Bureaucrats are often incentivized to manipulate information, which may have real consequences. Leveraging China's 2009 reform punishing economic data manipulation and counties' quasi-random reform exposure, I provide rich causal evidence showing that the reform led to: (1) a decrease in GDP growth manipulation amounting to 5% of reported GDP growth, driven by a reputational discipline effect; (2) an increase in local officials' development effort manifested in both policy changes and downstream impacts, consistent with an effort reallocation effect. These results thus highlight the far-reaching costs of such manipulation and the welfare implications of curbing it.

This paper analyzes modern political corruption through a cultural lens. Drawing on Confucianism that flourished in Imperial China emphasizing virtuous leadership, we show lower corruption in governments led by mayors with greater ancestral exposure to such culture. For causal identification, we leverage: (1) the irrelevance of Confucianism in leader selection and allocation; and (2) the exogeneity in Confucian academy location choices. We provide support for the virtuous leadership mechanism and rule out alternative explanations, such as changes in formal institutions, education inputs, or government revenue. We further show that the persistence of this effect stems from the intergenerational transmission of cultural norms. 

This paper tests the social contract view of the state within the realm of pollution control. Using a unique dataset on firm-level emission records from China, we find that firms with prior government subsidies achieved greater emission reductions after a national pollution control campaign, likely through output cuts and clean technology adoption. Such reductions, which we interpret as driven by reciprocity between firms and governments, were more pronounced for firms with a greater desire to reciprocate the government and paid off as higher future government subsidies. We find little support for other mechanisms, such as reduced financial constraints, political connections, forced reductions, and data manipulation. In aggregate, regions that previously granted more subsidies to firms witnessed greater reductions in pollution concentrations.

I test the role of economic incentives and command and control in reducing agricultural fires, a major source of air pollution in most rural regions across the world caused by burning crop residues after harvest. To tackle data shortage, I use high-resolution satellite data to construct a fine measure of agricultural fires as well as other geographic characteristics at 1 km × 1 km resolution for China. Using the staggered arrival of biomass power plants, which purchase crop residues as production inputs from nearby areas, as a shock to economic incentives, I find a more than 30% drop in agricultural fires in the vicinity of a plant after its opening relative to areas farther away. Such drop cannot be explained by structural transformation, migration, or enhanced regulation near the plant, and is consistent with an incentive-based explanation. I then examine the effectiveness of a command and control policy that bans agricultural fires within 15 km of airports. Using a spatial regression discontinuity design, I find no evidence that such a policy works. 

Chinese local leaders are frequently moved across prefectures. By combining local leader rotation data and comprehensive firm land parcel purchase data across prefectures from 2006 to 2016, this paper examines how firm-politician connections affect resource allocation and finds that a firm headquartered in a leader's previous work prefecture purchases three times more land parcels in that leader's new governing prefecture than the prefecture-year mean, at half the unit prices. Identification is from within-firm-year variation in various prefectures through exogenous politician rotation. Land usage efficiency is lower for these follower firms' land parcels. Land allocation distortion is also economically sizable.