重庆市主城区住房价格影响因子的空间异质性

doi:10.18402/resci.2017.02.15

摘要/Abstract

摘要：

以2015年重庆市主城区2449个住宅小区均价数据为基础,通过比较特征价格模型、空间扩展模型和地理加权回归模型,对住房市场进行模拟,寻求探索多中心山地城市住房价格影响因子空间异质性的最佳方法,进而对其空间异质性进行研究,以期为多中心城市的房地产市场管理提供有益借鉴。结果表明：①空间扩展模型和地理加权回归模型均对特征价格模型有所改进,地理加权回归模型在模拟效果与估计精度两个方面均优于空间扩展模型,并可将具体空间模式可视化,是探索空间异质性的最佳方法;②房龄、到城市中心的距离、到城市次中心的距离是影响住房价格最为重要的因子;地形位指数显著影响住房价格,且地形愈趋于平坦,房价越高;③各影响因子对住房价格的影响在不同空间有显著差异。重庆住房价格影响因子的空间格局相比单中心城市更为复杂,主要与重庆山水分割和“多中心、组团式”的城市格局密切相关;④多中心结构减弱了住房子市场的垄断,增加了住房的有效供给。优质公共设施在中梁山、铜锣山之间的狭长地带聚集,而郊区公共设施供应不足,需加大资金投入来改善郊区住房对优质公共设施的可达性。

关键词: 住房价格, 影响因子, 空间异质性, 空间扩展模型, 地理加权回归模型, 重庆市主城区

Abstract:

Based on price data of 2,449 housing projects in the main district of Chongqing in 2015,we used hedonic price modeling,spatial expansion modeling and geographically weighted regression （GWR）modeling to simulate the spatial heterogeneity in impact factors of housing price for this polycentric mountainous city. After comparing the three models,we found that the spatial expansion model and the GWR model performed better than the hedonic price model,and the GWR model performed better than the spatial expansion model when considering effectiveness and accuracy. In addition,the GWR model proved to be an effective method to explore spatial heterogeneity,which can reflect the spatial patterns of heterogeneity visually. A few variables,such as building age,distance to city center,and distance to city subcenters,were the most important factors affecting housing prices. The variable of TPI had a significant effect on housing price. When the terrain was flatter,the price would get higher,too. The effect of each factor on housing price varied spatially and significantly. The spatial pattern of polycentric cities such as Chongqing was more complex than for monocentric cities. The complexity was mainly considered to be closely related to the constraints of natural barriers and strategies of polycentric urban development in Chongqing. Polycentric development broke the monopoly of the housing submarket and increased the effective supply of housing in limited spaces of Chongqing. However,high-profile houses are still concentrated in the narrow valley floors between Zhongliang and the Tongluo Mountains where high-quality public facilities are located. Considering the lag in supply of public facilities in suburbs,substantial funds are needed to improve the accessibility of suburban housing to better public facilities.

Key words: housing price, impact factors, spatial heterogeneity, spatial expansion model, geographically weighted regression model, the main district of Chongqing

李颖丽, 刘勇, 刘秀华. 重庆市主城区住房价格影响因子的空间异质性[J]. 资源科学, 2017, 39(2): 335-345.

Yingli LI, Yong LIU, Xiuhua LIU. Spatial heterogeneity in factors affecting Chongqing housing prices[J]. Resources Science, 2017, 39(2): 335-345.

