Abstract
With the growing interest in using the technique of index decomposition analysis (IDA) in energy and energy-related emission studies, such as to analyze the impacts of activity structure change or to track economy-wide energy efficiency trends, the conventional single-level IDA may not be able to meet certain needs in policy analysis. In this paper, some limitations of single-level IDA studies which can be addressed through applying multilevel decomposition analysis are discussed. We then introduce and compare two multilevel decomposition procedures, which are referred to as the multilevel-parallel (M-P) model and the multilevel-hierarchical (M-H) model. The former uses a similar decomposition procedure as in the single-level IDA, while the latter uses a stepwise decomposition procedure. Since the stepwise decomposition procedure is new in the IDA literature, the applicability of the popular IDA methods in the M-H model is discussed and cases where modifications are needed are explained. Numerical examples and application studies using the energy consumption data of the US and China are presented.
1. Introduction
Index decomposition analysis (IDA) has been widely used to study the driving forces behind aggregate energy consumption changes, sectoral and economy-wide energy efficiency trends, and differences between countries in energy consumption or carbon emissions. See, for example, Ang and Zhang (2000), EERE (2003), Bataille et al. (2007), IEA (2007), OEE (2009), and Ang et al. (2010). In economy-wide studies, total energy consumption is often given as a sum of consumption in various energy consuming sectors. Each sector in turns comprises subsectors and so on. This leads to a well-defined hierarchy of energy consumption data, where IDA studies can be conducted at different levels of sector disaggregation. Generally, changes in the aggregate energy consumption at a specific level are often decomposed to give the contributions of factors linked to overall activity change, activity structure shifts, and energy efficiency improvement. The choice of level may vary from one study to another due to differences in study scope and objective, or in data availability and quality.
1 8. Discussion and conclusions
We discuss the limitations of single-level decomposition in IDA applied to energy and study how multilevel analysis can help to address these limitations. We introduce two multilevel decomposition models, the M-P model and the M-H model, and study their features. The M-P model, the multilevel decomposition model often used by researchers in the IDA literature, adopts the same calculation procedure as conventional single-level model. In contrast, the M-H model adopts a stepwise decomposition procedure which is totally new in the IDA literature. To extend from single-level to multilevel analysis using the M-H model, modifications via transformations are needed to some popular IDA methods. We further discuss the practical significance of the M-H model and present real cases to illustrate.
