Intertemporal and interfirm heterogeneity: Implications for pesticide productivity

This paper presents evidence concerning the existence and implications of heterogeneity effects for estimates of the marginal productivity of pesticides. The paper extends the work of Carpentier and Weaver (1996a), which considered fixed effects that homothetically shift the conditional mean of output. This paper considers the implications of time-varying marginal productivity, allowing for nonhomothetic shifts in production. The extent and structure of heterogeneity found in French crop agriculture is first presented to illustrate that persistent heterogeneity exists across these firms. Next, we review the implications of such heterogeneity for econometric estimates and show that estimates of marginal productivity of pesticides will be biased when based on models that do not explicitly account for these types of heterogeneity across firms and time. An alternative estimation approach is proposed, based on Chamberlain (1982 and 1984) and implemented to produce asymptotically unbiased estimates and to statistically validate the model specification. Existence of fixed firm effects is statistically confirmed and temporal variation of marginal productivity is found to be substantial.