SCIENTIA GEOGRAPHICA SINICA ›› 2010, Vol. 30 ›› Issue (5): 679-685.doi: 10.13249/j.cnki.sgs.2010.05.679

Previous Articles     Next Articles

Spatial Dependence,Heterogeneity and Economic Convergence of the Beijing-Tianjin-Hebei Metropolitan Region

DONG Guan-peng1,2, GUO Teng-yun1, MA Jing3   

  1. 1. Institute of Geographic Sciences and Natural Research, Chinese Academy of Sciences, Beijing 100101;
    2. Graduate University of Chinese Academy of Sciences, Beijing 100049;
    3. Department of Urban and Economic Geography, Peking University, Beijing 100871
  • Received:2010-03-05 Revised:2010-06-03 Online:2010-09-20 Published:2010-09-20

Abstract: The aim of this paper is to illustrate that spatial dependence and spatial heterogeneity do matter in the estimation of the β-convergence process on a sample of 140 county-level regions of the Beijing-Tianjin-Hebei Metropolitan Region(BTHMR) over the 2001-2007 period. As of the problem of spatial autocorrelation, spatial econometric tools such as spatial error model and spatial lag model were used. Concerning spatial heterogeneity, two methods are adopted, one of which is the spatial error spatial regimes model was estimated, the other is the GWR (geographically weighted regression) model estimated with Bayesian methods due to spatial heteroskedasticity and spatial outliers. Two spatial regimes, interpreted as spatial convergence clubs or core-periphery spatial pattern, are defined using Exploratory Spatial Data Analysis. Based on these techniques, several conclusions are draw as follows:① The estimation of Getis-Ord statistics shows that there has evolved a distinct core-periphery spatial structure. The core areas include Beijing, Tianjin and Tangshan mainly, while the periphery areas include mainly Zhangjiakou and Baoding, which are surrounding Beijing and Tianjin. ② The estimation of appropriate spatial regimes spatial error model shows that indeed the convergence process is different across the two regimes. As a matter of fact, there is no such a convergence process for the periphery regions, which surround Beijing and Tianjin; however, the core regions have a statistically significant β-convergence, and the speed of convergence associated with this estimation is 5.3% (the half-life is 13.2 years), far above 2% usually found in the convergence literature. The reason is that the spatial spillover effect among core areas is very large, and they are more similar in economic structure and have more proficient labors, better industrial infrastructure, making them easier to absorb the knowledge and technology spillovers.③ There also exists spatial structure instability inside the two spatial regimes on the β coefficient indicated by the GWR models estimated by Bayesian methods. Specifically, some of the core areas such as most counties of Beijing do not show β-convergence due to the special political status of Beijing, while most counties of Zhangjiakou have a statistical significant β-convergence. On one hand, the findings of the spatial regimes spatial error model are not contradictory with the results of the GWR model. The GWR model is designed to investigate the absolute spatial heterogeneity while the spatial regimes spatial error model is designed to explore the relative spatial heterogeneity which is on the assumption that areas in the same spatial regime are homogenous. On the other hand, the results of the GWR model is almost the same as the findings of the spatial regimes spatial error model indicating the findings from the spatial regimes spatial error model are robust. ④ With the appropriate spatial regimes spatial error model, this paper simulates the spatial spillover effect in BTHMR through random positive shocks to Tangshan in the core area and Quyang county of Baoding in the periphery area. The results show that there is a clear spatial pattern of the magnitude of spatial spillover effect which conforms to the law of geographical attenuation.

CLC Number: 

  • K901