The Application of Robust Regression to a Production Function Comparison
Abstract
The adequate representation of crop response functions is crucial for agronomic as well as agricultural economic modeling and analysis. So far, the evaluation of such functions focused on the comparison of different functional forms. In this article, the perspective is expanded also by considering different regression methods. This is motivated by the fact that exceptional crop yield observations (outliers) can cause misleading results if least squares regression is applied. In order to address this problem we also apply robust regression techniques that are not affected by such outliers. We evaluate the quadratic, the square root and the Mitscherlich-Baule function using the example of Swiss corn (Zea mays L.) yields. It shows that the use of robust regression narrows the range of optimal input levels across different functional forms and reduces potential costs of misspecification compared to least squares estimation. Thus, differences between functional forms are reduced by applying robust regression.