1# RECORD 80c7a3882a0d8e30de9d0f2d3d4f1558e6df1c502# TITLE: The Transitional Spatial Pattern of Housing Prices in Beijing: Factors and Implications3# YEAR: 2014.0 VENUE: nan DOI: 10.14246/irspsd.2.3_464# AUTHORS: Fang Wang; Xiao-Lu Gao5# PAGES: 17 CANDIDATE TABLE PAGES (1-based): [13]6# ABSTRACT: : The shift from the free-allocation housing welfare system to a market system was one of the most important parts of the economic system reform implemented in China from 1998 onwards. In recent years, however, soaring housing prices have become a serious public concern. In this study, we explored the spatial characteristics of housing prices, using spatial data on the average transaction price of Beijing’s housing blocks from 2005 and 2012, in order to reveal the factors influencing changes in housing price and address the way in which such changes in turn can affect land use, transportation, the living environment, and quality of life. The analysis revealed that the spatial autocorrelation effects associated with housing prices (and their increase) became significantly stronger between 2005 and 2012, especially in the central part of Beijing. Autocorrelation was, however, identified for both years. This finding explains the emergence of sub-area markets in recent years. It also provides evidence which suggests that developers or real estate agencies may have boosted housing prices purposely, which can lead to a biased market environment. Targeted policies are required in order to address, and avoid, this problem in the future. The spatial patterns of housing prices in different parts of Beijing were examined in detail. The differentiation of housing prices showed that Beijing has, in the past decade, gradually become a polynuclear city. By comparing the data for the two years using a multivariate linear regression model, the factors influencing housing price growth were analyzed. Besides location and the environment, it was found that housing policies related to property rights, the construction of the transportation network, and population change all played a key role in explaining changes in the spatial pattern of housing prices in Beijing in recent years.789===== PAGE 1 (score 7, numeric-paren 0, table-words 0, var-words 7) =====101146 12International review for spatial planning and sustainable development, Vol.2 No.3 (2014), 46-62 13ISSN: 2187-3666 (online) 14DOI: http://dx.doi.org/10.14246/irspsd.2.3_47 15Copyright@SPSD Press from 2010, SPSD Press, Kanazawa 16The Transitional Spatial Pattern of Housing Prices in 17Beijing: Factors and Implications 18Fang Wang1, 2 and Xiaolu Gao1* 191 Key Laboratory of Regional Sustainable Development Modeling, Institute of Geographic 20Sciences and Natural Resources Research, Chinese Academy of Sciences 212 University of Chinese Academy of Sciences 22* Corresponding Author, Email: gaoxl@igsnrr.ac.cn23Received 23 August, 2013; Accepted 24 Feberuary 2014 24Keywords: 25Housing Price, Spatial Autocorrelation, Spatial Structure, Beijing 26Abstract: 27The shift from the free-allocation housing welfare system to a market system 28was one of the most important parts of the economic system reform 29implemented in China from 1998 onwards. In recent years, however, soaring 30housing prices have become a serious public concern. In this study, we explored 31the spatial characteristics of housing prices, using spatial data on the average 32transaction price of Beijing’s housing blocks from 2005 and 2012, in order to 33reveal the factors influencing changes in housing price and address the way in 34which such changes in turn can affect land use, transportation, the living 35environment, and quality of life. 36The analysis revealed that the spatial autocorrelation effects associated with 37housing prices (and their increase) became significantly stronger between 2005 38and 2012, especially in the central part of Beijing. Autocorrelation was, however, 39identified for both years. This finding explains the emergence of sub-area 40markets in recent years. It also provides evidence which suggests that 41developers or real estate agencies may have boosted housing prices purposely, 42which can lead to a biased market environment. Targeted policies are required in 43order to address, and avoid, this problem in the future. 44The spatial patterns of housing prices in different parts of Beijing were 45examined in detail. The differentiation of housing prices showed that Beijing 46has, in the past decade, gradually become a polynuclear city. By comparing the 47data for the two years using a multivariate linear regression model, the factors 48influencing housing price growth were analyzed. Besides location and the 49environment, it was found that housing policies related to property rights, the 50construction of the transportation network, and population change all played a 51key role in explaining changes in the spatial pattern of housing prices in Beijing 52in recent years. 531.54INTRODUCTION55The top priority for every family, housing constitutes an issue that is also 56relevant to the welfare, social stability, and economic development of a nation. 57From 1998 onwards, China has undergone intense urban housing system 58@Licensee SPSD Press 59This open access article is published under a 60Creative Commons [Attribution-61NonCommercial-NoDerivatives 4.0 62International] license. 