Spatial Econometric Analysis On Determinants Of Foreign Direct Investment In China

After adopting the open-door policy, China has experienced a boom of inward foreign direct investment (FDI). FDI jumps from virtually zero in1979to an amount of US＄74.76billion in2007. Indeed, today China is the largest developing country recipient of FDI. While China’s record in attracting FDI in the past decades has been impressive, the regional distribution of FDI within China has been skewed. The bulk of FDI to China has gone to the south-eastern coastal belt, with only a small portion received by the hinterland. Given the important role of FDI in China’s growth, it is apparent that the unevenness in the geographical distribution of FDI is contributing to the skewed pattern of the country’s regional growth as well as other discrepancies between regions. In order to narrow or slow down the widening of the gap, China’s central government has adopted a series of measures which includes encouraging FDI in the Central and Western regions. The provinces in these regions also try to jump onto the bandwagon to attract FDI. An understanding of factors affecting the location of FDI can provide guidance to policymakers in identifying the obstacles that some regions must overcome to attract FDI.A large volume of research has investigated locational determinants of FDI, both empirically and theoretically. The analyses traditionally assume each region to be an isolated and homogeneous entity. The role of spatial effects (spatial dependence and spatial heterogeneity) is completely neglected, even though it is an important force in the process of location selection and ignoring it could result in serious misspecification. As a result, previously measured parameter estimates and statistical inferences are questionable. The present dissertation distinguishes itself from the most existing studies by explicitly reconsidering the question of regional FDI distribution from a spatial perspective and employing spatial statistical and spatial econometrical techniques.Although much has been written describing China’s overall achievement in attracting FDI and the general pattern of Chinese FDI, little work has been done analyzing quantitatively the geographical attributes of such investment. Based on a brief historical overview of FDI in China, this paper studies the space-time dynamics of distribution of FDI in mainland China using the methods of Exploratory Spatial Data Analysis (ESDA). Using a sample of Chinese provincial per capita FDI over 3to2007, Moran’s1statistic reveals positive global spatial autocorrelation, which is persistent over the whole period:regions with relatively high (low) per capita FDI are and remain localized close to other regions with relatively high (low) per capita FDI and that the spatial distribution of regional per capita FDI is not random. Moran Scatterplots show that most of the eastern provinces in the high-high quadrant and the western ones in the low-low quadrant, which suggests there are two distinct spatial clusters of high and low values of per capita FDI. Moreover, LISA statistics identify and assess the significant presence of local spatial autocorrelation. These findings suggest that spatial effects should be considered carefully using appropriate spatial specification and proper econometric tools to achieve reliable statistical inference.At present most of the empirical researches on the locational determinants of FDI are usually based on bilateral framework. There are few articles focusing on various forms of FDI behavior and spatial effects. Based on the"third-country effects", using spatial panel technique and a data set on31provinces during the period of1998-2007, we estimate both spatial lag model and spatial error model to examine the spatial effects for FDI in China. The inclusion of spatial variables into the analysis enables us to conclude that complex-vertical FDI strategy dominanted in China. In most specifications, the spatial lag and spatial error coefficients are highly significant, which means spatial effects matter for regional FDI distribution. A province’s FDI inflow is affected by neighboring provinces. Market scale and transportation infrastructure attract FDI, while higher wages, excessive governmental interference and lower degree of marketization deter it. The human resource variable we try is not found to have statistically significant relationships with the level of FDI inflows across provinces.Based on data of244Chinese cities in2006, we study the determinants of location choice of FDI in mainland China using spatial econometrical techniques. The empirical results reveal that spatial relationships between Chinese cities matter significantly. Chinese cities take advantage not only of local location factors but also of FDI flows received by their neighbors. Market scale, infrastructure and agglomeration are significant factors to attract FDI inflows, while the influence of labor cost is insignificant. More detailed study suggests location factors have significantly different effect in eastern, central and western cities.In fact, the province-level units in China are significantly different from each other in terms of social, economic and geographical aspects. The relationships between the level of FDI and various factors might also vary over space. Indeed, such variation or spatial nonstationarity in these relationships is common in spatial data. If there is spatial nonstationarity, then the global estimate of spatial relationships will misrepresent the real geographical and economic phenomena. Global estimation techniques such as OLS should then be replaced by local estimation techniques such as geographically weighted regression (GWR). Coventional regression analysis can only produce "average" and "global" parameter estimates rather than "local" parameter estimeters which vary over space in some spatial systems. Using the GWR technique to explore spatial nonstationarity in the development of FDI in Chinese provinces, it is found that the GWR model is better than the OLS model, confirming the existence of spatial nonstationarity. GIS has been used to visualize the spatial variability of parameter estimeters.Yangtze Delta has experienced tremendous regional development and FDI inflow in recent years. The Moran’s I statistic suggests that FDI of city-level in Yangtze Delta are positively spatially autocorrelated, combing the information in a Moran scatterplot, it indicates the presence of spatial heterogeneity in the form of the Core-Periphery pattern. To further understand the mechanisms of regional FDI flow, we analyze the city-level spatial variability of the effects of various factors on FDI. The GWR model has been applied to unravel spatial nonstationarity by extending the traditional multiple regression model. The estimates indicate that labor cost, human resource, market scale, industrial structure and economic agglomeration are important determinants of FDI distribution in Yangtze Delta. It has been quantitatively demonstrated that the relationships between the level of FDI and various factors exhibited considerable spatial variability.At last, we present main findings, discuss their implications for FDI policy and future research in the conclusion section.