The Calculation and Prediction of China's Energy Rebound Effect Based on the LMDI and SVAR Models

The energy rebound effect refers to the lower-than-expected energy savings caused by energy efficiency improvement. We used the LMDI to measure China's rebound effect from 2004 to 2018 and the SVAR model to predict it based on energy consumption data from 2010 to 2019. Four ICA methods accomplish the parameter recognition of the model. The conclusions were: (1) The energy rebound effect was characterized by "U"-type, with 86% left vertex in 2005 and 82% right vertex in 2018. The energy-saving effect slowly rose and then sharply decreased. (2) With the benchmark SVAR model using dcov, ngml, FastICA, and LiNGAM methods, the mean of rebound effect prediction results are 103.33%, 104.67%, 105.33%, and 105.67% in 1, 2, 4, and 6 years, respectively. However, it indicates that energy efficiency improvement results in energy consumption which is inconsistent with China's facts. (3) The expanded SVAR model, which adds the variables of economic structure and energy quality, still has an upward trend of 1, 2, 4, and 6 years, with an average of 79%, 85.33%, 87%, and 87.67% for the four methods which is more reasonable. However, the energy-saving effect still tends to decrease with improved energy efficiency, which needs further attention.