Regulatory impact of environmental standards on the eco-efficiency of firms

Abstract

In this paper, we propose an approach to implement environmental standards into Data Envelopment Analysis (DEA) and in this way to measure their regulatory impact on eco-efficiency of firms. One standard feature of basic DEA models lies in the exogeneity of inputs, desirable and undesirable outputs. Taking into account the environmental constraints, we therefore apply the bounded variable DEA model. The regulatory impact is assessed as difference in eco-efficiency scores before and after fictive introduction of an environmental standard. Furthermore, we distinguish between weak and strong disposability of undesirable outputs and develop corresponding models. Assessing the regulatory impact of environmental standards in advance provides support for environmental policymakers in choosing appropriate instruments and in adjusting the intensity of regulation.

Keywords:

• data envelopment analysis
• environmental studies
• energy
• regulation

Acknowledgements

The authors are thankful for helpful comments of the participants of the EWEPA 2011 Conference in Verona and the International Conference on Operations Research 2011 in Zurich.

Notes

1 Korhonen and Luptacik (2004) define eco-efficiency according to Heinz Felsner: ‘We are looking for eco-efficient solutions such that the goods and services can be produced with less energy and resources and with less waste and emission’.

2 For a survey of DEA studies in the area of energy and environment, see Zhou et al (2008).

3 The same procedure can also be applied to derive the regulatory impact on allocative and overall efficiency, respectively.

4 According to Shannon (1997), ‘there are several ways to describe the production possibility set of a firm, one refers to a transformation function f’.

5 Limitation on absolute amount of undesirable outputs.

6 Note that the regulatory impact can be computed for single firms (difference in single eco-efficiency scores) and the whole industry (difference in average eco-efficiency scores), given the sample remains unchanged.

7 For a more detailed description of the analytical model framework the interested reader is referred to Korhonen and Luptacik (2004).

8 The Undesirable Output Model from Cooper et al (2007) is provided in Appendix A.

9 This condition was not specified by Shephard (1970), Kuosmanen (2005) and Podinovski and Kuosmanen (2011). We add it to describe the technology under weak disposability via inputs. Starting from a given technology (y^g, y^b)∈P(x), bad outputs need to be decreased to (θy^b). Consequently, inputs need to be increased to ((1)/(θ)x).

10 The advantage of a fictive sample is that the data can be used for all models presented in the paper and thus achieves comparability.

11 For a graphical illustration, see Figure 2.

12 DMU D is again projected on the intersection between efficiency frontier and environmental standard (). For a graphical illustration, see Figure 1.

13 Recall that the environmental standard was set to (Emission/Output)⩽3.5.

14 Moreover, it is interesting to compare the results of SBM Model B in Table 2 and SBM Model B weak in Table 4. Despite referring to the same firms and efficiency frontier, the average efficiency score decreased from 0.80 to 0.76 when taking into account weak disposability of undesirable outputs.