Evaluating the heterogeneous effects of on-site scientist-government collaboration on Yangtze River protection and restoration using causal machine learning

China has introduced the On-site Government and Research Institution Collaboration (OGRIC) for the Yangtze River Environmental Protection and Restoration to address the gap between environmental research and the Yangtze River's remediation requirements. We used causal machine learning and panel data from 117 cities in the Yangtze River Basin from 2016 to 2019 to assess the effect of the OGRIC on water quality improvement. We found that the OGRIC was efficient in reducing point source pollutants, including chemical oxygen demand, biochemical oxygen demand, and total phosphorus, demonstrating the value of the pattern of on-site cooperation between the government and scientists in pollution control. The results revealed that cities with higher initial pollution, higher economic development, and lower infrastructure development benefitted more from the OGRIC. Policy suggestions for improving the OGRIC are presented, including strengthening the scientist-government collaboration, focusing on non-point source pollution, and providing increased financial support for areas with low development levels.