A Study On DEA-based Performance Evaluation And Resource Allocation For Parallel Production System

Resource allocation and target setting is a classic problem in economics. Under the paradigm of economic researches, it highly depends on competitive market and existence of price information. Data envelopment analyisis (DEA) as a nonparametric mathematical programming tool gains advantage in handling problems without price information, and with multiple inputs and outputs. Therefore, it was originally applied to the evaluation of non-profit organization. Furthermore, in economic researches, we often assume that the decision makers always make the optimal decisions. Thus, inefficiency is not a useful concept sometimes. However, the performance variation and inefficiency indeed exists in the operation of a real organization. Therefore, DEA can be applied to improve the ability to the evaluation of the past operation, and also make decision for the future. The topic of this paper is a case in point.Based on the literature review, this paper systematically summarizes the characteristics of the current researches, and points out the insufficiencies within those methodologies. In cases where the inner operation information is not available, this paper proposes models to perform evaluation. In addition, when the black box of decision making unit is opened, this paper also studies various models to effect an efficient resource allocation and target setting scheme with equality consideration.Chapter 1 firstly summerizes the fundamental theory of DEA, which includes the main principles for efficiency evaluation, models, and theorecial applications. Secondly, based on the literature review, this chapter systematically summarizes the characteristics of extant researches in DEA-based resource allocation and target setting, and the main mothodologies applied in the literature. In the mean time, the insufficiencies are pointed out. In the end, the significance of this dissertation is demonstrated.Chapter 2 aims at the bottleneck for the application of network DEA, namely, lacking internal operational data. This paper proposes ?rstly to classify the production activities in light of value analysis to form core production unit and non-core production unit. Secondly, the input data of core production unit are estimated based on Pareto distribution, and then used by extended DEA model for evaluation purpose.Chapter 3 provides a radial measurement of efficiency for production process of parallel structure under variable returns to scale assumption, and proposes procedure without affected by multiple optima for estimating returns to scale. The equivalence between the traditional black box approach and the proposed parallel approach is established, which ensures the consistency in dealing with returns to scale. Finally, an empirical study of the pollutant treatment processes in China is presented.Chapter 4 proposes methodology for resource allocation and target setting based on DEA from the perspective of efficiency. It deals with organization that can be modeled as consisting of several production units, each of which has parallel production lines. The previous studies in the DEA literature only deal with reallocating/allocating organizational resources to production units and set targets for them. In their researches, the production unit is treated as a black box. In such circumstances, how to arrange the production at production unit level is not clear. This paper serves to generate resource allocation and target setting plan for each production unit by opening the black box.Chapter 5 studies models for resource allocation and target setting within DMUs. It is an extended research from the previous chapter. Since that the existing papers mainly focus on resource allocation at organizational level, it proposes bi-criteria (efficiency and equality) DEA models based on efficiency and equality analysis, while incorporating the characteristics of resource allocation and target setting within DMUs, that is, (1) fixed input and fixed output; (2) the relationship among Sub-DMUs in choosing weights when evaluating performance; (3) the relationship between the Sub-DMUs and DMU in choosing weights when evaluating performance.Chapter 6 summarizes all the work of this dissertation and presents some useful directions for future researches.