论文部分内容阅读
国外大量研究通过估算地产价格梯度来解析城市空间结构的特征与发展,2004年以来国内城市的相关研究也得到关注。本文回顾了地产价格梯度研究的主要方法、基本假设与模型改进。研究中主要采用了特征价格模型与重复销售模型两种方法,其中特征价格模型使用频率最高,非参数估计、样条函数等能够揭示价格梯度复杂特征的新方法也得到了应用。城市空间结构预设主要包括单中心假设、非单中心或多中心假设和无中心预设三种,单中心模型中,欧式距离变量应用最普遍,使用交通花费、交通时间或虚拟变量取代或改进距离变量,能够取得更优的估计结果。空间自相关带来的异方差问题容易导致模型估计的无效,如何检验和修正空间自相关得到了广泛的探索。最后从样本选取、变量设置、空间预设和模型应用方面展望了国内研究的方向。
A large number of foreign studies have analyzed the characteristics and development of urban spatial structure by estimating the gradient of real estate prices. Since 2004, relevant studies on domestic cities have also drawn attention. This article reviews the main methods, basic assumptions and model improvements of the real estate price gradient. Two methods, the feature price model and the repeated sales model, are used in the study. Among them, the most frequently used feature price model, non-parametric estimation, spline function and other new methods that can reveal complex features of price gradient have also been applied. The presupposition of urban spatial structure mainly includes single-center assumption, non-single-center or multi-center assumption and no-center preset. In single-center model, the application of the European distance variables is the most common, and the traffic cost, traffic time or dummy variables are substituted or improved Distance variables, to obtain better estimates. Heteroscedasticity problems caused by spatial autocorrelation easily lead to invalid model estimation, and how to test and correct spatial autocorrelation has been widely explored. Finally, the direction of domestic research is prospected from the aspects of sample selection, variable setting, space pre-setting and model application.