The paper presents a dynamic model for the analysis of the National Agricultural Research Systems (NARS) strategy. In a context of increasing globalisation, both intersectoral and international technology spill-ins may greatly affect the NARS design, and this paper proposes an analytical framework for its study when major spill-ins are present. These spill-ins are calculated by applying the Yale Technology Concordance (YTC) methodology. A VAR/VEC model is then specified to detect empirically, given the calculated spill-ins, what NARS design prevails in a dynamic framework. The model is estimated for the Italian agriculture data and provides evidence of close dependency on external R&D sources. * I wish to thank Renato Balducci, Pierpaolo Pierani, Franco Sotte and Alessandro Sterlacchini for their useful suggestions.