Modeling for pesticide productivity measurement

This paper revisits claims about implications of production function specification for pesticide productivity measurement and presents two extensions of the popular damage control specification, along with an empirical application. One extension eliminates bias caused by relying solely on economic data and shows how to include variables that represent the pest population when they are necessary to avoid bias. The second extension allows for the possibility of phytotoxicity. These extensions generalize the damage control specification by eliminating bias and allowing for a range over which the first and third stages of production may occur. The main contributions of the analysis are to clarify existing misconceptions about pesticide productivity modeling and to provide extensions to the damage control specification that permit greater realism for empirical analysis.