Voluntary programs to encourage refuges for pesticide resistance management: lessons from a quasi-experiment

Economists often treat pesticide resistance as a common-pool resource problem. While pecuniary economic incentives are the standard prescription for open-access market failures arising with such resources, non-pecuniary behavioral approaches are also effective in some cases. Yet non-pecuniary instruments have not previously been evaluated for managing pesticide resistance. I empirically evaluate the performance of such an intervention to manage pest resistance to genetically engineered Bacillus thuringiensis (Bt) corn. Regulations by the US Environmental Protection Agency mandate refuges to delay the evolution of Bt resistance. To satisfy Bt product permitting requirements with the EPA, the agricultural company Monsanto piloted a social marketing program to promote refuge in 17 North Carolina counties in 2013-2014. Using seed sales data from 2013 to 2016 for the whole of North Carolina, I estimate multiple econometric models, including linear difference-in-differences (DID), fractional DID, discrete changes-in-changes, and matching to identify the average treatment effect (ATE) of the program on growers’ refuge planting decisions. Results suggest that if it had covered all corn growers in North Carolina, the intervention would have led the average grower to plant between 2.6% and 5.8% (depending on model) more refuge in the 2014 season immediately following the program, than would have been the case without the program. The ATE roughly halves in 2015 and vanishes by 2016. Econometric analysis suggests the program increased by 12% the average probability of planting any refuge in 2014. Evidence is mixed for effects of the intervention on grower compliance with mandated refuge thresholds. Informed by behavioral economics research on other environmental and resource policies, I discuss the implications of these findings for pesticide resistance management.