能源消费影响因素分解方法的比较研究

A Comparative Study of Energy Consumption Decomposition Methods

在能源消费研究中经常要将其变化分解为各种因素,通过各种因素的影响程度来分析影响能源消费的变化原因。自20世纪70年代以来,已经发展了多种分解方法,这些方法的假设前提与算法都有较大差异。本文从能源消费变化量与能源强度两个角度介绍了各种分解算法,对其进行了对比,并运用中国制造业的相关数据比较了分解的结果。能源消费变化量的分解方法中,Shapley算法与M-E算法分解结果相同,但是Shapley算法按照各因素的贡献加权来计算各因素对能源消费量的影响,更适合多因素分解;Se-HarkPark算法与AWT-PDM算法能较好地体现经济结构对能源消费量的影响,但AWT-PDM算法由于权数的确定问题,使得该算法存在不能分解的剩余项。能源强度变化的分解方法分为乘法分解和加法分解两大类,乘法分解反映的是能源强度变化率,加法分解反映的是能源强度变化量。Fisher算法与LMDI算法均是对能源强度的完全分解,而Laspeyres算法与AMDI算法均存在剩余项;方法是对AMDI算法的改进。研究者在研究能源问题或环境问题时,应根据研究要求及所掌握的数据选择恰当的分解方法。

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.