Conference Agenda

What role do incentives play in the performance of bureaucrats? This paper estimates the effect of promotion incentives for bureaucrats in reducing water pollution in the Yangtze river in China, using a regression discontinuity design. The RD approach exploits the age threshold implicit in China’s bureaucratic promotion system where promotion prospects dwindle sharply after the age of 56.5 years. Since 2017, environmental performance was explicitly added as a criteria in promotion decisions of mayors and party secretaries. We find that polluting firms in cities governed by younger bureaucrats emit 3% less pollution than those governed by bureaucrats older than 56.5 years. This young–old pollution gap exists only in polluting industries, not in nonpolluting ones, and it emerges only after 2017, when the policy was implemented. Using firm level pollution data, we find that younger bureaucrats with stronger promotion incentives tighten water regulation by increasing inspections of polluting firms and promoting the entry of cleaner firms. Linking promotion incentives with housing values, a back-of-the-envelope calculation suggests that strong bureaucratic incentives related to protecting the Yangtze River are associated with an economic benefit of approximately 3.9 trillion CNY (550 billion USD).

Firms often face strong incentives to evade environmental regulation, particularly when compliance costs are high and enforcement relies on on-site inspections. This study investigates how rainfall facilitates strategic pollution behavior. Leveraging daily variation in precipitation across Chinese industrial firms equipped with Continuous Emissions Monitoring Systems (CEMS), we identify the causal impact of rainfall on firms’ water pollutant discharges. We find that precipitation significantly increases pollution emissions: a 100 mm rise in daily rainfall leads to a 2% increase in chemical oxygen demand (COD) and a 3.5% increase in ammonia nitrogen (NH₄–N), alongside a 23.9% rise in wastewater discharge volume. These effects are most pronounced during heavy rainfall and are concentrated among manufacturing firms subject to stricter regulation. Firms strategically time discharges based on weather forecasts, exploiting enforcement frictions and natural dilution effects. Citizen monitoring and stronger local enforcement substantially mitigate this behavior, whereas central inspections offer limited deterrence without persistent local incentives. By disentangling firm-level emissions from rainfall-induced non-point pollution, our study provides novel evidence on firms’ exploitation of natural conditions to evade environmental regulation.

This paper provides novel and systematic evidence on the impact of water collection times on women’s education. I employ a robust methodological framework combining panel two-way fixed effects (FE) and an instrumental variable (IV) approach. I construct a novel way to instrument water collection time at the local level by exploiting variation in climate change-induced drought intensification. Using harmonized DHS panel data covering 439 subnational regions across Africa, Latin America, Eastern Europe, and Asia from 1990 to 2019, the analysis shows that longer water collection times have a negative and quantitatively meaningful effect on women’s educational attainment. A battery of sensitivity tests demonstrates that these results are robust to alternative specifications and additional controls. Ongoing work is exploring potential mechanisms and further robustness checks, which we aim to include before the Conference.

Environmental monitoring is essential for detecting pollution and enforcing regulations, yet systematic evidence on monitoring equity remains limited. We examine whether water quality monitors in the United States are strategically placed to avoid minority communities. Comparing regulatory monitors (placed by state agencies) against randomly-placed monitors from the EPA's National Rivers and Streams Assessment, we find that strategic monitors are 4 to 5 percentage points less likely to appear in census tracts with above-median shares of Hispanic, Black, or Native American residents. Random monitors show no such pattern. This disparity persists after controlling for stream availability, geography, and socioeconomic factors, and translates to lower monitoring intensity in disadvantaged areas. Our findings are difficult to reconcile with stream geography and residential sorting alone and are consistent with regulatory siting decisions. Because water quality monitoring is the primary mechanism for identifying impaired waters and triggering enforcement under the Clean Water Act, these disparities imply systematic underdetection of pollution in minority communities and inefficient allocation of regulatory resources.