Ivory Power? University regulations, knowledge transfers, and firm innovation

This paper uncovers the short-term effect of academic knowledge transfers on a firm’s innovation inputs and outputs. Not only do academic knowledge transfers require the intent of the university to send knowledge, they also require the intent of the firm to seek that knowledge in the first place. This makes transfers a highly endogenous variable shaped by a firm’s strategic choices. This selection into treatment is governed by a firm’s latent innovation ability biasing my regression estimates if I fail to fully control for those confounders. I use two ways to combat this confounding. First, I use panel data with two-way fixed effects for time and firm and control for a comprehensive set of start-of-period firm-level variables that covers most factors that shape firm choices but fail to cover for temporal changes in latent innovation abilities that arise among other things due to changes in the firm’s staff. Second, I use a shift-share instrument that exploits quasi-random variation from changes to universities’ laws and regulations of intellectual property rights. I show that these university-pushed supply shocks of academic knowledge were beyond the control of firms and hence act as an exogenous shock to them via their idiosyncratic exposures. I use Swiss firm-level survey data that spans from 1999 to 2018 and covers 4293 firms. I augment this data with university-level administrative data to construct the shocks, the instrument, and to conduct falsification tests on random shock assignment. The results show that academic knowledge transfers raise a firm’s innovation input and output within the next 3 years where the effect on the former is larger. On the input side, transfers raise the propensity to spend on internal R&D by 14 percentage points (pp) and the internal R&D expenses by 16%. On the output side, they raise the propensity to develop a novel or improved product by 8 pp and the sales from novel or improved products by 7%. While the extensive margin effects are statistically significant at a 5 ‰ level, I fail to reject the absence of an effect for the intensive margin even at a 19% level. Overall, this study shows that academic knowledge is not only relevant via untargeted spillovers but also highly useful once targeted even in the short run.