Forecasting Chinese CO2 emission using a non-linear multi-agent intertemporal optimization model and scenario analysis

Haitao Xu,Xiongfeng Pan,Shucen Guo,Yuduo Lu
DOI: https://doi.org/10.1016/j.energy.2021.120514
IF: 9
2021-01-01
Energy
Abstract:The whole community have paid a lot of attention to whether China can achieve its emission target. The paper adds to the existing literature on emission forecast by considering consumption preference, knowledge capital and technological innovation mechanism with in a non-linear multi-agent inter temporal optimization model (NL-MIOM) which can further improve the accuracy of CO2 emission prediction. The historical fitting test shows that the MAPE value of NL-MIOM model is 1.81%, which is lower than the GM, NGM, ARIMA, OGM, SVR and BR-AGM models. By using this model, we forecast the CO2 emissions and energy consumption structure in China under different scenarios from 2018 to 2035. We find that China's CO2 emissions will peak around 2032, 2029 or 2027 with 12.34, 11.59 or 11.17 billion tons CO2 emissions under the benchmark scenario, Preference A (American consumption preference) scenario and Preference B (Japanese consumption preference) scenario. Based on the methodology of LMDI decomposition, we identify the main factors affecting China's CO2 emissions. The results show that the technical progress is the main reason for the reduction of CO2 emissions in the historical stage, pre peak stage and post-peak stage. Moreover, we also forecast the energy use of 14 different industries in China under different scenarios. (c) 2021 Elsevier Ltd. All rights reserved.
What problem does this paper attempt to address?