Experimental Design
Our experiment involves several steps, including recruiting relevant state partners and local soil and water conservation districts and co-designing the features of the experiment (especially with respect to recruitment and selection processes).
The initial recruitment phase began in January 2024, with partner organizations in each state publicizing the program and the application process. We piloted the application and selection process for two states (North Dakota and Virginia) in February 2024, with additional rounds of application and selection for all states to follow. As part of their applications, farm producers provide basic data about their farm and the number of acres they would like to enroll in the program. In addition, applicants must specify which practices they plan to adopt, for which crops and acres. (Our program excludes producers from receiving multiple sources of payments for implementing climate-smart or conservation practices on the same acreage.) Applicants must also agree to participate in a detailed data collection process (using either the COMET Farm or COMET planner software) to help researchers quantify the greenhouse gas emissions associated with the farm and how other production practices may have changed after adopting CSAPs.
To evaluate treatment effects of participation in the program, the total amount of funding requested across the pool of applicants must exceed the total amount of funding available so that we are able to select applicants into treatment and control groups. Equity in participation in USDA programs has been an important issue historically, and in this project we allocate more funding per acre to limited resource (LR) producers and historically underserved (HU) producers. (Both groups are defined by USDA.) We have also set targets for the share of payments that must go to LR and HU producers and the share of farmer participants who are LR or HU producers. Hence, for each round of applications we will generate a funding target based on the amount of total funding available (split equally among four states), the planned number of rounds of applications in each state, and the equity targets. Before selection takes place for each round, we will estimate the number of producers who will need to be selected to reach that funding target. A random sampling procedure will be used to sort applications, and treated producers will be those who appear first among the randomly sorted list. Depending on the number of applicants and funding amount requested, limited resource producers and historically underserved producers may be selected at a greater frequency than non-underserved producers to meet equity targets.
In addition, we are working with a survey firm to obtain information about farms that have not applied for funds through our project to serve as a secondary control group.