Is the spatial distribution of China’s population excessively unequal?: A cross-country comparison

Abstract

Is the spatial distribution of China`s population excessively unequal? So far, China has strictly controlled domestic migration to slow down the phase of urban growth. Such public action implicitly assumes that China`s population distribution in space is exceedingly unequal, so diseconomies of agglomeration dominate positive externalities from it. The validity of this assumption, however, is subject to scientific testing and requires empirical evidence that the spatial distribution of China`s population substantially deviates from a certain optimal range or a widely accepted reference level. Given the lack of such an empirical test, this study is motived to fill the gap.In detail, I test whether the spatial inequality of China`s population distribution deviates upward from that of other countries when controlling for a set of socioeconomic variables. For the purpose of this hypothesis testing, I estimate the following fixed effects model, and interpret region-specific fixed effects (µ_i) as a systemic bias in spatial distribution of population:y_(i,t)=x_(i,t)^` ß+µ_i+e_(i,t) In the model, y_(i,t) and x_(i,t) are spatial inequality of population distribution and a vector of control variables for country i and time t, respectively, and ß and e_(i,t) refer to a vector of parameters to be estimated and error terms, respectively.Primary data for analysis (panel data for 65 countries) is 0.25°×0.25° global population grids for five years (1990, 1995, 2000, 2005, and 2010). I estimate the grids for China from China`s official census data and county-level statistics, while for other countries I directly use the grids excerpted from the Gridded Population of the World version 3 (GPWv3) dataset (SEDAC, 2014). I consider two measures of spatial inequality of population distribution (y_(i,t)). One is the spatial Gini coefficient, measuring spatial inequality across cities; the other is Moran`s I index, measuring spatial inequality across clusters of cities. My preliminary analysis shows that the spatial Gini coefficient for China is not biased upward, while Moran`s I index is. In other words, the distribution of China`s population is not excessively unequal at the grid cell level, but those grid cells with high population counts tend to be highly agglomerated, compared with other countries. This results suggests that the spatial inequality of China`s population distribution is more obvious at the regional level than at the city (or county) level.