Is China's regional inequality ethnic inequality?

What is the relevant geographic unit for the study of comparative economic development and regional inequality? Recent research suggests the importance of traditional ethnic homelands as a unit of study, and the relevance of geo-climactic features in shaping the contours of these homelands. I assess the relevance of ethnic homelands in understanding China's persistent regional inequality. I estimate spatial Gini coefficients for nighttime lights emissions across five different levels of spatial aggregation: provinces, counties, a 2.5 by 2.5 decimal degree grid, traditional ethnic homelands and perturbed ethnic homelands (which retain the centroid of traditional homelands but have altered borders). Across multiple specifications and 21 years of nighttime lights data, I find inequality is highest at the traditional ethnic homeland level. Subsequent analyses show that inequality in precipitation levels and the caloric suitability of land are also highest at the ethnic homeland level, suggesting that ethnic inequality reflects in part systematic differences in the quality of agricultural land across the traditional homelands of Han and non-Han ethnicities.