Power upward, pollution downward: how organizational centralization improves air quality in Chinese communities

The structural design of government organizations is a critical institutional factor shaping the effectiveness of public governance. This study leverages high-resolution 1-km grid air pollution data spanning over 200,000 communities—including urban cores, urban-rural fringes, town centers, town-rural fringes, and special-designated areas as classified by National Bureau of Statistics—to examine the environmental consequences of bureaucratic reorganization. Exploiting China’s 2016 vertical management reform of environmental agencies as a quasi-natural experiment, we apply difference-in-differences models complemented by an instrumental variable strategy based on the frequency of warfare during the Song Dynasty. Results indicate that organizational centralization significantly reduced air pollution levels at the community scale. Decomposition analysis reveals that these pollution reductions did not stem from industrial shutdowns or output contractions, as proxied by nighttime light intensity, but instead point to broad-based improvements in regulatory compliance. Policy comparison further demonstrates that the pollution abatement effect of centralization exceeded that of a major supply-side energy intervention—the “coal-to-gas” switching policy—by approximately 72%. Mechanism analysis shows that organizational centralization was particularly effective in strengthening environmental enforcement in areas characterized by severe elite capture. Economic magnitude estimates indicate that centralization generated aggregate economic benefits totaling $19.94 billion, averaging $1.994 billion annually—a finding that remains robust across a range of social discount rates. This study contributes to the broader literature on institutional design by demonstrating that reallocating regulatory authority upward within governmental hierarchies can yield substantial and measurable improvements in local environmental quality.