Portal de Periódicos da CAPES

Economic growth and growth linkages in China 1994–2003

2010; Routledge; Volume: 2; Issue: 3 Linguagem: Inglês

10.1080/17538960903529477

1753-8971

Ari Kokko, Christer Ljungwall, Patrik Gustavsson Tingvall,

China's Socioeconomic Reforms and Governance

Abstract This paper investigates to what extent income growth in the Chinese provinces is linked to growth and income levels in neighboring provinces. We find that the rate of income growth in a province is positively related to income and growth in neighboring provinces. However, we find no evidence of such positive interdependence between growth in rich coastal provinces and their immediate inland neighbors. This suggests that there has been little synchronization in economic growth rates between these regions, and/or that the immediate hinterland of the coastal growth centers might have been bypassed as China's manufacturing sector has migrated westward. Keywords: domestic integrationeconomic growthspatial interdependenceChina's economy Notes 1. The conclusion falls back to Tobler's law roughly saying 'everything is related to everything but things far away are less related'. 2. This idea was widely used as a planning tool in developing countries during the import-substitution era in the 1950s and 1960s. Reidel (Citation1976) criticized Hirschman's view on backward linkages for implicitly assuming that demand creates its own supply. Instead, Reidel argued that in a country engaged in international trade, such an assumption would be naïve. 3. In his groundbreaking work Principles of economics, Marshall (Citation1920) argued that the decision of where to locate economic activities is affected by three categories of technology spillovers: (1) knowledge spillovers that are 'in the air'; (2) forward and backward linkages; and (3) labor market pooling. The first two concepts implicitly comprise distance, whereas the size of cities is central to the latter. Reasons to believe that knowledge is locally bounded are conveyed as five 'stylized facts' by Dosi (Citation1988), and are further developed by Feldman (Citation1994), and Baptista and Swann (Citation1998). 4. China is administratively divided into 31 provincial units which fall into three categories: provinces (a total of 22), autonomous regions (a total of five), and municipal cities (a total of four). Tibet is excluded from our sample because of data constraints and the province of Chongqing is kept within Sichuan. 5. For a detailed description on methodology, see Anselin (Citation1988), Anselin and Florax (Citation1995) and Anselin (Citation1995). 6. Estimates are performed in SpaceStat based on non-parametric estimation using 10,000 permutations. 7. Since in Table 2 we not are analyzing higher order contiguity we here and in subsequent analysis apply the inverse distance as preferred distance measure. 8. See e.g. Hsiao (Citation2003) and Baltagi (Citation2008). 9. Examples of empirical spatial growth studies include Basile (Citation2008), Garrett et al., (Citation2007) and Moreno and Trehan (Citation1997), where the latter include spatial growth regressions with and without initial income on the right hand side. 10. For example, Weeks and Yao (Citation2002) apply a system GMM estimator and find evidence of a nationwide divergence process in China. Similarly, Pedroni and Yao (Citation2006) find evidence of divergence on both a national basis as well as in various regional and political subgroups. Lei and Yao (Citation2008) take a long-term perspective and find no convergence in the pre-reform period, but a slow (1–2% per year) convergence in the post-reform period. Having a regional perspective on China, Maasoumi and Wang (Citation2007), Anderson and Ge (Citation2009), and Takashi and Ryoichi (Citation2007) find convergence in some regions of China, but no convergence or even divergence in other areas. Finally, Zhang, Liu, and Yao (Citation2001) find evidence for provincial convergence toward their own steady state, as well as evidence of big shocks affecting the regional income distribution. 11. Notably, some studies such as Fagerberg (Citation1994) and Durlauf and Quah (Citation1999) report inconclusive country evidence regarding the impact of population growth on per capita income. The possible endogeneity of population growth has been further discussed in Barro and Sala-i-Martin (Citation1995). At present, however, there does not appear to be any consensus regarding the issue. 12. An alternative measure of capital growth is defined as investments net of capital depreciation. 13. As there are possible endogeneity problems as well as an obvious time lag between finishing education and making an impact on the labor market, we use one year lagged values. 14. In Instrumental Variable (IV)-models the coefficient on the spatial lag may exceed unity. For details, see e.g. Anselin (Citation1988). 15. Following Anselin (Citation1988), we apply spatial lags of the exogenous variables as instruments. 16. Substituting the inverse distance matrix for a row standardized first order contiguity matrix lowers the coefficient of the spatial lag, but does not upset the significance of the results, indicating that the findings are robust with respect to choice of distance weight procedures. 17. Note that the inclusion of the spatial variable makes the significance of FDI and openness disappear. 18. See Ljungwall (Citation2004) and references therein. 19. Moran's I is –0.39 and significant at the 1% level. Estimations based on a robust permutation approach using 10,000 bootstrap permutations. 20. Moran's I is –0.01, p-value 0.49. Estimations based on a robust permutation approach using 10,000 bootstrap permutations. 21. The original version of Christaller's central place theory was published in German in 1933, while Von Thünen's localization theories were first published (also in German) previously in 1826. The references in the text are to the first English translations of the works. 22. One may note that the negative coefficient is a partial correlation. That is, both the hinterland and the growth pole may co-vary due to nationwide and period-specific nationwide trends. In addition, both regions have their own local growth factors.