Acta Geographica Sinica ›› 2009, Vol. 64 ›› Issue (10): 1214-1220.doi: 10.11821/xb200910007

• Original Articles • Previous Articles     Next Articles

Driving Force Analysis of Residential Land Price in Beijing Based on Statistical Methods

WANG Zhen1,  GUO Huaicheng1,  HE Chengjie1,  LI Na1,  YU Yajuan2,  LIU Hui1,  FENG Changchun3   

  1. 1. College of Environmental Science and Engineering, Peking University, Beijing 100871, China;
    2. School of Chemical Engineering and the Environment, Beijing Institute of Technology, Beijing 100081, China;
    3. College of Urban and Environmental Sciences, Peking University, Beijing 100871, China
  • Received:2009-01-23 Revised:2009-07-14 Online:2009-10-16 Published:2009-10-16
  • Supported by:

    National Basic Research Program of China (973 Program), No.2005CB724205; Chinese Government Scholarships for Postgraduates, No.[2007]3020.


Statistical methods were employed in this paper to explore the driving forces of residential land prices in Beijing, including accessibility, land intensity, cultural and sport infrastructure and new transport methods. Box-Cox transformations, T-test, Pearson correlation, factor reduce and ridge regression were carried out to identify the key factors that influence the residential land price. Distances to the nearest CBD (P = 0.265), to the nearest road (P = 0.529), to the nearest schools (P = 0.202), to the nearest parks (P = 0.105) and to the nearest hospitals (P = 0.706), which had a low correlation with residential land price, were excluded by Pearson correlation test. Independent samples T-test showed that cultural and sport infrastructure (P = 0.003) and urban subways (P = 0.000) had statistical significant influence on residential land price. Thus, factors including distances to the central area and railway stations, plot ratio, public bus lines within 1 km, urban subways as well as cultural and sport infrastructure were studied in this paper. Factor reduce found that all the remaining factors could be divided into 4 groups. This result was used as one piece of judgment for the regression results, which should use at least one factor of each group. Ridge regression is one of the least-squares refinement methods. In this method, a biased constant is employed to find out a biased estimator, which helps to enhance the precision compared with least-squares methods. It has been proven that the ridge regression method is stable and valid when independent variables are highly correlated. Thus, the multicollinarity among the independent variables in this paper could be resolved by ridge regression analysis. Results of ridge regression indicated that the effects of the studied factors mentioned above accounted for 73.2% change of the independent variable Y in Beijing, and among which, the distance to the central area was the primary factor influencing the price of residential land, followed by the plot ratio. A negative correlation between distance and land price and a positive correlation between plot ratio and land price appeared respectively. Accessibility factors such as bus lines within 1 km had considerable effects on residential land price. Besides, urban subways and cultural and sport infrastructure had a significant value added function to residential land around. Based on statistical analysis, suggestions were proposed in this paper: (1) Land use rate could be improved by enhancing the accessibility and value of suburban areas via land use pattern change and urban subway construction to maximize the land use value. (2) Land use pattern of low efficiency such as 'urban village' could be presented to raise the intensification level of land use to optimize the urban function, thus the urban entity value increased. (3) Cultural and sport infrastructures could help to enhance the additional value of residential land price.

Key words: ridge regression, driving forces, residential land price, Beijing