DEA Methodology And Research Considering Undesirable Outputs

At the 75th session of the United Nations General Assembly,the Chinese government proposed that China will increase its national contribution and adopt stronger policies to achieve the ambitious goal of carbon neutrality by 2060.If China’s development is evaluated at this time,it is inevitable to pay attention to the improvement of environmental efficiency.However,reducing carbon emissions will inevitably result in economic losses,so balancing economic development and carbon reduction is a difficult issue.The treatment of environmental factors in the field of efficiency evaluation has been one of the classic problems that researchers have paid much attention to.The core of the problem is to deal with undesirable outputs in a more realistic way and construct a reasonable evaluation objective,so as to solve the problem that the efficiency score of the decision making unit is closer to the effective value(relative to the case when undesirable outputs are not considered)that occurs in the conventional data envelopment analysis(DEA)model from the efficiency perspective when undesirable outputs are introduced.Through the study of the existing literature,this paper systematically summarizes the main features as well as the shortcomings of the current relevant studies in the field of efficiency evaluation.On this basis,this paper introduces utility theory,and focuses on a comparative study of efficiency and utility perspectives,which not only provides a new perspective for dealing with undesirable outputs,but also has strong practical significance in balancing economic development and emission reduction.This paper is divided into four sections.The first part of this paper is the foundation of the study,which briefly introduces the research background,significance,content,and the methods used in this paper.Literature review covers both efficiency evaluation and utility theory.Firstly,a review and categorization of current research in the field of efficiency evaluation dealing with undesirable outputs and indicator selection is presented.Then the relevant concepts of utility theory and its research combined with DEA are introduced.The second part focuses on the theory and methodology of indicator selection and the treatment of undesirable outputs from an efficiency perspective,which introduces the challenges of indicator selection in the context of carbon neutrality from an efficiency perspective,and then proposes a DEA indicator selection and efficiency evaluation process based on machine learning methods and Monte Carlo experiments.Finally,based on the DEA indicator selection and efficiency evaluation process,a comparative study of the three types of DEA methods for dealing with undesirable outputs from the efficiency perspective is presented.The third part is devoted to the treatment of undesirable outputs from a utility perspective,examining the theory of DEA considering undesirable outputs and focusing on a comparative study of efficiency and utility perspectives.First,the conventional objective(efficiency)is replaced by utility.Based on the different preferences of decision makers for operational performance and environmental performance,the utility-based DEA model is constructed,and the corresponding strategies and the empirical results are given.After that,this paper focuses on a comparative study of the efficiency and utility perspectives,comparing and analyzing the advantages of the DEA model from the utility perspective over the efficiency perspective based on the theoretical analysis and empirical results.The fourth part is the summary and research outlook of the whole paper,which not only summarizes the innovations and shortcomings of the research content,but also clarifies the research directions.The main innovations of this paper include:(ⅰ)From the utility perspective,considering the fairness between operation and environment,bridging the fairness gap between environment and operation in the presence of undesirable outputs,thus better applying to efficiency evaluation in the context of carbon neutrality;(ⅱ)Based on machine learning theory,a DEA indicator selection and efficiency evaluation process is proposed to promote the intelligent development of DEA.