A review of Cunningham, J. A., Lehmann, E. E., Menter, M., & Seitz, N. (2019). The impact of university focused technology transfer policies on regional innovation and entrepreneurship. The Journal of Technology Transfer, 44(5), 1451-1475.
I. Research question and its importance
This paper investigates whether regional entrepreneurial and innovative outcomes have been affected by a legislative change in intellectual property rights of inventions made by scientists, in the empirical context of German universities. There are three reasons the topic is important. First, it contributes to an ongoing discussion concerning the efficacy of replicating successful university focused technology transfer policy initiatives in different contexts. Second, as it’s always the query of policy maker whether or not one policy works in its intended direction, this paper provides an approach of impact evaluation for innovation policy. Finally, prior studies have not focused on examining the simultaneous effects on entrepreneurial and innovative outcomes of university focused technology transfer policies. In other words, the authors attempt to document a dynamic relationship between the policy and its effect.
II. Method, finding, and limitation
The authors collect a unique panel data set of 75 public universities in 62 regions in Germany within a 15-year time period between 1998 and 2012. Their identification strategy is a far-reaching legislation change in Germany, reforming the old “professor’s privilege” which substantially shifted property rights to universities and constituted the end of the Hochschullehrerprivileg that academic employees held all rights of their inventions and research outcomes themselves. Subsequent to the legislative change, an initial positive effect on universities as measured by start-ups and patents was observed. However, the effect changed over time, leading to some unintended consequences such as plateauing and declining periods as measured by patent rate.
III. Future research
Although the authors document a dynamic trend, they fail in making causal claim of their finding. Precisely speaking, their choice of specification was static model (i.e., two points in time), which inherently prohibits a causal claim over longer time horizon. Moreover, their findings were based on observations on different points of time. Do they control for serial correlation amongst these observations? For example, the patent rate in t1 might have correlation with the rate in t2 but the authors unfortunately ignore this, leaving their results less robust. Meanwhile, as they acknowledge in the final section, a similar trend has occurred with business registrations along their observation. It’s therefore unclear to what extent their finding was not spurious correlation unless they rule out all possible alternative explanations. To address this issue, it’s preferable to find a counterfactual (e.g., neighbour countries that share similar characteristics with the treated but was not affected by the new policy) and assess the average treatment effect (ATE) of this German legislative change. Finally, it’d be even better is that they deploy a dynamic model (e.g., refer to marketing literature) to assess the impact over time by considering serial correlation, hence can arrive at a more robust conclusion.