In this paper, we address identification concerns linked to spillover effects. The literature is still scanty on this issue. To this aim, we propose an innovative strategy combining a counterfactual approach with spatial models. Specifically, we suggest an original revision of spatial propensity score matching, extending it to handle continuous treatment and weighting the spatial lags by the products' distances over a "product-country space". Adopting this strategy, we aim to address the bias in treatment selection and effectively manage interference and spillover effects from policy interventions and from government actions in other sectors. Using standard data, we then test our model in an empirical application that assesses the causal impact of agricultural policy support on product trade performance. Results show that not considering spillover effects leads to underestimating treat-ment assessment.
Nenci, S., Vurchio, D. (2023). Modeling country-sectoral spillovers in generalized propensity score matching: An empirical test on trade data. ECONOMIC MODELLING, 124 [10.1016/j.econmod.2023.106293].