Explaining olive growers’ participation in producer organisations: insights from a machine learning technique

Abstarct Although Italy is among the leading global producers of olives and olive oil, the sector remains structurally fragmented, posing challenges for farmers in terms of scaling production, improving bargaining power, and managing market risks. To address these issues, the Common Agricultural Policy (CAP) has promoted the formation of Producer Organisations (POs) to enhance competitiveness. Despite institutional incentives, however, many Italian olive growers still refrain from joining POs. This study identifies the key characteristics that distinguish PO-affiliated olive growers from non-members in Italy. Drawing on both theoretical and empirical literature on farmer aggregation and employing machine learning (ML) techniques, the research offers a data-driven approach to complement traditional theory-based models. While many findings align with existing studies, the analysis also reveals new variables, particularly those related to asset specificity, that influence the likelihood of PO participation. The study presents both policy and methodological implications. First, the results can help policymakers and PO managers better target specific groups of farmers who may require additional support to achieve CAP objectives. Second, it demonstrates the value of ML-based variable selection and preprocessing in agricultural economics, offering a replicable method for identifying relevant factors in complex, high-variability sectors.