Damage control productivity: why econometrics matters

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. Key words: pesticides, productivity, panel data, damage control. The distinct roles of direct versus damage con-trol inputs would seem to warrant estimating marginal productivities based on production functions that involve asymmetric functional roles for the two types of inputs. The impor-tance of such an asymmetric specification has received considerable attention in studies of the productivity of pesticides in agriculture. Lichtenberg and Zilberman (LZ) hypothesized that earlier econometric estimates reported by Headley (1968) and Carlson were biased be-cause of their symmetric treatment of pesticides in the production function. LZ proposed use of the asymmetric specification found in past nor-mative studies and showed that if the true pro-duction function had such an asymmetric form, then use of other functional forms which treat pesticides symmetrically would result in up-ward biased estimates of the productivity of pesticides. Misspecification of ecosystem pro-cesses or the roles of ecosystem state variables may also result in biased estimates (Blackwell and Pagoulatos). While these papers reiterate the simple truth that specification matters, the true functional form remains unknown. For this reason, the LZ In alphabetical order, the authors are respectively, researcher