Modeling Heterogeneous Farm Responses to European Union Biofuel Support with a Random Parameter Multicrop Model

Abstract

Although there is now widespread evidence of substantial variability in economic agents' responses to economic drivers in many applied economics fields, this variability has been largely overlooked by econometric agricultural production models. This article sets out to fill this gap by providing methodological contributions and empirical results. First, we consider panel data multicrop models featuring random intercept and slope parameters to account for the heterogeneous responses of crop producers to economic drivers. Second, we show that Monte Carlo expectation‐maximization algorithms are particularly well‐suited to estimating this type of model. Third, based on an application of our empirical modeling framework with a sample of French grain crop producers, we demonstrate substantial variability in farmers' responses to economic incentives. Fourth, we use the estimated model and a simple “statistical calibration” procedure to build farm‐specific simulation models, which are then used to evaluate the effects of the rapeseed price increase induced by European Union (EU) biofuel support. Our simulation results demonstrate that ignoring the variability in the considered farmers' responses to the economic incentives results in significant overestimation of the increases in rapeseed yield levels and variable input use levels induced by EU biofuel support, as well as significant underestimation of the variability in the congruent increases in rapeseed acreages.

Modeling Corners, Kinks, and Jumps in Crop Acreage Choices: Impacts of the EU Support to Protein Crops

“…vectors are independently distributed across time.Combined with the fact that vector it z doesn't contain any lagged endogenous variable, this serial independence assumption implies that our MEMC model can be interpreted as a reduced form model as regards the dynamic features of the modelled choices. Indeed, we hypothesize that random parameters i γ capture the effects on farmers' production choices and performances of the stable crop rotation schemes that these farmers rely on.15 Koutchadé et al (2018) provide empirical results confirming this hypothesis with a sample of arable crop producers located in an area contiguous to the one considered in our application.…”

“…Ranges of expectations of the acreage flexibility parameters are theoretically consistent. Conditions Once we have estimated the parameters characterizing the distribution of the random parameters i γ , we can "statistically calibrate" those parameters for each farmer in our sample and thus obtain a set of farmer specific "calibrated" models that can then be used for simulation purposes (Koutchadé et al, 2018). In this study, the specific parameter i The estimated farmer specific models allow us to compute fitting criteria, Sim-R², which are reported in Table 5.…”

“…As in Koutchadé et al (2018) we adopt a random parameter approach for accounting for farmers' and farms' unobserved heterogeneity. We assume that the parameters of farmers' production choices, including those driving farmers' responses to economic incentives, are farm specific.…”

“…Relying on the MNL acreage choice models also enables us, following Koutchadé et al (2018), to go one step further and to account for farmers' unobserved heterogeneity by considering a random parameter version of our ERS-MEMC model. Estimating ERS models with multiple regimes is challenging mostly because their likelihood function involves integration of expectations over the probability distribution of multivariate latent error terms (e.g., Pudney, 1989).…”

Do Risk Preferences Really Matter? The Case of Pesticide Use in Agriculture

“…We used annual data on winter wheat over the period 1993-2010. For greater details on these data, see e.g., Carpentier andLetort (2011), Femenia andLetort (2016), and Koutchadé, Carpentier and Femenia (2018).…”