Spatial-temporal dynamics of regional convergence at County level in Jiangsu

The dynamics of regional convergence include spatial and temporal dimensions. Spatial Markov chain can be used to explore how regions evolve by considering both individual regions and their geographic neighbors. Based on per capita GDP data set of 77 counties from 1978 to 2000, this paper attempts to investigate the spatial-temporal dynamics of regional convergence in Jiangsu. First, traditional Markov matrix for five per capita GDP classes is constructed for later comparison. Moreover, each region’s spatial lag is derived by averaging all its neighbors’ per capita GDP data. Conditioning on per capita GDP class of its spatial lag at the beginning of each year, spatial Markov transition probabilities of each region are calculated accordingly. Quantitatively, for a poor region, the probability of moving upward is 3.3% if it is surrounded by its poor neighbors, and even increases to 18.4% if it is surrounded by its rich neighbors, but it goes down to 6.2% on average if ignoring regional context. For a rich region, the probability of moving down ward is 1.2% if it is surrounded by its rich neighbors, but increases to 3.0% if it is surrounded by its poor neighbors, and averages 1.5% irrespective of regional context. Spatial analysis of regional GDP class transitions indicates those 10 upward moves of both regions and their neighbors are unexceptionally located in the southern Jiangsu, while downward moves of regions or their neighbors are almost in the northern Jiangsu. These empirical results provide a spatial explanation to the "convergence clubs" detected by traditional Markov chain.