The main objective of this work is to examine the hypothesis of the inverse relationship between farm size and productivity in Brazil using non-parametric kernel methods. While parametric estimators are considered global, kernel regressions use a sample of data close to a point to adjust the estimation, which allows focusing on local peculiarities of the data distribution, allowing the data “to speak for themselves”. Both methods are applied and compared using aggregated data from the 2006 Agricultural Census. Among the main results, it was observed that the hypothesis of the inverse relationship between area and productivity was not corroborated in the fully parametric analysis; however, when the different non-parametric estimators are compared, the inverse relationship hypothesis is observed for small farmers, but not for medium and large farmers in most regions of the country.
inverse relationship; productivity; farm size; kernel regression