Index decomposition analysis with multidimensional and multilevel energy data

- We study IDA methods to dealing with multidimensional and multilevel energy data.
- We consider both the additive and multiplicative decomposition analyses.
- Two important properties of IDA methods are investigated.
- LMDI-I is the recommended method to deal with energy data with multiple attributes.

Index decomposition analysis (IDA) is a popular tool for analyzing changes in energy consumption over time. Traditionally, a typical IDA study uses a single dimensional energy dataset, such as industrial energy consumption by industrial sector or transportation energy consumption by transport mode. More recently, there have been a growing number of studies using more sophisticated datasets, e.g. energy consumption by geographical region and by economic sector in a single dataset. For IDA studies using energy data with multiple attributes, intermediate decomposition results can be generated using subsets of the entire dataset, and these results provide further insight into the energy system and problem studied. To ensure that these intermediate results are consistent and meaningful, the IDA method used should ideally satisfy two properties: perfect in decomposition at the subcategory level and consistency in aggregation. It is shown that the logarithmic mean Divisia index method I (LMDI-I) satisfies these two properties in both additive and multiplicative decomposition analysis. It is therefore the recommended IDA method when dealing with energy data with multiple attributes.