AGRICULTURAL ECONOMICS Agricultural Economics 35 (2006) 171–181 Economic versus environmental improvement potentials of Danish pig farms Mette Asmild a, ∗ , Jens Leth Hougaard b a Department of Economics and Natural Resources, The Royal Agricultural University, Copenhagen, Denmark and Industrial Economics Division, Nottingham University Business School, Jubilee Campus, Wollaton Road, Nottingham, NG8 1BB, UK b Department of Economics, University of Copenhagen, Copenhagen, Denmark Received 29 January 2004; received in revised form 10 March 2005; accepted 8 September 2005 Abstract This article demonstrates how economic and environmental improvement potentials of Danish pig farms can be estimated using Data envelopment analysis (DEA). To avoid some of the problems associated with the definition of undesirable outputs, environmental variables are included as nutrients applied with manure and nutrients removed with the crops. These environmental variables are combined with a series of economic variables, including the costs of fertilizer. Furthermore, subvector efficiency and a two-step analysis are used in order to consider economic and environmental (technical) efficiency separately. The empirical results show considerable improvement potentials, especially on the environmental variables. Sensitivity studies validate that these results are robust to possible uncertainties in the measurement of the environmental variables. JEL classification: C14, C61, D24, Q12, Q53 Keywords: Data envelopment analysis; Subvector efficiency; Pig production; Undesirable outputs; Economic and environmental improvement potential 1. Introduction Environmental regulation of agricultural production usually implies a trade-off between economic performance and the ex- tent of environmental regulation. This trade-off is crucial for the Danish pig sector, which is highly export-oriented (over 80% of physical production) and, at the same time, has a tendency toward concentration of production, resulting in potential envi- ronmental problems in the form of nutrients surpluses (Petersen, 1996). Environmental improvements, such as decreased nitrate leaching, that do not affect the cost of pig production, would therefore be ideal. In this article we analyze a data set of 290 Danish pig produc- ers, based on their annual production accounts. Data envelop- ment analysis (DEA) is used to estimate their economic as well as environmental improvement potentials. See, e.g., Charnes et al. (1978) and Charnes et al. (1994) for general DEA refer- ences and, e.g., Piot-Lepetit and Rainelli (1996) for an applica- tion of DEA to pig production. The main environmental problem associated with Danish pig production is nutrient surpluses. A straightforward way to deter- mine the nutrient surpluses on fields is to estimate the nutrients added to the fields through manure and fertilizer and subtract ∗ Corresponding author. Tel.: +44 115 951 5484; fax: +44 115 846 6667. E-mail address: mette.asmild@nottingham.ac.uk (M. Asmild). the amounts removed with the harvested crops. Unfortunately, data availability prevents direct estimation of the surpluses, as the production accounts used only includes the cost of fertilizer but not its nutrient content. DEA is well suited to deal with this problem, as it is capable of handling multiple inputs and multiple outputs, measured in different units. In the literature, environmental factors are usually included in DEA as undesirable outputs (here nutrient surpluses). However, the way in which undesirable outputs are modeled influences the results of the analysis (see, e.g., Scheel, 2001). In this article, we suggest a model formulation in which estimated nutrient contents in the manure and in the harvested crops enter as inputs and outputs, respectively, together with a number of economic variables, including the cost of fertilizer. This particular model formulation overcomes data limitations and furthermore avoids the problems associated with modeling undesirable outputs. We also apply a two-step subvector DEA approach, which can, among other things, model the behavioral assumption that the farmer has sequential preferences in the sense that improve- ments in technical efficiency of economic variables is the pri- mary goal. Only when this improvement potential has been achieved are any additional improvements in technical effi- ciency of the environmental variables become relevant. In other words, we determine how much each pig producer could in- crease his nutrients removal without influencing either direct c  2006 International Association of Agricultural Economists