基于空间扩展模型和地理加权回归模型的城市住房价格空间分异比较

doi:10.11821/dlyj201507013

摘要/Abstract

摘要：

鉴于已有研究主要集中探讨住房价格的空间依赖性,较少涉及空间异质性对住房特征价格的影响,也很少尝试构建不同计量模型来比较模型间刻画住房价格影响因素空间分异的准确性,以长沙市中心城区为研究区,采用空间扩展模型和地理加权回归模型比较分析城市住房价格影响因素的空间分异,结果表明：① 空间扩展模型和地理加权回归模型都表明,长沙市中心城区的住房属性边际价格随着区位的变化而变化,揭示住房价格影响因素具有显著的空间异质性;小区环境、交通条件、教育配套、生活设施等因素对住房价格的影响强度存在明显的空间分异。② 地理加权回归模型和空间扩展模型都能对传统特征价格模型进行改进,但地理加权回归模型在解释能力和精度方面都超过空间扩展模型;对属性系数估计空间模式的分析,地理加权回归模型形成的结果比采用坐标多义扩展的空间扩展模型更为复杂和直观。

关键词: 空间扩展模型, 地理加权回归模型, 住房价格, 空间异质性

Abstract:

Prior researches mostly focused on spatial dependence among house prices, ignoring the effects of spatial heterogeneity on house hedonic price; also, few comparative studies based on econometric models were conducted to obtain the accuracy of spatial variances on influencing factors of housing price. Considering the problems above and taking the center of Changsha as a research objective, this paper adopts spatial expansion model and geographic weighted regression model (GWR) to examine the spatial variances of factors that influence housing prices. The main findings are: (1) the analysis of spatial expansion model and GWR model shows that marginal prices of house attributes in the center of Changsha vary in different locations, indicating that the factors are significantly spatially heterogeneous, and some factors such as community environment, transportation conditions, educational facilities and living facilities have obvious spatial variances. (2) Spatial expansion model and GWR model can both modify traditional hedonic model; however, GWR model has stronger explanatory power and is more accurate in simulating the results than spatial expansion model; as for the analysis of attribute coefficient estimation spatial mode, the results from GWR model are more complicated and objective than those from spatial expansion model which adopts coordinate-based polysemy extension.

Key words: spatial expansion model, geographical weighted regression model, house price, spatial heterogeneity

孙倩, 汤放华. 基于空间扩展模型和地理加权回归模型的城市住房价格空间分异比较[J]. 地理研究, 2015, 34(7): 1343-1351.

Qian SUN, Fanghua TANG. The comparison of city housing price spatial variances based on spatial expansion and geographical weighted regression models[J]. GEOGRAPHICAL RESEARCH, 2015, 34(7): 1343-1351.

图/表 6

图1

表1

表2

表3

空间扩展模型估计结果"

模型	回归系数	Std. Error	t值		Sig.
（常数项）	9.118	0.221	41.237		0.000
Q-envi	0.18	0.023	3.895		0.003
N-bus	0.042	0.011	3.977		0.003
D-xriver	-13.231	0.692	3.817		0.003
D-CBD	-727.698	0.032	-4.298		0.002
D-railway	-0.111	0.026	-7.913		0.000
$v$	1.055	0.104	6.924		0.000
$v 2$	1.054	0.227	5.894		0.000
Q-envi $× u 2$	5.224E-5	0.000	11.653		0.000
D-xriver $× u$	-10.638	0.012	-8.718		0.000
D-xriver $× v$	-26.434	0.169	-4.801		0.001
D-xriver $× v 2$	-5.511E-5	0.000	-4.534		0.001
D-CBD $× v$	0.051	0.015	3.374		0.007
D-CBD $× v 2$	-0.002	0.001	-3.646		0.004
D-railway $× u$	-0.111	0.004	-4.916		0.001
D-railway $× v$	-0.112	0.012	-3.909		0.003
D-railway $× u 2$	-0.111	0.215	-3.919		0.003
D-railway $× v 2$	-0.114	0.000	-3.919		0.003
D-railway $× uv$	-0.112	0.242	-3.912		0.003
R²	R²（adj）	Std. Error	F值	Sig.	Durbin-Watson
0.798	0.792	0.119	135.358	0.000	1.996

