大连海事大学硕士学位论文
摘要
由 CO2过多排放引起的环境问题已经成为世界上最严重的环境问题之一，其产生的
后续效应给人类的日常生活带来了恶劣的影响，因此如何实现碳减排成为当今世界的一
个重要难题，人类前进的一个重大挑战就是气候变化所带来的负面影响。实现碳减排要
着眼于各个领域，而航运业在国际贸易中的地位举足轻重，有超过 80%的货物是通过航
运完成的。港口作为重点耗能单位，已成为实现碳减排的重要突破口。每年的航运消耗
的燃油大约有 3亿吨，排放的污染物严重影响了生态环境，尤其是近海域人民的生活环
境。因此科学的测度中国上市沿海港口公司的碳排放绩效、分析其时空演变特征以及预
测其未来的发展趋势十分必要，有利于推动中国早日实现碳中和，根据碳排放绩效可以
为中国低碳港口的建设提供参考依据。
由于港口公司数据查找的困难性和特殊性，本文基于 2005-2020年数据，选择中国
沿海 13个上市港口公司作为研究对象，采用考虑非期望产出的超效率 SBM-DEA模型
对港口公司的投入产出进行处理，得到各上市港口公司的碳排放效率值，并结合各年份
的效率值从年份、省域以及区域分别进行静态分析，以描述上市港口公司的效率值在不
同层面的时间和空间上的变化；然后采用 Global-Malmquist-Luenberger（GML）指数对
投入产出指标进行处理得到 GML指数，就全要素生产率及技术进步和技术效率对碳排
放效率在不同层面上进行动态的时空演变分析；采用灰色 GM（1,1）模型对 2005-2020
年效率均值进行预测，其预测值通过了后验差检验，因此用同样的方法预测 2021-2026
年的碳排放效率值，通过 2005-2020年的预测值计算出与各年实际值的误差和相对误差，
将误差划分状态空间，利用马尔可夫链建立一步概率转移矩阵，再利用灰色马尔可夫模
型确定未来碳排放效率预测值的变动区间，对其预测值进行修正，最终得到修正的预测
值。将前后得到的数据进行对比，经过灰色马尔可夫模型修正后的预测值更加符合实际，
因此利用灰色马尔可夫模型对 2021-2026年的预测值进行残差修正得到更加准确的预测
值。结果显示：(1)2005-2020年中国沿海上市港口公司的碳排放效率值呈现稳定中小幅
波动变化，但整体水平不高，因此上市港口公司的碳排放绩效在未来的发展中仍然需要
较大的提升，且近几年的趋势在缓慢下降，并无回升趋势。(2)从空间上来看，对于不同
区域和省域的上市港口公司的碳排放效率值存在较大的差异，东南沿海港口企业效率值
最高，珠三角和长三角的效率值次之，最低的是西南沿海上市港口公司的效率值。空间
格局呈现出“南高北低”“东高西低”的特征。(3)根据灰色马尔可夫的预测结果，中国沿海
上市港口公司的碳排放效率值变化幅度较小，比较稳定，但是未来 6年的预测值稳定在
0.68左右，相较于前几年有所下降，并无回升的趋势，因此未来港口的碳排放绩效的发
展并不乐观。
-
I
中国沿海上市港口公司碳排放绩效时空演变及预测
关键词：碳排放绩效；超效率 SBM-GML模型；灰色-马尔可夫模型；时空演变；
预测趋势
II
大连海事大学硕士学位论文
Spatial and temporal evolution and prediction of carbon emission
performance of listed coastal port companies in China
Abstract
The environmental problems caused by excessive CO2 emissions have become one of the
most serious environmental problems in the world, and its follow-up effects have had a bad
impact on human daily life. Therefore, how to achieve carbon emission reduction has become
an important problem in today's world. A major challenge for human progress is the negative
impact of climate change. The realization of carbon emission reduction should focus on
various fields, and the shipping industry plays an important role in international trade, with
more than 80% of goods completed by shipping. As a key energy consuming unit, port has
become an important breakthrough to achieve carbon emission reduction. The annual fuel
consumption of shipping is about 300 million tons, and the pollutants discharged have
seriously affected the ecological environment, especially the living environment of people in
the near sea area. Therefore, it is very necessary to scientifically measure the carbon emission
performance of China's coastal port enterprises, analyze their temporal and spatial evolution
characteristics and predict their future development trend, which is of great significance to
China's realization of carbon neutralization and provide a reference basis for China's
realization of green port construction.
Due to the difficulty and particularity of data search of port enterprises, based on the data
from 2005 to 2020, this thesis selects 13 listed port enterprises along the coast of China as the
research object, processes the input and output of port enterprises by using the super
efficiency SBM-DEA model considering unexpected output, obtains the carbon emission
efficiency value of each listed port enterprise, and makes a static analysis from year, province
and region combined with the efficiency value of each year, to describe the time and space
changes of the efficiency value of listed port enterprises at different levels; Then, the
Global-Malmquist -Luenberger (GML) index is used to process the input-output indicators to
obtain the GML index, and the dynamic temporal and spatial evolution of total factor
productivity, technological progress and technological efficiency on carbon emission
efficiency is analyzed at different levels; The Grey GM (1,1) model is used to predict the
average efficiency from 2005 to 2020, and its predicted value passes the posterior error test.
Therefore, the carbon emission efficiency from 2021 to 2026 is predicted by the same method.
The error and relative error between the predicted value from 2005 to 2020 and the actual
value of each year are calculated, the error is divided into state space, and the one-step
probability transfer matrix is established by Markov chain, Then, the Grey-Markov model is
used to determine the variation range of the predicted value of future carbon emission
efficiency, and the predicted value is modified to finally obtain the modified predicted value.
III