Empirical analysis in South African agricultural economics and the R-Square disease

Abstract

The South African agricultural sector underwent a significant amount of institutional and structural changes during the past two decades, especially in the aftermath South Africa’s first democratic elections in 1994 and the deregulation of the agricultural marketing environment in 1996/97. These changes meant that South African agricultural economics scholars had to adapt to these changes. The increased need towards more quantified output in agricultural economic research has led agricultural economic scholars to “borrow” econometric models from their fellow scholars abroad to apply to South African research problems in order to fulfil the need for more quantified research output. However, the development of econometrics has over the years given rise to a disenchantment with the way in which econometrics have been applied in economic research. Consequently it is believed that a large body of literature has entered the public domain without being properly reviewed because South African agricultural economic scholars do not have the necessary insight and knowledge of the problems believed to be at the root of the disenchantment with the manner in which econometrics have been applied. The general objective of this dissertation is to investigate the disenchantment with the manner econometrics has been applied in economic and agricultural economic scholarship in order to identity the main drivers of this disenchantment, and to use this knowledge gained to evaluate the application of econometrics in South African agricultural economic scholarship as portrayed in Agrekon, one of South Africa’s agricultural economics peer review journals. The study is conducted by means of a review of the literature on the history of econometrics, the development of econometric methodologies and the disenchantment with econometrics in economics and agricultural economics. Applied econometrics portrayed in Agrekon is evaluated by means of a survey of papers published in this journal. The main findings of this study revealed that the key drivers of disenchantment can mainly be ascribed to the following: <ul> <li>The misuse of statistical significance tests in applied studies.</li> <li>Problems with data underlying econometric analyses.</li> <li>The problems associated with replication. <li>Data mining <li>The “Black box ideology” in applied econometrics and <li>Scholasticism and associated preference falsification.</li> </ul> A survey of papers published in Agrekon based on a sample of 65 papers, which were sampled by means of stratified random sampling, revealed that elements behind the disenchantment with econometrics are present in South African agricultural economic scholarship. It was also found that the data underlying econometric analyses are a major point of concern in South African agricultural economics and it seem as if South African agricultural economics scholars have adopted a lackadaisical attitude towards data. The study concludes with recommendations for future studies into to the application of econometrics in South African agricultural economics.