Research On Theory And Application Of Data Envelopment Analysis In Energy Economy

With the nation’s attention gradually focused on the relationship between economic development and natural environment, environment economics and energy economics have recently developed tremendously. The energy efficiency issue is an important branch of the energy economy. In order to increase the value of energy return on energy invested and make use of energy more efficiently, it is important to choose an efficient way of the energy consumption. On the one hand, it is beneficial to the nation’s energy security. On the other hand, it meets the national call for energy saving and emission reduction, which plays an important role in the construction of harmonious social environment.Data envelopment analysis (DEA) has a wonderful prospect of application in energy economy.However,there exists a variety of problems in the application of traditional data envelopment analysis models. Traditional method of constructing production frontier in data envelopment analysis seems to be much more complex, in order to solve this problem, we proposed a new DEA production frontier algorithm, which is called the rotation algorithm. Further, we explained the meaning of rotation algorithm. Respectively, from the two-dimensional and high-dimensional perspective, we built the production frontier of traditional four DEA models, and proved the theoretical basis of the algorithm. Through practical examples, we proved that, compared to traditional methods, such as vertex and extreme direction method, Graham scanning method, rotation algorithm is much simpler and has a very wide practical value. Traditional DEA models cannot deal with the negative data problem. We pointed out that the negative data problem can be addressed by the addtive-based DEA model. Furthermore, we compared the properties of the traditional additive-based data envelopment analysis models and proposed two generalized DEA models, i.e., the generalized additive-based DEA model and the generalized additive-based super-efficiency DEA model. The virtues of the new models are two-fold:one is that they inherited the properties of the traditional additive-based DEA models. The other is that many new additive-based DEA forms can be derived from the generalized additive-based model. Finally, we gave two numerical examples to show the application of the generalized additive-based model. Traditional DEA models cannot deal with the uncertain data problem. The chance-constrained DEA model is used to address uncertain data. We transformed the chance-constrained DEA model into a robust DEA model. The main attributes of the new robust DEA model can be elaborated as two folds. One is that it is built based on the ellipsoidal uncertainty, so that many virtues of the ellipsoid could be used. The other is that this model is computational tractable. Finally, we applied this new model to the vendor performance problem. Finally, we applied the dynamic slacks-based measure model (DNSBM) to evaluate the operational performance of 31 electric power-supply companies (EPCs) in China from 2010 to 2012. It is indicated that:first, the macro-economic environment influences the policy of separation of transmission division (T) and the distribution division (D). Second, regional economic development level has a significant impact on the performance of the EPCs in China. In the end, EPCs in China should focus on enhancing the operating efficiency of D when deciding to separate T and D in the future.Finally, we focused our attention on the integer-valued DEA problem (IDEA) which was on the center of debate recently. We first analyzed the IDEA axioms, discovered some limitations in the well-known IDEA model (KKM) and developed a rectified version (RKKM). The RKKM model is superior to the original, both in theory and in application. We theoretically compared RKKM properties with the LVM model. Results showed that the LVM model’s optimal solution is no better than that of the RKKM when outputs were assumed to be integer-valued. Additionally, the optimal solution of the two models is equivalent when outputs were assumed to be real-valued. We note that the RKKM model might lose some optimal points and overestimate the efficiency in certain cases. We proposed the RDI model to help to retrieve these missing points and promote proper efficiency. We also compared the properties of the RDI and RKKM models, finding that the optimal point obtained by the RDI model is no better than that of the RKKM. Based on the properties of the RDI and RKKM models, we proposed a three-step method for solving the IDEA problem that may be seen as an improvement over previous ones. The famous 42 universities’example was cited to illustrate the approach.