Benchmarking With Data Envelopment Analysis

Global surveys of the most popular management instruments reported by Bain & Company Inc has consistently listed benchmarking among the most important tools.Danish Professor Peter Bogetoft,an outstanding scholar in benchmarking and professor in the department of economics of Copenhagen Business School,defines benchmarking as a managerial tool that improves performance by identifying and applying best documented practices in the business world.In production economics,it is necessary to evaluate the performance of firms’ or organizations’ subunits before identifying and applying best documented practices,discriminate efficient and inefficient units,thereby determining how to improve performance.Data Envelopment Analysis(DEA)is a nonparametric methodology to measure performance.It can formulate various Pareto production frontiers by imposing different axioms,providing a powerful theoretic instrument for benchmarking.From the viewpoint of the definition of benchmarking,this dissertation attempts to further explore benchmarking using DEA.Previous DEA benchmarking studies mainly focus on how to identify best documented practice,in other words,benchmark selection.Few of them have considered resource restrictions,performance improvement in weighted space with undesirable outputs and the Pareto optimality of benchmark selection in Direction Distance Function(DDF)framework to date.In addition,how to apply the best documented practice(i.e.,benchmark realization)has not been investigated fully in literature.Therefore,this dissertation explores benchmarking with DEA,combining context-dependent DEA,Distance Friction Minimization(DFM)model,DDF and principal-agent theory to fill research gaps mentioned above as follows:(1)DEA benchmark selection from a resource restriction perspective.This chapter divides all units into different nested efficient frontiers using context-dependent DEA,and describes the resource restriction constraint using the slacks of input reductions and output expansions in each performance improvement process.The DEA closest target model is used to select benchmark in each performance improvement process.An Algorithm is established to determine the complete benchmarking path.Finally,the proposed approach is applied to the benchmark selection of China’s transport sector empirically.(2)DEA benchmark selection with a modified DFM model.This chapter points out that the objective function of DFM model is inconsistent with benchmark selection.In particular,this chapter extends DFM model into environmental benchmark selection and environmental efficiency evaluation by incorporating undesirable outputs.Lastly,the new DFM model is employed to select environmental benchmarks and measure environmental efficiencies of China’s regional industry.(3)DEA benchmark selection with endogenous direction in DDF.This chapter determines endogenous directions that guarantee the selected benchmarks along this direction locate on the Pareto production frontier.A benchmark index is defined to describe the effort to realize the selected benchmark.In particular,the benchmark index is positive,weak monotonic,translation invariant,and reference-set dependent.Finally,an empirical illustration to China’s transport sectors is conducted.(4)Benchmark realization from a principal-agent perspective.This chapter regards managers and subunits as players in an incentive game,where the efforts of subunits and the disutility of efforts are considered.An optimal reimbursement scheme is proposed to motivate subunits to realize best documented practice.The contributions of this dissertation are as follows:(1)the effects of technology level,environmental influence,and governmental policy on resource restriction are identified,which is employed to benchmark selection practice;(2)the DFM model is modified and extended to environmental benchmark selection and environmental efficiency evaluation,which perfects benchmark selection studies using DFM models;(3)An endogenous direction is proposed to ensure Pareto optimality of the selected benchmarks using DDF and DEA,which proves the practicality of DDF in benchmark selection theoretically;(4)Principal-agent theory is first employed to DEA benchmarking,suggesting a novel perspective to benchmarking in the future.