We examine the determinants of innovation cooperation in general, and with science in particular. First, the focus was on innovative firms: the ones that may decide to cooperate in innovation. Second, the cooperation with science was only defined for firms which decide to cooperate. A trivariate model with double sample selection was estimated to address the potential bias due to self-selected subsamples. Innovative firms that cooperate with science had a better in-house capability, were larger, and received more R&D support than innovators that cooperate with other partners. Furthermore, their probability of cooperating with science increased with the level of protection.

This paper evaluates the effectiveness of public subsidies in fostering R&D activities of Italian firms using 2008 CIS data. Given the observational nature of data, the endogeneity of the subsidies is taken into account. To estimate the effect of subsidies on firms' R&D activities, the propensity score matching is applied, controlling for determinants of public support and innovation activities. Furthermore, the heterogeneity of effects is investigated using a subgroup analysis. The underlying rationale is that the R&D funding impact is more realistic if its effect is measured on homogeneous groups, instead of measuring an average effect on a heterogeneous group of firms. The results from various propensity score matching lead to an increment of the R&D intensity of about 1.5 percentage points, and such a result, closely matches earlier empirical findings for Italy. In contrast, the subgroup analysis suggests that this is not a widespread finding; but, isolated to some subgroups.