Using non-parametric methods in econometric production analysis: an application to Polish family farms

Research output: Contribution to conferencePaperResearch

Econometric estimation of production functions is one of the most common methods in applied economic production analysis. These studies usually apply parametric estimation techniques, which obligate the researcher to specify the functional form of the production function. Most often, the Cobb-Douglas or the Translog production function is used. However, the specification of a functional form for the production function involves the risk of specifying a functional form that is not similar to the “true” relationship between the inputs and the output. This misspecification might result in biased estimation results—including measures that are of interest of applied economists, such as elasticities. Therefore, we propose to use nonparametric econometric methods. First, they can be applied to verify the functional form used in parametric estimations of production functions. Second, they can be directly used for estimating production functions without specifying a functional form and thus, avoiding possible misspecification errors. We use a balanced panel data set of farms specialized in crop production that is constructed from Polish FADN data for the years 2004-2007. Our analysis shows that neither the Cobb-Douglas function nor the Translog function are consistent with the “true” relationship between the inputs and the output in our data set. We solve this problem by using non-parametric regression. This approach delivers reasonable results, which are on average not too different from the results of the parametric estimations but many individual results are rather different.
Original languageEnglish
Publication date2011
Number of pages12
Publication statusPublished - 2011
EventCongress of the European Association of Agricultural Economists 2011 - Zürich, Switzerland
Duration: 30 Aug 20112 Sep 2011


ConferenceCongress of the European Association of Agricultural Economists 2011

Number of downloads are based on statistics from Google Scholar and

No data available

ID: 35162415