This article contributes to the literature on the effects of universities on local innovation ecosystems. In particular, it examines the extent to which technology transfer policies implemented by universities impact the cognitive proximity between academic and non-academic agents at local level. To do so, we consider both academic and non-academic patents data for a panel of Italian Provinces (NUTS3) in the timespan 1999–2009. By means of fractional response, we find a positive effect, which is stronger in cases where the initial level of proximity is low. However, results suggest that the effect of local technology transfer policies does not depend on the level of research performance of the university, except for top-tier universities, for which the impact is relatively higher. Finally, as a further robustness check, a zero-inflated beta model and a recursive fractional probit approach are employed in order to take into account the potential existence of different data generating processes and the potential presence of an endogeneity problem between the cognitive proximity index and the adoption of local technology transfer policies.