I 摘要 随着国内外空间计量经济学应用领域的不断扩展，空间权重矩阵的构建方法 日益受到重视。作为直接反映空间单元间空间溢出效应的核心部分，其构建方法 研究对于空间计量经济学的理论发展与应用拓展具有重要的价值。本文根据中国 500kV及以上电力线路的投运时序，提出了基于电网拓扑结构等基础设施的空间 权重矩阵构建方法。通过在能源经济领域的各类空间计量模型中的应用，探讨其 引致的空间溢出效应的变化机理，并结合具体研究问题的经济含义提出具有操作 性的政策建议。论文的主要特点和结论如下： ①以电网拓扑结构变化对区域能源强度的影响为例，揭示区域基础设施连接 紧密程度对空间相关性变化影响的内在机理。根据1998-2014年中国30个省份高 压电网建设的时序顺序与拓扑结构构建空间权重矩阵，实现了基础设施动态建设 与空间权重矩阵常数元素赋值的对应。使用SDM面板模型，并采用内生变量的滞 后一期值代替原变量克服内生性偏误，得到了更为稳健的结果。研究结果表明： 考虑电网拓扑结构的空间溢出效应后，电力占比与要素结构提升对能源强度改善 的影响得以体现，对外贸易的“污染天堂”效应被发现；《大气污染防治计划》配 套的12条高压电力线路的投运会改变中国各省份能源强度的空间同质性，使得电 力占比、产业结构、R&D研发水平对能源强度的影响变大，人均FDI存量、产权 结构、要素结构、城市化率对能源强度的影响变小；基于电网拓扑结构的空间权 重矩阵能反映区域能源强度地理空间邻接关系的主要信息，通过强化和调整各省 间的电网拓扑结构能够改变其空间溢出关系，促进政策目标的实现。 ②以碳排放指标的省际空间相关性为基础，揭示省间高压电网拓扑结构对碳 排放指标空间相关性的影响机理。依据1998-2014年中国30个省份高压电网省间 联络线的投运时序构建空间权重矩阵，采用动态空间杜宾面板模型克服静态空间 面板模型的忽略变量偏误。研究发现：省间高压电网互联使得各连接省份间的碳 排放指标呈现出空间同质性的特点；计及高压电网的动态空间溢出效应后，能源 强度降低对减少碳排放指标起着关键作用；提高电力占比与改善产业结构对减少 碳排放指标起着重要作用。加强省间高压电网互联有助于强化省间碳排放指标的 空间相关性。 ③提出能源流出需求指数反映中国能源生产/消费的区域不平衡特征，构建基 于电网拓扑结构的空间权重矩阵因子引导空间溢出路径，以二者之积构建嵌套空 间权重矩阵，并采用动态空间Dubin模型，综合反映能源回弹效应的空间溢出效 应。发现：能源强度每下降1吨标准煤/万元，长短期能源回弹效应分别降低18.794 重庆大学博士学位论文 II 与17.867个百分点；电力占比每上升1个百分点，长短期能源回弹效应分别降低 9.360与8.898个百分点。人均能耗每增加1吨标准煤/人，长短期能源回弹效应分 别提升10.343与9.833个百分点；城市化水平每上升1个百分点，长短期能源回 弹效应分别提升2.928与2.784个百分点。电网拓扑结构决定了能源回弹效应的空 间溢出路径，能源流出需求指数影响了各影响因素与能源回弹效应变动的量化关 系。本文构建的嵌套空间权重矩阵，揭示了能源生产/消费的区域不平衡与能源基 础设施拓扑结构对空间溢出路径的共同影响。发现提升能源利用效率与电力占比 的政策措施对控制能源消费总量至关重要，加强能源生产大省与能源消费大省的 能源基础设施互联，对平衡全国的能源回弹效应有重要的价值。 本文提出的空间权重矩阵构建方法不仅实现了能源经济指标空间相关性机理 分析的方法创新，还为政策制定者量化分析省级能源基础设施互联对节能减排政 策的影响提供了测量方法。 关键词：电网拓扑结构，空间权重矩阵，能源强度，碳排放指标，能源回弹效应 英文摘要 III ABSTRACT The specification of spatial weight matrix draws more and more attentions with the extension of spatial econometricat home and abroad. Spatial weight matrixreflect directly the spatial spillover effect of space units, the study of itsspecification method isof great value to the development and extension of spatial econometric. The paper presented a new specification method of spatial weight matrix based on the operating sequence of Chinese power lines above500kV that form the topological structure of power grid as a representative of infrastructure. Through the application of the specified spatial weight matrixin different spatial econometric models in different areas of energy economics, the changing mechanism of spatial spillover effect among spatial units is discovered and the operational policy recommendations are proposed according to the economic implications of the specific questions. The main features and conclusions are as follows: ①The effectto the regional energy intensity caused bythe topological structure of Chinese power grid is taken as an example to reveal the changing mechanism of the spatial correlations induced by the topological structure of infrastructure. The spatial weight matrix is specified based on the topological structure consisting ofinter-provincial high voltage power linesthat operated at different time during the period 1998-2014 in China, the specification transformeveryentry of spatial weight matrix to a constant elementaccording to the dynamic topological structure of infrastructure.A more robust result is obtained using the SDM panel model through replacing the original endogenous explanatory variable with lag one.The finding is that: increasing