Performance measurement in public and regulated organisations: Evaluation and improvement of productive efficiency using data envelopment analysis.

Data envelopment analysis (DEA) represents a new and innovative way of measuring performance in the public sector. First suggested by Farrel (1957), linear programming techniques have made it possible to measure technical efficiency through the estimation of non-parametric production frontiers. DEA measures technical efficiency, it does not need information on prices or costs and produces a single efficiency criterion using data purely on measured volumes of inputs and outputs, including qualitative ones. The type of DEA programme used in this thesis is one suggested by Banker (1984) allowing for variable returns to scale in the construction of the production frontier. It is argued that DEA is superior to other attempts to measure the performance of public sector organisations. It has been traditional to use partial productivity ratios to measure efficiency and these have recently become popular in the public sector as 'performance indicators'. Other techniques used are regression analysis and the estimation of econometric production frontiers. It is demonstrated that DEA has a number of advantages over these methods. This is not to say that DEA does not have its drawbacks or that there are not practical problems with its use. These have been explored through the conduct of two case studies, one of Post Office Counters and the other of the Area Electricity Boards. Themes explored during the course of the case studies are as follows. How useful is DEA as a performance measurement tool in the context of Post Office Counters. Can more information on efficiency be yielded by clustering the datasets into smaller groups which have a. common factor and re-estimating the frontier. How robust is the DEA isoquant given that it is a frontier constituted on the basis of maximum observations and is susceptible to outliers, and can this be overcome. Also undertaken in this thesis is one of the first British studies using DEA in a dynamic context. The case study of Counters was conducted using data that was derived from 1281 Crown Post Offices for a 13 week period from September-November 1989. The input used was labour (in hours) and the outputs were a quality variable (average waiting time) and the outputs were ten different categories of transactions. The results were examined for the whole of Counters and on a regional basis. In the study of the Area Electricity Boards a time-series approach was taken on a pooled data set for 12 AEBs from 1969-88. Three inputs and four outputs were used. Technical efficiency between AEBs is compared, the change in efficiency over time is examined and conclusions reached about the implications for regulation now they have been privatised