Experimental Design
This experiment uses a randomized encouragement design, which was implemented to address the issue of noncompliance with treatment assignment that often arises in randomized controlled trials. This approach is also well known as a basis for instrumental variable (IV) estimation.
-Instrumental Variable (IV): Randomized provision of two packages of entomophagous insects in 2022, along with standardized instructions on how to use them.
- Treatment Variable: Actual number of entomophagous insect packages used by each farmer during the 2022 season (0, 1, 2, …).
- Outcomes: Subsequent adoption of entomophagous insects, use of chemical pesticides, crop yields, crop types, and farm revenues.
In 2022, the sampling frame was constructed from a list of 1,000 cotton farmers maintained by the Department of Organic Agriculture of the Ministry of Agriculture of the Kyrgyz Republic (now the Ministry of Water Resources, Agriculture and Processing Industry). These farmers had never purchased entomophagous insects from the Ministry, which at that time was the sole authorized seller.
We relied on this list because, for instrumental variable (IV) estimation, the relevance condition requires that encouragement strongly affects subsequent use of entomophagous insects. By assuming that farmers who had never purchased were also those with no prior experience using these insects, the list provided an appropriate basis for the encouragement design. From this list, 300 farmers were randomly sampled.
Random assignment was conducted in Excel using computer-generated assignments, and out of the 300 farmers originally sampled, 4 did not participate, leaving 296 cotton-farming households that ultimately took part in the experiment. The intervention itself—the randomized provision of entomophagous insects with usage instructions—was implemented by the Ministry of Agriculture of the Kyrgyz Republic (now the Ministry of Water Resources, Agriculture and Processing Industry) in 2022. In addition, the Ministry also conducted two follow-up surveys: the first between April 1 and May 15, 2023, and the second between July 10 and July 21, 2023.
We are therefore registering this study in the AEA RCT Registry in order to document and conduct independent follow-up surveys on the medium- and longer-term impacts of the government-led intervention. Our team will conduct the independent recall-based follow-up survey between October 1 and November 30, 2025. Because the intervention and earlier follow-ups were government-led, this trial is registered retrospectively.