Matters
指出在非实验数据中，损害控制投入的规格设定会限制技术结构，并提出一种对称形式来估计损害控制生产率，利用面板数据方法消除异质性偏差。以法国谷物农场农药生产率的估计为例，发现固定效应很重要，且农药的边际生产率远低于以往研究。
Abstract For nonexperimental data, asymmetric specifications of the role of damage control inputs impose restrictions of homotheticity or nonjointness on the underlying technology. A generalized symmetric form where damage control inputs affect factor diminution functions motivates use of symmetric specifications. Panel data and estimation methods are presented to produce robust estimates of damage control productivity that are free of heterogeneity bias. Estimates of the productivity of pesticides for French cereal farm production illustrate that fixed firm and time effects are important. Estimates indicate substantially smaller marginal productivity of pesticides than reported by past studies.