Project1:DEA models for Parallel System: Application to Manufacturing Industry of China from the Perspective of Auditing
Conventional data envelopment analysis (DEA) considers a system or decision-making unit (DMU) as a “black box” in calculating its efﬁciency and does not take the operation of individual components into account. However, in the real world, there are systems which are composed of independent component units. In recent studies on parallel DEA, inputs/outputs of the system are the sum of those of all its component units, which is not always true. This paper proposes a new parallel DEA model where each input/output of the system is not the sum of those of all its components under an assumption of variable returns to scale (VRS). The proposed approach is then applied to the manufacturing industry of China. Previous applications on manufacturing industry only discuss efficiency improvement for inefficient DMUs and barely analyze the efficient DMUs. However, from the perspective of auditing, the audit department should not only focus on inefficient DMUs or DMUs of the lower efficiency, but also pay attention to efficient DMUs or DMUs of the higher efficiency. Based upon this consideration, we propose a method to help the audit department select audit objects scientifically
Project2：Behind the Black-box DEA-a New Viewpoint of the Two-stage Structure
Before we know the internal structure and the internal data of an organization, the standard DEA model (“black-box” model) can only tell us to what extent the initial inputs (final outputs) of the organization should be reduced (increased). It cannot tell us, however, what adjustment on the intermediates should be made. In this project, we try to answer this question for serial two stage organizations, using their intermediates data. In other word, we try to provide the complete information about the efficiency improvement along the supply chain. In the popular “relative” two-stage DEA models, the authors put emphasis on the relationship between the two stages, by requiring that the weights of the intermediates are equal across different stages, but they generally lead to efficiency results less than those obtained by the black-box model. Knowing this inconsistency of efficiency scores between the two kinds of models, a researcher may be confronted with an “information dilemma”: he will inevitably deny his earlier efforts on efficiency evaluation after he gets the complete internal information of the organization.We try to derive the coordination mechanism within a serial two-stage organization based on available data, keeping the overall efficiency of our new model equal to that of the black-box one.
Project3：A Two-stage DEA Model with Partial Intermediates: Application to China Dairy Supply Chains
This project is aimed at evaluating efficiency of China dairy supply chains with a given scale of upstream. After examining the data of 2011’s “China’s dairy yearbook”, we construct a two stage non-cooperative model with some outputs in the first stage flowing out of the system, some additional inputs flowing into the second stage, and especially some of the intermediates only partly entering the second stage. By this model, we not only get the efficiency scores of each dairy chain, but also provide the optimal portion of the intermediates which partly go into the second stage and partly flow out of the given chain.