研究半参数空间自回归(SAR)面板模型的估计方法，对于丰富和发展非参数空间计量理论不仅有着重要的现实意义，而且具有重大的学术价值。目前关于这方面的研究主要集中于面板数据的纵剖面时间序列(即个体内)为相互独立随机过程的情形，与经济学、环境学等领域中的现实问题存在较大差距。本项目将从面板数据个体内存在相关性的角度下来研究半参数SAR面板模型的估计问题，具体研究内容包括：为个体内存在相关性的部分线性单指标SAR面板模型、个体内存在相关性的部分线性变系数SAR面板模型、个体内存在相关性的部分线性可加SAR面板模型分别建立有效估计方法，并在一些正则条件下拟推导这些估计方法的大样本性质；进一步，将相关理论研究成果应用于碳排放影响因素分析中。通过本项目的研究，目的在于为半参数空间面板模型提供一些新方法和新理论，同时也为应用研究提供更加有力的分析工具。

The study on estimation method of semiparametric spatial autoregressive panel model not only has great practical significance, but also has important academic value for enriching and developing nonparametric spatial econometrical theory. At present, the research in this field mainly focuses on the case that the longitudinal section time series of panel data (i.e. within individuals) are independent stochastic processes, which has a big gap compared with the actual problems in economics, environtology and so on. This project plan to research estimation methods of semiparametric spatial autoregressive panel models which have some correlation structure within individuals. The specific research contents include: we will construct efficient estimation methods respectively for partially linear single index spatial autoregressive panel model with correlation within individuals, partially linear varying coefficient spatial autoregressive panel model with correlation within individuals and partially linear additive spatial autoregressive panel model with correlation within individuals, and the asymptotic properties of these models estimators were derived under some rule conditions. Furthermore, we use above theoretical achievements to study influence factors of carbon emissions. By research of this project, we tend to establish the new methods and new theories for spatial nonparametric econometrics, and offer application researcher more powerful research tools.