Nonparametric and Semiparametric Estimation: An Analysis of Multiproduct Returns to Scale

The nonparametric and semiparametric regression models based on kernel density estimation are reviewed, and they are applied to the analysis of returns to scale in dairy farms using a multioutput cost function. Semiparametric tests of constant returns to scale and homogeneity of production are developed, and their implementation show that both of these hypotheses are rejected. Kernel estimation of the local degree of multiproduct returns to scale confirms that the sample's dairy farms are characterized by a considerable degree of scale economies.