Parametric latent class analysis with färe-primont TFP decomposition: application to french grazing livestock farms

The objective of the article is to extend the latent class approach currently available to measure efficiency with stochastic frontier analysis (SFA), to the measurement of total factor productivity (TFP) and its components using Färe-Primont index which is multiplicatively complete and transitive. The application will be to French grazing livestock farms. From an empirical point of view, the applications of the Färe-Primont TFP index to agriculture are rare. From a methodological point of view, the extension of the econometric latent class model to Färe-Primont TFP index has not been done so far. We apply the new methodology to French beef cattle farms, dairy farms, sheep or goat farms over the period 2002-2015. For each of the three types of farms, we will classify farms based on their pollution intensity that is to say the use of polluting inputs (fertilizers, pesticides, veterinary costs, energy)