Abstract: Based on the patent dataset at the province level in China from 1990 to 2008, the authors attempt to apply spatial Markov chains to investigate the space-time evolutionary characteristics of regional innovation convergence in China. The results indicate: (1)The process of regional innovation convergence in China has been globally characterized by “convergence clubs” since 1990, but this trend from 2000 to 2008 is sharpened and statistically different from that of the period from1978 to 1990. (2) The regions and their neighbors that both experience upward mobility are located in southeast coastal areas of China, while the regions or their neighbors that move downwardly are mostly found in central and western areas of China. (3) Regional innovation class transitions in China are highly constrained by their geographical neighbors, the regions are positively influenced by highly innovative regions and are negative influenced by less innovative regions. These empirical analyses provide a spatial explanation to "convergence clubs" of regional innovation in China.

Key words: regional innovation convergence, space-time evolutionary characteristics, spatial Markov chains, patent, China