Accounting for Latent Cropping Management Practices Choices in Crop Production Models: a Random Parameter Hidden Markov Model approach

In this article, we account for cropping management practices (CMPs) in economists' production functions to evaluate pesticide uses responsiveness to price changes in a context of heterogeneous technology. CMPs being latent in most economists' datasets and CMP changes entailing adjustment costs, we consider a hidden Markov model to describe the dynamics of farmer's CMP choice. We also account for farmers' unobserved heterogeneity by considering a random parameter model for our production function. An illustration on French winter wheat producers of La Marne area uncovers very high-yielding, high-yielding and low-input CMPs. The characteristics of the low-input CMPs we uncover are very close to those tested by agronomists in the area covered by our data. We also show that input uses differences between low-input and more conventional CMPs are too small for taxes on chemical inputs to imply large relative profitability effects and thus to encourage farmers to adopt less intensive practices.