China's interregional capital mobility: A spatial econometric estimation

•The traditional Feldstein-Horioka test is extended to be a spatial model.•FH coefficients in this paper differ to those in the existing literature.•A proxy variable is use to demonstrate capital mobility directly.•Four spatial weight matrices are devised from the perspectives of geographical features and economic characteristics.

Free interregional mobility of capital refers to the efficiency of resource allocation. Using a general spatial autocorrelation model that extends the traditional Feldstein-Horioka (FH) test, we analyze Chinese provincial data for the periods before and after 1993, and show that the FH coefficients differ to those in the existing literature. During China's period of reform and opening up between 1979 and 1992, the FH coefficient is significantly negative or not statistically significant. However, for the period after 1994, the FH coefficient is significantly higher than that shown in prior studies. This study explains this phenomenon from the perspective of Chinese-style decentralization. The FH test is an indirect method of showing regional capital mobility. Therefore, we need some form of direct measurement. Thus, we use the rate of the proportional change in physical capital stock as a proxy variable to demonstrate capital mobility directly, and use it to analyze various factors affecting Chinese provincial capital mobility. When investigating these factors, and to establish a realistic foundation on which the spatial effects take place, we devise four spatial weight matrices from the perspectives of geographical features and economic characteristics. Lastly, in order to promote regional capital mobility in China, the Chinese central government needs to take urgent measures to change local governments' strong preferences for GDP under the pressure of political promotion.