63https://creativecommons.org/licenses/by-nc-nd/4.0/ 646566===== PAGE 13 (score 23, numeric-paren 0, table-words 7, var-words 6) =====676858 69IRSPSD International, Vol.2 No.3 (2014), 46-62 70 714.2 72Linear regression model 73This study used a semi-log linear regression model, which has a better 74imitative and interpretation quality. Regression analysis was conducted, and 75multicollinearity detected by a stepwise regression technique (significance 76level is 0.05) in the SPSS16.0 (Table 5 and Table 6) using the data from 2005 77and 2012, respectively. The results show that 13 variables entered the model 78in 2005 and 14 variables in 2012. Whilst both models have good explanatory 79power (adjusted R2=0.771 in 2005, adjusted R2=0.549 in 2012), the imitative 80quality was found to be better in 2005 than in 2012, probably because the 81influencing factors were more complicated in 2012. Not all the variables 82entered into the model in 2005 had strong multicollinearity. In 2012, the three 83property levels in the variables had good multicollinearity. However, through 84testing – by changing the variables randomly in the model – the three property 85types were all found to have strong significance and their regression 86coefficients were found to change little. As such, the models are considered 87reliable. 88Table 5. Regression results of housing price in 2005 89 90B 91Std. 92Error 93Beta 94t 95sig 96Tolerance VIF 97 98(Constant) 998.530 .057 100 101149.613 .000 102 103Sub-center 104distance 105-.034 .003 106-.360 -12.189 .000 .338 1072.955 108Property costs 109.079 110.006 111.261 11212.265 .000 .653 1131.530 114Apartment 115.416 116.052 117.373 1188.004 119.000 .136 1207.371 121Third-level 122hospital 123.107 124.023 125.089 1264.577 127.000 .781 1281.281 129Urban 130center 131distance 132-.011 .002 133-.140 -4.588 134.000 .315 1353.175 136Building age 137-.025 .004 138-.114 139-6.429 140.000 .944 1411.060 142Key high school 143.071 144.025 145.054 1462.801 147.005 .780 1481.281 149Plot ratio 150.013 151.004 152.068 1533.678 154.000 .864 1551.158 156Number 157of 158bus 159station 160.006 161.002 162.058 1633.141 164.002 .854 1651.171 166Park 167.060 168.020 169.056 1703.000 171.003 .854 1721.171 173Cottage 174.373 175.086 176.091 1774.331 178.000 .661 1791.513 180Common residence .169 181.048 182.160 1833.545 184.000 .144 1856.939 186Subway station 187.039 188.019 189.038 1902.085 191.037 .897 1921.115 193Dependent Variable: ln (2005 housing price) 194Table 6. Regression results of housing price in 2012 195 196B 197Std. 198Error 199Beta t 200sig 201Tolerance VIF 202 (Constant) 2039.927 .097 204 205102.196 .000 206 207 208Sub-center 209distance 210-.033 .001 211-.399 -24.193 212.000 .330 2133.031 214Urban 215center 216distance 217-.016 .001 218-.244 -12.482 219.000 .235 2204.247 221222223===== PAGE 14 (score 13, numeric-paren 0, table-words 1, var-words 9) =====224225Fang Wang and Xiaolu Gao 22659 227 228Property costs 229.041 230.002 231.184 16.388 232.000 .715 2331.399 234Cottage 235.518 236.092 237.248 5.613 238.000 .046 23921.834 240Building age 241.005 242.001 243.077 7.174 244.000 .787 2451.271 246Third-level 247hospital 248.075 249.011 250.073 6.911 251.000 .796 2521.256 253Common 254residence 255.333 256.091 257.303 3.666 258.000 .013 25976.013 260Greening ratio 261.002 262.000 263.041 4.043 264.000 .884 2651.131 266Population density .001 267.000 268.029 2.230 269.026 .546 2701.830 271Subway station 272.038 273.011 274.035 3.599 275.000 .977 2761.024 277Apartment 278.282 279.091 280.229 3.089 281.002 .016 28260.963 283Number of bus 284station 285.002 286.001 287.047 2.833 288.005 .327 2893.054 290Plot ratio 291-.005 .002 292-.026 -2.676 293.007 .916 2941.091 295Key high school 296.031 297.012 298.027 2.652 299.008 .851 3001.175 301Dependent Variable: ln (2012 housing price) 3024.3 303Analysis of influencing factors 3044.3.1 305Location and traffic 306Location characteristics play a very important role in determining housing 307prices. This study found that the housing price decreased as distance to the 308urban center increased; as such, prices obeyed the rent distance attenuation 309law. A 1-km increase in the distance to the urban center was found to lead to a 3101.1% decrease in price per square meter in 2005 and a 1.6% decrease in 2012, 311respectively. At the same time, the variable describing the distance to the 312nearest sub-center also exerted a powerful, negative influence on housing 313prices in both of the years studied. Furthermore, the influence coefficients for 314the sub-center variable were greater than those generated by the urban center 315variable, indicating that sub-centers exert greater influence than the urban 316center, and thus that the polycentric pattern of Beijing has, at least initially, 317now been established. 318In addition, variables relating to transport convenience within the location 319characteristics were also found to have a significant and positive impact on 320housing prices. The variables relating to the proximity of a subway station and 321to the number of bus stations were both approved in a significance test for 322both of the years studied. Holding all other conditions the same, the unit 323housing prices of housing blocks that have a subway station within 1500 m 324were found to be about 3.9% and 3.8% higher per square meter in 2005 and 3252012, respectively, than those without a subway station (we used the subway 326station plan to 2015 for the 2012 subway station data). Further, each additional 327bus station was found to raise the housing price by 0.6% per square meter in 3282005 and 0.2% in 2012. Convenient transport access is good for trade 329development and population centralization; as such, it will stimulate the 330development of commercial housing. Transport will also directly influence 331the level of convenience experienced in people’s daily life – for instance, 332commuting and shopping, constituting a significant factor to consider when 333residents purchase houses. 334