The allocation mechanism for carbon emissions permit(CEP)is an institutional guarantee for advancing the development of China’s unified carbon trading market.The initial allocation of carbon quotas fails to solve new...The allocation mechanism for carbon emissions permit(CEP)is an institutional guarantee for advancing the development of China’s unified carbon trading market.The initial allocation of carbon quotas fails to solve new inequalities stemming from subsidizing cleaner production.This paper constructs a theoretical framework that describes China’s progressive decline in carbon intensity,calculates the equilibrium solution on the neoclassical saddle point path using the shooting method,and studies the income distribution imbalance caused by cleaner production subsidies and the reallocation mechanism of carbon emissions permit The main conclusion is that the incremental cleaner production subsidy policy meets the goal of maximizing welfare on the saddle point path,but it may lead to over-investment in the clean sector,thus causing the income distribution imbalance among entities.Further research suggests that the amount of carbon emissions permit acquired by the clean sector should be higher than the actual emissions in the trading market and that,as the cleaner support increases,the share of carbon emissions permit acquired by the sector should be constantly increased through reallocation mechanism.This helps achieve the Pareto improvement in all parties’economic benefi ts.展开更多
针对现有的研究大多将短序列时间序列预测和长序列时间序列预测分开研究而导致模型在较短的长序列时序预测时精度较低的问题,提出一种较短的长序列时间序列预测模型(SLTSFM)。首先,利用卷积神经网络(CNN)和PBUSM(Probsparse Based on Un...针对现有的研究大多将短序列时间序列预测和长序列时间序列预测分开研究而导致模型在较短的长序列时序预测时精度较低的问题,提出一种较短的长序列时间序列预测模型(SLTSFM)。首先,利用卷积神经网络(CNN)和PBUSM(Probsparse Based on Uniform Selection Mechanism)自注意力机制搭建一个序列到序列(Seq2Seq)结构,用于提取长序列输入的特征;其次,设计“远轻近重”策略将多个短序列输入特征提取能力较强的长短时记忆(LSTM)模块提取的各时段数据特征进行重分配;最后,用重分配的特征增强提取的长序列输入特征,提高预测精度并实现时序预测。利用4个公开的时间序列数据集验证模型的有效性。实验结果表明,与综合表现次优的对比模型循环门单元(GRU)相比,SLTSFM的平均绝对误差(MAE)指标在4个数据集上的单变量时序预测分别减小了61.54%、13.48%、0.92%和19.58%,多变量时序预测分别减小了17.01%、18.13%、3.24%和6.73%。由此可见SLTSFM在提升较短的长序列时序预测精度方面的有效性。展开更多
基金The authors express their gratitude to the Youth Program of National Natural Science Foundation of China(71904131)the 2019 Youth Talent Program for Publicity,Thought and Culture by the Publicity Department of the CPC Central Committee,and the basic research expenses of Beijing municipal universities for the Capital University of Economics and Business for their funds。
文摘The allocation mechanism for carbon emissions permit(CEP)is an institutional guarantee for advancing the development of China’s unified carbon trading market.The initial allocation of carbon quotas fails to solve new inequalities stemming from subsidizing cleaner production.This paper constructs a theoretical framework that describes China’s progressive decline in carbon intensity,calculates the equilibrium solution on the neoclassical saddle point path using the shooting method,and studies the income distribution imbalance caused by cleaner production subsidies and the reallocation mechanism of carbon emissions permit The main conclusion is that the incremental cleaner production subsidy policy meets the goal of maximizing welfare on the saddle point path,but it may lead to over-investment in the clean sector,thus causing the income distribution imbalance among entities.Further research suggests that the amount of carbon emissions permit acquired by the clean sector should be higher than the actual emissions in the trading market and that,as the cleaner support increases,the share of carbon emissions permit acquired by the sector should be constantly increased through reallocation mechanism.This helps achieve the Pareto improvement in all parties’economic benefi ts.
文摘针对现有的研究大多将短序列时间序列预测和长序列时间序列预测分开研究而导致模型在较短的长序列时序预测时精度较低的问题,提出一种较短的长序列时间序列预测模型(SLTSFM)。首先,利用卷积神经网络(CNN)和PBUSM(Probsparse Based on Uniform Selection Mechanism)自注意力机制搭建一个序列到序列(Seq2Seq)结构,用于提取长序列输入的特征;其次,设计“远轻近重”策略将多个短序列输入特征提取能力较强的长短时记忆(LSTM)模块提取的各时段数据特征进行重分配;最后,用重分配的特征增强提取的长序列输入特征,提高预测精度并实现时序预测。利用4个公开的时间序列数据集验证模型的有效性。实验结果表明,与综合表现次优的对比模型循环门单元(GRU)相比,SLTSFM的平均绝对误差(MAE)指标在4个数据集上的单变量时序预测分别减小了61.54%、13.48%、0.92%和19.58%,多变量时序预测分别减小了17.01%、18.13%、3.24%和6.73%。由此可见SLTSFM在提升较短的长序列时序预测精度方面的有效性。