图/表 7

图1

表1

表2

表3

表4

表5

图2

参考文献 24

[1]	Wen H Z,Tao Y L.Polycentric urban structure and housing price in the transitional China:Evidence from Hangzhou[J]. Habitat International,2015,46:138-146.
[2]	温海珍,张之礼,张凌. 基于空间计量模型的住宅价格空间效应实证分析:以杭州市为例[J]. 系统工程理论与实践,2011,31(9):1661-1667.
	[Wen H Z,Zhang Z L,Zhang L.An empirical analysis on spatial effects of the housing price based on spatial econometric model:Evidence from Hangzhou City[J]. Systems Engineering-Theory & Practice,2011,31(9):1661-1667.]
[3]	Bourassa S C,Cantoni E,Hoesh M.Spatial dependence,housing submarkets,and house price prediction[J]. The Journal of Real Estate Finance and Economics,2007,35(2):143-160.
[4]	郑思齐. 公共品配置与住房市场互动关系研究述评[J]. 城市问题,2013,(8):95-100.
	[Zheng S Q.Review of the interaction between public goods and housing market configuration[J]. Urban Problems,2013,(8):95-100.]
[5]	董冠鹏,张文忠,武文杰,等. 北京城市住宅土地市场空间异质性模拟与预测[J]. 地理学报,2011,66(6):750-760.
	[Dong G P,Zhang W Z,Wu W J,et al.Spatial heterogeneity in determinants of residential land price:Simulation and prediction[J]. Acta Geo-graphica Sinica,2011,66(6):750-760.]
[6]	李峰清,赵民. 关于多中心大城市住房发展的空间绩效-对重庆市的研究与延伸讨论[J]. 城市规划学刊,2011,195(3):8-19.
	[Li F Q,Zhao M.A discussion on housing development in multi-cente cities:Phenomenon & planning implications of Chonging[J]. Urban Planning Forum,2011,195(3):8-19. ]
[7]	Fotheringham A S,Charlton M E,Brunsdon C.Geographically weighted regression:A natural evolution of the expansion method for spatial data analysis[J]. Environ Plan A,1998,30(11):1905-1927.
[8]	Brunsdon C,Fotheringham A S,Charlton M.Some notes on parametric significance tests for geographically weighted re-gression[J]. Journal of Materials Science,1999,39(3):497-524.
[9]	Casetti E.Generating models by the expansion method:Appli-cations to geographical research[J]. Geographical Analysis,1972,4(1):81-91.
[10]	Fik T J,Ling D C,Mulligan G F.Modeling spatial variation in housing prices:A variable interaction approach[J]. Real Estate Economics,2003,31(4):623-646.
[11]	Bitter C,Mulligan G F.Incorporating spatial variation in housing attribute prices:A comparison of geographically weighted regre-ssion and the spatial expansion method[J]. Journal of Geo-graphical Systems,2007,9(1):7-27.
[12]	汤庆园,徐伟,艾福利. 基于地理加权回归的上海市房价空间分异及其影响因子研究[J]. 经济地理,2012,32(2):52-58.
	[Tang Q Y,Xu W,Ai F L.A GWR-based study on spatial pattern and structural determinants of Shanghai’s housing price[J]. Economic Geography,2012,32(2):52-58.]
[13]	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.
	[Sun Q,Tang F H.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.]
[14]	隋雪艳,吴巍,周生路. 都市新区住宅地价空间异质性驱动因素研究-基于空间扩展模型和GWR模型的对比[J]. 地理科学,2015,35(6):683-689.
	[Sui X Y,Wu W,Zhou S L.Drive pattern on the spatial heterogeneity of residential land price in urban district:A comparison of spatial expansion method and GWR model[J]. Scientia Geographica Sinica,2015,35(6):683-689.]
[15]	王梦玮,刘勇,刘秀华. 多中心山地城市住房价格空间格局研究-以重庆主城区为例[J]. 西南大学学报(自然科学版) ,2016,38(5):133-138.
	[Wang M W,Liu Y,Liu X H.Spatial pattern and determinant factors of housing price in polycentric moun-tainous cities-the case of Chongqing city prope[J]. Journal Press of Southwest University(Natural Science Edition) ,2016,38(5):133-138.]
[16]	安居客重庆二手房房产信息网. 重庆普通商品住宅小区均价 [EB/OL].(2015-12-01)[2016-05-18].
	[Anjuke. Price Data of Ordinary Commercial Housing projects in Chongqing[EB/OL].(2015-12-01)[2016-05-18].
[17]	重庆市地理信息中心. 重庆市地理信息公共服务平台[EB/OL].(2015-04-24)[2016-05-20].
	[Chongqing Geographic Information Center. Chongqing Public Service Platform of Geographic Information[EB/OL].()[2016-05-20].
[18]	于璐,郑思齐,刘洪玉. 住房价格梯度的空间互异性及影响因素-对北京城市空间结构的实证研究[J]. 经济地理,2008,28(3):406-410.
	[Yu L,Zheng S Q,Liu H Y.The spatial variation and affecting factors of the housing price gradients:The case of Beijing[J]. Economic Geography,2008,28(3):406-410.]
[19]	余建辉,张文忠,董冠鹏. 北京市居住用地特征价格的空间分异特征[J]. 地理研究,2013,32(6):1113-1120.
	[Yu J H,Zhang W Z,Dong G P.Spatial heterogeneity in the attributes prices of residential land in Beijing[J]. Geographical Research,2013,32(6):1113-1120.]
[20]	焦贝贝,石培基,刘春芳,等. 黄土高原低山丘陵区农村居民点分布与地形因子关系研究-以兰州市七里河区为例[J]. 资源科学,2013,35(8):1719-1727.
	[Jiao B B,Shi P J,Liu C F,et al.The distribution of rural settlements in relation to land form factors in low hilly land on the loess plateau[J]. Resources Science,2013,35(8):1719-1727.]
[21]	张星星,刘勇,杨朝现. 重庆山地城市空间扩展形态的定量研究[J]. 西南大学学报(自然科学版),2015,35(10):119-124.
	[Zhang X X,Liu Y,Yang C X.Quantifying spatial patterns of urban expansion in mountainous cities:The case of Chongqing[J]. Journal Press of Southwest University(Natural Science Edition),2015,35(10):119-124.]
[22]	Weiss A.Topographic Position and Landforms Analysis[C]. San Diego,USA:Proceedings of ESRI User Conference,2001.
[23]	沈体雁,冯等田,孙铁山. 空间计量经济学[M]. 北京:北京大学出版社,2010.
	[Shen T Y,Feng D T,Sun T S.Spatial Econo-metrics[M]. Beijing:Peking University Press,2010.]
[24]	Hurvich C M,Simonoff J S,Tsai C L.Smoothing parameter selection in nonparametric regression using an improved Akaike information criterion[J]. Journal of The Royal Statistical Society Series A Statistics in Society,1998,60(2):271-293.

