Deforestation, contributing to around 10% of global GHG emissions, is a focal issue at U.N.COP15. In 2023, the European Union established anti-deforestation regulations, effective by 2024, mandating companies to provide due diligence statements concerning supply chain traceability to prevent deforestation. Indonesia, a primary producer of coffee and cocoa, argues that this law unfairly burdens smallholders, emphasizing the significant effort and time required to transition to sustainable technology. In Lampung province, the agroforestry system emerges as a pivotal strategy to conserve endemic tree species and sustain commercially viable crops, with agricultural training acknowledged for enhancing smallholder farmers’ skills (FAO, 2018). Conventionally, local training by invited experts is the norm; however, these settings restrict farmers' exposure to practical technology applications and anticipated benefits of improved practices. The risks inherent in new agricultural technologies can deter adoption; yet, firsthand witnessing of potential benefits may encourage farmers to take such risks for anticipated future gains. Understanding effective information dissemination methods among farmers and leveraging local networks for more effective knowledge transmission is vital. Nevertheless, previous studies delving into social networks and desired outcomes often encountered endogeneity issues, arising from unobservable characteristics correlated with both outcomes and network variables, possibly misestimating network effects. Overcoming these limitations may be feasible through a randomized-controlled trial (RCT) experiment, which assigns randomly selected participants in a treatment group to interact with other randomly selected farmers –making network assignments exogenous.

This paper explores an experiment aligned with this premise, combining the RCT method with locational effects to assess both network outcomes and the promotion of agroforestry practices among coffee and cocoa farmers in Lampung. Randomly selected farmers will be allocated to training program held in three distinct, more developed and agriculturally-advanced locations. The research questions are: (1) Does the training alter the structure of local information networks, and (2) Does the training influence the uptake of agroforestry and conservation practices? It is hypothesized that through peer influence, training will catalyze agroforestry adoption among farmers. Additionally, it is anticipated that farmers trained in more advanced locations will exhibit superior performance on agroforestry indicators, having been exposed to best practices in those regions.