32 2 Vol.32 No. 2 2012 2 ECONOMIC GEOGRAPHY Feb. 2012 1 1 2 3 2010 12 1 014 OLS CBD OLS OLS F293.3 A 1000-8462(2012)02-0052 - 07 A GWR- Based Study on Spatial Patter n and Str uctur al Determinants of Shanghai s Housing Price TANG Qing - yuan 1,XU Wei 1,2,AI Fu - li 3 (1.The center for modern Chinese city studies,east China Normal University,Shanghai 200062,China;2. Department of Geography,University of Lethbridge,Lethbridge T1K 3M4,Alberta,Canada;3.Academy of Disaster Reduction and Emergency Management,Being Normal University,Being 100875,China) Abstract: Based on average housing price of 1014 residential quarters within the Outer Ring of Shanghai in December 2010, this paper establishes a geographically weighted regression model, and compares with least square method based on overall situation. It reveals the spatial differentiation of Shanghai housing price and impacts of different factors. According to the study. the effects of unit change in housing price influencing factor are ranked from high to low in order of building completion year, CBD, greening rate, distance to parks,distance to metro stations, distance to schools, and distance to supermarkets. In the meantime, GWR model provides better results than the traditional OLS model in goodness of fit and parameter estimation when spatial dependency is present in urban housing data, which help to reveal the complicated relationship between housing price and determinants over space. Moreover, the visualization tools allow to map the effects of model coefficients across urban landscape in detail, which traditional OLS methods are not on par with. Key words: housing price;spatial pattern;geographically weighted regression;shanghai 1 2011-08 - 16; 2011-12 - 03 11JJDZH006 747509H Lethbridge (Grant # 13253) 1980 E- mail tqy395@yahoo.com.cn
2 53 [1] 2010 12 1 014 [2] [3-5] [6] [7-8] Fotheringham Geographical Weighted Regression GWR [9] R &D GWR [10] 2009 OLS GWR, 100 10 [11] [13] 3 GWR [12] GWR GWR 3 GIS GWR 2 2007 4 2009 CBD ArcGIS www.anjuke.com 10 2 75% 663.5km 2 10% 3
54 32 ε i 4 OLS GWR β 0 β i k y i = β 0 + Σβ i x i + ε i i=1 GWR OLS y i x i OLS Tobler OLS [14] Geoda 0.542 GWR 0.6099 GWR 1 3D k y i = β u 0 i v + i Σβ u i i v i x ik + ε i i=1 u i v i i β i β i GWR i CBD [15] w = exp - 1 d Σ Σ 2 2 b b GWR GWR y i = β u 0 i v + i Σβ u 1 i v i x rate + Fotheringham GWR b GWR Σβ u 2 i v i x years + Σβ u 3 i v i x sub + Fig.1 1 3D 3D visualization of housing price distribution in Shanghai
2 55 Σβ u 4 i v i x sch + Σβ u 5 i v i x sup + Σβ u 6 i v i x CBD + Σβ u 7 i v i x park SAM Spatial Analysis in Macroecology OLS GWR 3 GWR 1 GWR OLS 1 OLS GWR Tab.1 Comparison between OLS and GWR Sigma AIC 2 R Adj F(R 2 ) P- value Residual OLS 0.226-129.8 0.408 100.331 0.00 51.31 GWR 0.029-530.9 0.666 14.937 0.00 25.09 OLS 40.8% GWR OLS - 530.9-129.8 Fotheringham 3 GWR GWR OLS 2 OLS OLS OLS 3 GWR Fig.3 Residual Plot of GWR 2 3 OLS GWR OLS CBD OLS 2 GWR 2 OLS Fig.2 Residual Plot of OLS GWR GWR Bi- squared 10% GWR 5 GWR 66.6% OLS 51.31 25.09 Sigma 5.1
56 32 Tab.2 2 OLS Coefficients of OLS model estimation Β coefficient t- statistic 1.0E+01 4.4E+02 4.0E- 03-4.7E+00-4.0E- 03 7.9E+00 1.6E- 05 2.3E+00-3.4E- 05-2.6E+00-8.3E- 06-1.2E+00 1.2E- 05 1.4E+00 CBD - 7.0E- 05-1.5E+01 Tab.3 3 GWR Coefficients of GWR model estimation 0.89% 5 5.2 CBD CBD CBD 7% 6 1.0E+01 1.0E+01 1.0E+01 1.1E+01 1.1E+01 1.4E- 03 2.7E- 03 4.0E- 03 5.9E- 03 1.6E- 02-2.8E- 02-1.4E- 02-8.9E- 03-4.2E- 03 7.3E- 03-1.5E- 04-2.0E- 05 <.00001 4.0E- 05 1.7E- 04-1.9E- 04-7.0E- 05-2.0E- 05 2.0E- 05 3.8E- 04-2.0E- 04-5.0E- 05 <.00001 2.0E- 05 1.2E- 04-2.0E- 04-6.0E- 05-3.0E- 05 1.0E- 05 1.3E- 04 CBD - 1.8E- 04-1.1E- 04-7.0E- 05-3.0E- 05 7.0E- 05 CBD CBD CBD CBD 1% 0.4% 7 CBD 5.3 4 OLS 1km Fig.4 4 Impacts of greening rate on the housing price Fig.5 5 Impacts of completion year on the housing price
2 57 Fig.6 6 CBD Impacts of CBD on housing prices Fig.7 7 Impacts of supermarket on housing prices 3.4% 8 GWR 9 5% 7% 10 182 1km 3% 3km 2 6% 6 ArcGIS 10 2 8 Fig.8 Impacts of Metro on Housing Prices 5.4 7 OLS GWR GWR
58 32 Fig.9 9 Impacts of schools on housing prices 10 Fig.10 Impacts of parks on housing prices [1] Stephen A Samaha Wagner A Kamakura. Assessing the market value of real estate property with a geographically weighted stochastic frontier model[j]. Real Estate Econmics 2008 36 (4) 717-751. [2] Wu F L. Housing provision under globalisation a case study of Shanghai[J].Environment and Planning A 2001 33 1 741-1 764. [3] Stephen malpezzi Gregory H Chun Richard K Green. New place - to - place housing price indexes for US metropolitan areas and their determinants[j]. Real Estate Economics 1998 26 (2) 235-274. [4]. [J]. 11 2010 6 43-50. Fig.11 Spatial variations in housing price trend in the [5] Allen C. Goodman. Hedonic prices indices and housing inner city of Shanghai markets[j]. Jouranl of Urban Economics 1978 5 471-484. [6] Li Z G Wu F L. Tenure - based residential segregation in post - OLS GWR reform Chinese cities a case study of Shanghai[J]. Transactions of the Institute of British Geographers 2008 7 404-419. [7] Wu F L.Sociospatial differentiation in urban China evidence from Shanghai's real estate markets[j]. Environment and Planning A OLS 2002 34 1 591-1 615. GWR [8]. [J]. 2011(2) 33-40. [9] Tu J Xia Z G. Examining spatially varying relationships between OLS land use and water quality using geographically weighted regression I Model design and evaluation[j]. Science of the Total Envi- ronment 2008 407 358-378. [10]. R&D GWR [J]. 2007(2) 145-153. GWR [11]. GWR [J]. 2009 23 (10) 20-25. GWR [12]. GWR [J]. 2010 30(3) 472-478. [13].. [J]. 2009(9) 26-27. [14]. [J]. 2011 31(11) 1 906-1 911. [15]. [J]. 2005 (4) 79-81.