特征分类	解释变量	变量描述	均值	标准差
因变量
	住房价格PRICE	住宅小区均价的对数值	8.786	0.186
自变量
结构特征	容积率FAR	住宅小区容积率的对数值	1.060	0.411
	绿化率Greening rate	住宅小区绿化率的对数值	3.390	0.274
	房龄 Building age	住宅小区房龄的对数值	2.274	0.585
邻里特征	交通条件 N-Bus stations D-Metro stations	500m范围内公交站点数的对数值	1.827	0.630
	交通条件 N-Bus stations D-Metro stations	到最近轻轨站的距离的对数值	6.614	0.814
	教育设施D-Good primary schools D-Good middle schools	到最近重点小学的距离的对数值	7.174	1.006
	教育设施D-Good primary schools D-Good middle schools	到最近重点中学的距离的对数值	7.339	0.931
	医疗设施D-Top hospitals D-Common hospitals	到最近三甲医院的距离的对数值	7.797	1.115
	医疗设施D-Top hospitals D-Common hospitals	到最近普通医院的距离的对数值	6.051	0.926
	公园设施D-Municipal Parks D-Local Parks	到最近市级公园的距离的对数值	7.470	0.947
	公园设施D-Municipal Parks D-Local Parks	到最近区级公园的距离的对数值	7.360	0.877
区位特征	临近城市中心距离D-CBD	到城市中心（解放碑商圈）的距离的对数值	8.929	0.817
	临近城市次中心距离D-Subcenters	到最近城市次中心（观音桥商圈、沙坪坝商圈、杨家坪商圈、南坪商圈）距离的对数值	8.145	1.059
	临近河流距离D-River	到河流的距离的对数值	7.424	1.092
地形特征	地形位指数 TPI	高程值与其邻域高程平均值的差值的对数值	4.632	1.776

变量	B值	标准差	标准化B值	T值	Sig.	VIF
（常数项）	9.768***	0.124	-	78.854	0.000	-
FAR	0.042***	0.009	0.088	4.745	0.000	1.056
Greening rate	0.055***	0.013	0.081	4.278	0.000	1.096
Building age	-0.146***	0.007	-0.403	-19.951	0.000	1.257
N-Bus stations	0.025***	0.007	0.083	3.747	0.000	1.511
D-Metro stations	-0.011**	0.005	-0.047	-2.275	0.023	1.330
D-Good primary schools	-0.021***	0.005	-0.115	-4.256	0.000	2.258
D-Good middle schools	-0.014***	0.005	-0.073	-2.805	0.005	2.077
D-Top hospitals	-0.029***	0.004	-0.175	-5.979	0.000	1.948
D-Common hospitals	-0.014***	0.004	-0.070	-3.220	0.001	1.448
D-Municipal Parks	-0.007**	0.004	-0.034	-1.563	0.028	1.558
D-Local Parks	-0.003	0.005	-0.018	-0.792	0.428	1.442
D-CBD	-0.041***	0.006	-0.179	-6.605	0.000	2.260
D-Subcenters	-0.030***	0.005	-0.170	-6.480	0.000	2.120
D-River	-0.018***	0.004	-0.104	-4.300	0.000	1.807
TPI	-0.035*	0.019	-0.034	-1.832	0.067	1.048
R²=0.391 R²（adj）=0.386 Std.Error=0.156 F值=79.907 Sig.=0.000

