Evaluating policy impact under sparse and staggered adoption. A synthetic difference-in-differences application to EU rural development measures

This paper investigates how farmers respond to targeted policy measures under the EU Common Agricultural Policy (CAP), focusing on whether voluntary adoption, often driven by private motivations, also leads to outcomes of societal interest. To address the methodological challenges posed by staggered and sparse treatment adoption, the study employs a Synthetic Difference-in-Differences (SDID) approach. A theoretical framework is developed to distinguish between private and societal outcomes of policy adoption. The empirical analysis uses a balanced panel of Italian farms from 2014 to 2022 and focuses on selected second-pillar CAP measures. Results reveal that while some measures significantly affect private outcomes (e.g., farm income or productivity), their impact on societal outcomes (e.g., environmental indicators) is weaker and more volatile. The paper discusses key challenges in identifying and estimating effects with few, heterogeneous treated units and staggered policy uptake. While SDID is well-suited to such contexts, its real-world application may face practical limitations. • This paper explores the application of Causal Inference (CI) logic to the evaluation of policies with variable granularity and adoption sparsity. • An appropriate theoretical framework is elaborated to model agents’ behavioural response to the policy then generating the outcome of societal interest. • To account for staggered policy adoption, a Synthetic Difference-in-Differences approach is adopted. • As an exemplary case the response of farmers to some selected 2014-2022 EU RDP measures is investigated by using a balanced sample of Italian farms. • Results highlight advantages and limits of SDID estimation, provide evidence on the limited effectiveness and efficiency of these policies and point to possible improvements.