A government's choice of environmental strategy plays an important role in the coordinated governance of regional air pollution. Based on changes in China's environmental policies and on changes in environmental indicators over the years, this paper selects regional haze data from the years 2005, 2009, 2013, and 2017; uses social network analysis to describe the structural characteristics of a spatial correlation network in China; measures the level of coordination using a population gravity model; and further analyzes the influence of the overall structural characteristics of spatial networks on the level of coordination. The results show that the spatial association of regional haze presents a typical "central edge" network structure. The Beijing-Tianjin-Hebei region and the Yangtze River Delta region are the largest emitters in China. The coordination level of haze control in China showed a fluctuating upward trend, but the overall level of coordination is relatively low, and there is still great room for improvement. Based on the above characteristics, using the provincial panel data from 2005-2017, a two-zone spatial Durbin model was built to empirically test the impact of changes to the environmental performance assessment system on local coordinated haze-control decisions and their stage characteristics. The overall sample results show that there was a "race to the bottom" among Chinese provinces during the study period. When the haze control intensity in neighboring areas was relaxed, the regional governments also tended to relax their own environmental regulation intensity. The time-based analysis results further show that with the improvement of the environmental performance assessment system, the strategy selection of coordinated governmental haze-management presents the possibility of a "race to the top".