基于先验信息的社会碳成本贝叶斯不确定性分析
    点此下载全文
引用本文:曾惠芳1,熊培银2.基于先验信息的社会碳成本贝叶斯不确定性分析[J].经济数学,2020,(3):183-188
摘要点击次数: 155
全文下载次数: 0
作者单位
曾惠芳1,熊培银2 (1.湖南科技大学 商学院,湖南 湘潭 411201
2.湖南科技大学 信息与电气工程学院,湖南 湘潭 411201) 
中文摘要:针对气候变化及经济影响存在的巨大不确定性,研究了气候变化不确定性以及先验信息对社会碳成本的影响.在贝叶斯理论框架下,采用指数分布刻画气候变化的分布特征,假设尾部变化率是一个随机变量,给出其伽玛先验分布,推导了气候变化分布的贝叶斯先验预测分布.并分别基于指数分布以及帕累托先验预测分布计算了社会碳排放成本.模拟分析发现,在未融合先验信息的情况下,由于尾部概率很小,不管是否修正消费与气候变化之间的关系,截尾社会碳成本和未截尾社会碳成本几乎重合.然而,在利用贝叶斯方法融合先验信息的情况下,社会碳成本容易受到先验信息的影响.但是,通过修正消费与气候变化之间的关系后,发现社会碳成本受先验信息的影响比较少.
中文关键词:气候变化  社会碳成本  贝叶斯分析  不确定性
 
Bayesian-Uncertainty Analysis for the Social Cost of Carbon Based on Prior Information
Abstract:According to numerous uncertainties in climate change and its economic impacts, we study the effect of potentially severe climate change and prior knowledge on the social cost of carbon. Under Bayesian framework, it is supposed that the probability distribution for temperature change is the exponential distribution. Combining data with gamma prior information, we derive the prior predictive distribution for temperature change. We find that the prior predictive distribution is fat tailed because of prior information. It indicates that the prior information makes the possibility of rare event more large. Furthermore, we derive the Social Cost of Carbon (SCC) based on exponential distribution and the prior predictive distribution. The results show that, while no Bayesian learning, the curve of SCC under limit case is almost overlapping with the curve of SCC under un-limit case if the true climate sensitivity is distributed exponential, since observations near the mean provide evidence against fat tails. However, considering the Bayesian learning, the SCC under un-limit case changes more rapidly than the SCC under limit case if the true climate sensitivity is fat tailed because of the prior knowledge.
keywords:climate change  social cost of carbon (SCC)  Bayesian analysis  uncertainty
查看全文   查看/发表评论   下载pdf阅读器