表3

表4

图2

参考文献 17

[1]	沈铁雁, 冯等田, 孙铁山. 空间计量经济学. 北京: 北京大学出版社, 2011.
	[Shen Tieyan, Feng Dengtian, Sun Tieshan. Spatial Econometrics.Beijing: Peking University Press, 2011.]
[2]	Anselin L, Varga A, Acs Z J.Geographic and sectoral characteristics of academic knowledge externalities. Papers in Regional Science, 2000, 79(4): 435-443.
[3]	Raymond Y C T. Estimating neighborhood effects in house prices: Towards a new hedonic model approach. Urban Studies, 2002, 39(7): 1165-1180.
[4]	Dubin R, Pace R K, Thibodeau T G.Spatial autoregression techniques for real estate data. Journal of Real Estate Literature, 1999, 7(1): 79-95.
[5]	Steven C, Bourassa S C.Spatial dependence, housing submarkets, and house price prediction. Journal of Real Estate Finance and Economics, 2007, 35(2): 143-160.
[6]	Brasington D M, Hite D.Demand for environmental quality: A spatial hedonic analysis. Regional Science and Urban Economics, 2005, 35(1): 57-82.
[7]	Anselin L, Lozano-Gracia N.Spatial hedonic models. Palgrave Handbook of Econometrics, 2009, 2(26): 1213-1250.
[8]	Anselin L.Spatial Econometrics: Methods and Models. Berlin: Springer Science & Business Media, 1988.
[9]	Schnare A, Struyk R.Segmentation in urban housing markets. Journal of Urban Economics, 1976, 3(2): 146-166.
[10]	Quigley J.Consumer choice of dwelling, neighborhood and public services. Regional Science and Urban Economics, 1985, 15(1): 41-63.
[11]	周敏, 甄峰. 基于空间分析的城市商品住宅价格空间分布研究: 以南京市2007年开盘在售商品住宅为例. 现代城市研究, 2008, 23(7): 47-53.
	[Zhou Min, Zhen Feng.The spatial distribution of commodity housing and price in Nanjing: Based on the spatial analysis. Modern Urban Research, 2008, 23(7): 47-53.]
[12]	孟斌, 张景秋, 王劲峰, 等. 空间分析方法在房地产市场研究中的应用: 以北京市为例. 地理研究, 2005, 24(6): 956-965.
	[Meng Bin, Zhang Jingqiu, Wang Jinfeng, et al.Application of spatial analysis to the research of real estate: Taking Beijing as a case. Geographical Research, 2005, 24(6): 956-965.]
[13]	龙奋杰, 郑思齐, 王轶军, 等. 基于空间计量经济学模型的城市公共服务价值估计. 清华大学学报: 自然科学版, 2009, 42(12): 2028-2031.
	[Long Fenjie, Zheng Siqi, Wang Yijun, et al.Value estimates of local public services using a spatial econometric model. Journal of Tsinghua University: Science and Technology, 2009, 42(12): 2028-2031.]
[14]	王霞, 朱道林. 地统计学在都市房价空间分布规律研究中的应用: 以北京市为例. 中国软科学, 2004, (8): 152-156.
	[Wang Xia, Zhu Daolin.The application of geostatistics in the study on spatial distribution discipline of metropolitan house prices: A case study of Beijing. China Soft Science Magazine, 2004, (8): 152-156.]
[15]	Cassetti E.Generating models by the expansion method: Applications to geographical research. Geographical Analysis, 1972, 4(1): 81-92.
[16]	Cassetti E, Jones J P.Spatial aspects of the productivity slowdown: An analysis of US manufacturing data. Annals of the Association of American Geographers, 1987, 77(1): 76-88.
[17]	Brunsdon C, Fotheringham A S, Charlton M E.Geographically weighted regression: A method for exploring spatial nonstationarity. Geographical Analysis, 1996, 28(4): 281-298.

特征分类	解释变量	变量描述及计算方法	均值
因变量	住房价格（ZFJG）	样本楼盘建筑面积销售均价CPI指数平减后的对数值	8.76
内生特征	容积率（R-plot）	样本楼盘容积率的对数值	1.15
	绿化率（R-green）	样本楼盘绿化率的对数值	3.66
	物业管理（Q-property）	样本楼盘物业管理资质,分为四级：一级（4分）、二级（3分）、三级（2分）、三级暂定（1分）,取对数	1.35
	小区环境（Q-envi）	样本楼盘环境质量,五级评分,量化为1~5分,取对数	1.15
邻里特征	交通条件（N-bus）	样本楼盘500 m以内公交站点数的对数值	2.28
	教育配套（D-school）	样本楼盘到最近重点小学、中学距离的对数值	7.21
	医疗配套（D-hospital）	样本楼盘到最近三甲医院距离的对数值	7.62
	生活配套（D-smarket）	样本楼盘到最近大型超市距离的对数值	7.41
区位特征	邻近CBD距离（D-CBD）	样本楼盘到最近CBD距离的对数值	8.88
	邻近湘江距离（D-xriver）	样本楼盘到湘江最近距离的对数值	8.26
	邻近火车站距离（D-railway）	样本楼盘到火车站或火车南站最近距离的对数值	8.92

模型	回归系数	Std. Error		t值		Sig.
（常数项）	9.574^***	0.283		33.782		0.000
R-plot	-0.010	0.020		-0.247		0.805
R-green	-0.012	0.024		-0.332		0.740
Q-property	0.017	0.117		0.398		0.691
Q-envi	0.662^***	0.059		11.179		0.000
N-bus	0.029^**	0.013		2.154		0.033
D-school	-0.019^*	0.015		-1.907		0.086
D-hospital	-0.040^*	0.016		-1.923		0.083
D-smarket	-0.037^*	0.014		-1.931		0.082
D-CBD	-0.090^***	0.023		-3.879		0.000
D-xriver	-0.045^***	0.010		-4.317		0.000
D-railway	-0.054^**	0.022		-2.393		0.018
R²	R²（adj）	Std. Error	F值	Sig.	Durbin-Watson
0.686	0.680	0.123	125.572	0.000	1.954

	回归系数	t值	25%分位数	中位数	75%分位数	Sig.
（常数项）	9.868	25.472	9.796	9.823	9.959	0.000
R-plot	-0.008	-0.425	-0.011	-0.007	-0.004	0.680
R-green	-0.010	-0.443	-0.013	-0.009	-0.008	0.667
Q-property	0.002	0.018	-0.006	0.001	0.012	0.986
Q-envi	0.262	10.199	0.257	0.264	0.269	0.000
N-bus	0.086	2.894	0.083	0.086	0.088	0.016
D-school	-0.043	-1.87	-0.045	-0.043	-0.041	0.091
D-hospital	-0.061	-2.219	-0.063	-0.060	-0.059	0.051
D-smarket	-0.054	-2.34	-0.055	-0.054	-0.053	0.041
D-CBD	-0.198	-4.651	-0.199	-0.198	-0.196	0.001
D-xriver	-0.085	-3.678	-0.092	-0.083	-0.075	0.004
D-railway	-0.100	-2.246	-0.102	-0.100	-0.099	0.049
R²	R²（adj）		Std. Error		Sig.
0.914	0.892		0.082		0.000