变量	标准化B值	T值	Sig.	变量	标准化B值	T值	Sig.
Building age	-0.419***	-20.853	0.000	D-Good primary schools×uv	13.323**	2.281	0.023
N-Bus stations	154.038***	3.290	0.001	D-Good middle schools×u	-237.704**	-2.055	0.040
D-Good primary schools	-11.534**	-1.965	0.050	D-Good middle schools×u²	0.092***	3.844	0.000
D-Good middle schools	-115.442**	-1.992	0.047	D-Top hospitals×v	-0.136***	-4.693	0.000
D-Common hospitals	-101.617*	-1.891	0.059	D-Common hospitals×u²	-208.321*	-1.940	0.052
D-CBD	-0.119***	-3.740	0.000	D-Common hospitals×u	0.052**	2.478	0.013
D-Subcenters	-0.245***	-9.680	0.000	D-Municipal Parks×v	-0.012*	-0.695	0.087
FAR×u	0.089*	1.644	0.100	D-Municipal Parks×u²	-0.015*	-0.683	0.095
FAR×u²	0.077***	4.232	0.000	D-CBD×u	516.342***	2.746	0.006
FAR×v²	0.084*	1.645	0.100	D-CBD×v	-66.668***	-5.625	0.000
FAR×uv	0.089*	1.945	0.100	D-CBD×u²	231.223**	2.476	0.013
Greening rate×u²	0.074***	3.939	0.000	D-Subcenters×u	463.472***	3.444	0.001
Building age×u	-451.021***	-4.488	0.000	D-Subcenters×v	-52.512***	-5.313	0.000
Building age×u²	-224.870***	-4.474	0.000	D-Subcenters×u²	239.688***	3.549	0.000
N-Bus stations×u	307.443***	3.284	0.001	D-River×uv	0.050*	1.369	0.071
N-Bus stations×u²	0.077***	3.738	0.000	TPI×v	-0.045**	-2.455	0.014
D-Metro stations×uv	-0.041**	-2.112	0.035	v	0.159***	7.226	0.000
D-Good primary schools×u	12.446**	2.123	0.034	u²	0.076***	3.106	0.002
D-Good primary schools×v	-0.128***	-5.129	0.000	R²=0.459; R²（adj）=0.454; Std.Error=0.156 F值=69.875 Sig.=0.000

变量	平均值	标准差	最小值	最大值	下四分位值	中值	上四分位值	Sig.
（常数项）	9.721	0.660	8.073	13.153	9.282	9.823	10.086	0.005
FAR	0.025	0.017	-0.021	0.069	0.016	0.023	0.037	0.015
Greening rate	0.044	0.037	0.008	0.121	0.019	0.039	0.071	0.084
Building age	-0.167	0.021	-0.198	-0.065	-0.182	-0.166	-0.154	0.021
N-Bus stations	0.015	0.014	-0.021	0.042	-0.002	0.016	0.026	0.017
D-Metro stations	-0.021	0.020	-0.065	0.038	-0.035	-0.019	-0.009	0.018
D-Good primary schools	-0.021	0.019	-0.073	0.037	-0.035	-0.019	-0.005	0.022
D-Good middle schools	-0.015	0.032	-0.071	0.083	-0.029	-0.005	0.010	0.029
D-Top hospitals	-0.024	0.033	-0.115	0.025	-0.029	-0.017	-0.003	0.019
D-Common hospitals	-0.011	0.010	-0.038	0.035	-0.015	-0.012	0.018	0.010
D-Municipal Parks	-0.006	0.018	-0.089	0.079	-0.010	-0.001	0.007	0.013
D-CBD	-0.026	0.072	-0.503	0.248	-0.053	-0.017	0.036	0.006
D-Subcenters	-0.012	0.025	-0.298	0.180	-0.028	-0.008	0.007	0.026
D-River	-0.011	0.032	-0.051	0.095	-0.015	-0.012	0.036	0.038
TPI	-0.041	0.074	-0.271	0.093	-0.088	-0.034	0.023	0.083
R²=0.683; R²（adj）=0.669; Std.Error=0.022; F值=3.337; Sig.=0.000

	实际值90%置信区间	实际值95%置信区间	实际值99%置信区间	R²	接近实际值的比例
特征价格模型	0.287	0.295	0.363	0.415	0.497
空间扩展模型	0.351	0.407	0.431	0.449	0.516
GWR模型	0.399	0.450	0.500	0.609	0.598