the proportion of power consumption and factor structure can reduce regional energy intensity when spatial correlationscaused bythe topological structure of power grid were considered, “pollution heaven effect” from foreign trade was found; the operating of 12 high voltage power lines will change the spatial homogeneity of inter-provincial regional energy intensity, the effects of the proportion of power consumption, industrial structure and R&D level will be strengthened, the effects of per capita FDI stock, ownership structure, factor structure and urbanization rate will be weakened; the regression results using the specified spatial weight matrix based on the topological structure of power grid can reflect the major information that arespatial correlations of regional energy intensity caused by the spatial adjacency relationship. 重庆大学博士学位论文 IV The spatial spillover effectof regional energy intensity could be changed by adjusting the topological structure of inter-provincial power grid, and promote the policy objective to realize. ②The changing mechanism of spatial correlations caused by inter-provincial high voltage power lines is illustrated with spatial correlations of carbon dioxide emission index among provinces. The spatial weight matrix is specified according to the operating sequence of inter-provincial high voltage power lines in China during the period 1998-2014, and the dynamics spatial Dubin panel model is adopted to overcome the ignored bias error induced by the ignored explanatory variable in the static one. The finding is that: the interconnection of inter-provincial high voltage power grid makes carbon dioxide emission index to show spatial homogeneity; the action that reduce provincial energy intensity is the key influence to reduce carbon dioxide emission indices, resulted from the spatial spillover effects caused by inter-provincial topological structure of power grid and the actions that increase the proportion of power consumption and adjust industrial structure are also important measure. Strengthening the interconnection of inter-provincial high voltage power grid canpromote spatial correlations among provincial carbon emissions indices. ③The paper reflects the inter-provincial imbalance of energy product/consumption by constructing energy outflow demand index; the spatial weight matrix elements specified by the topological structure of power grid configure the spatial spillover path. Dynamic spatial Dubin model with the nested spatial weight matrix by the products of the two is used to analyze the spatial spillover effect. The findings are: Energy rebound effects of long & short-term will reduce by 18.794%, 17.867% respectively for the reduction of energy intensity by 1 ton of standard coal equivalent per 10,000 Yuan; will reduce by 9.360%, 8.898% respectively for the increasing of electricity proportion by 1%. The two effects will increase by 10.343%, 9.833% respectively for increasing of energy consumption by 1 ton of standard coal equivalent per capita, and will increase by 2.928%, 2.784% respectively for increasing of urbanization rate by 1%. The topological structure of power grid configure the spatial spillover paths, and energy outflow demand index affects the quantitative relations betw