A Comparative Study of Energy Consumption Decomposition Methods
能源消费影响因素分解方法的比较研究

Growth in energy consumption is a concern shared by many countries. Increases and decreases in output, energy efficiency and industrial structural adjustment cause changes in energy consumption. Here, we described various decomposition algorithms from the view of energy consumption and energy intensity change and compare results for decomposition across China's manufacturing industry. The factor decomposition method is a way to analyze reasons for changes that affect energy consumption. The energy decomposition method can be divided into the amount of energy consumption, and energy intensity. The Shapley algorithm, M-E algorithm, Se-Hark algorithm and AWT-PDM algorithm are common decomposition methods for energy consumption. The results of the two methods are the same, but the Shapley algorithm is more suitable for multi-factor decomposition. The Se-Hark Park algorithm and AWT-PDM algorithm better reflect the impact of the economic structure of energy consumption. The defect in the AWT-PDM algorithm is that the method cannot decompose the amount of change in energy consumption completely. The method of energy intensity decomposition is divided into multiplication decomposition, and sum decomposition. Multiplication decomposition reflects change in the rate of energy intensity, the sum decomposition reflects the amount of energy intensity change. The Laspeyres algorithm, Fisher Algorithm, AMDI algorithm and LMDI algorithm are common methods of energy intensity decomposition. Among these algorithms, the Fisher algorithm and LMDI algorithm can decompose the change in energy intensity completely, but the result of the Laspeyres algorithm and AMDI of algorithm results in residual items. Fisher and the refined Laspeyres algorithm makes up for the shortcomings of the Laspeyres algorithm. The results of the refined Laspeyres algorithm are equal with the Shapley algorithm. The decomposition of the AMDI algorithm results in small residual items and the LMDI algorithm which is a further development of the AMDI algorithm eliminates residual items.