Eco-Efficiency Analysis of Swedish Regions: A second-stage DEA approach

Abstract: This paper uses a two-stage data envelopment analysis (DEA) approach to measure and evaluate the relative eco-efficiency and the influencing factors of 17 out of 21 regions in Sweden during the period of 2008 and 2015. In the first stage of the DEA, a standard CCR model is used to calculate the relative eco-efficiency of the regions given multiple inputs and outputs. The model contains the variables labour, capital and energy consumption as desirable inputs, regional GDP as a desirable output and Co2 emissions as an undesirable output. On average, the results show that four regions are eco-efficiency leaders, eight are eco-efficiency followers and the remaining five are eco-efficiency moderates. The main finding from the first stage of the DEA is that the regions in general should concentrate on energy consumption and capital stock reduction to improve eco-efficiency. However, the results differ slightly across the regions. The second stage of the DEA constitutes a fractional logit regression model, which is used to analyse the factors influencing differences in eco-efficiency across regions. The results indicate that enterprises’ production value of goods has a significant and negative effect on eco-efficiency, whereas environmental awareness and environmental sector have positive and significant effects on eco-efficiency. Furthermore, net income and population density are insignificant. Thus, there is potential to improve eco-efficiency in regions via a reduction in energy consumption and in capital stock while keeping the output constant. A concluding remark is that guidance of national legislation should be better implemented at the regional level. Alternatively, regional legislation based on the regions’ environmental prerequisites should be implemented to a greater extent.