with F. Poege, K. Hoisl, D. Harhoff and D. Dorner
We examine how collaborator loss affects knowledge workers in corporate R&D. We argue that such a loss affects the remaining collaborators not only by reducing their team-specific capital (as argued in the prior literature) but also by increasing their bargaining power over the employer, who is in need of filling the gap left by the lost collaborator to ensure the continuation of R&D projects. This shift in bargaining power may, in turn, lead to benefits, such as additional resources or more attractive working conditions. These benefits can partially compensate for the negative effect of reduced team-specific capital on productivity and influence the career trajectories of the remaining collaborators. We empirically investigate the consequences of collaborator loss by exploiting 845 unexpected deaths of active inventors. We find that inventor death has a moderate negative effect on the productivity of the remaining collaborators. This negative effect disappears when we focus on the remaining collaborators who work for the same employer as the deceased inventor. Moreover, this group is more likely to be promoted and less likely to leave their current employer.
with S. Baruffaldi
We establish the importance of physical capital in knowledge production. Exploiting adverse events in research laboratories, we find that scientists experience a persistent reduction in research output if they lose specialized physical capital---equipment and material they created over time for a particular research purpose. In contrast, they recover in productivity if they only lose generic physical capital. Scientists in older laboratories, who presumably lose more obsolete physical capital, are more likely to change their research direction and recover. These findings suggest that scientists' investments into their own physical capital yield lasting returns but also create path dependence.
with D. Byrski, and M. J. Higgins
Prior literature has established a link between changes in market size and pharmaceutical innovation; whether a link exists with scientific research remains an open question. If upstream research is not responsive to these changes, the kinds of scientific discoveries that flow into future drug development could be disconnected from downstream demand. We explore this question by exploiting the effects of quasi-experimental variation in market size introduced by Medicare Part D. We find no causal relationship between market size and biomedical research in the decade following the implementation of Medicare Part D. While many factors have been shown to motivate scientists to conduct research, this result suggests that changes in market size provide no such incentive. We do find, however, limited support for a response by corporate scientists conducting applied research. Implications for pharmaceutical innovation policy are discussed.
with A. Danzer and C. Feuerbaum
While economic theory suggests substitutability between labor and capital, little evidence exists regarding the causal effect of labor supply on inventing labor-saving technologies. We analyze the impact of exogenous changes in regional labor supply on automation innovation by exploiting an immigrant placement policy in Germany during the 1990s and 2000s. We find that an increase in the low-skilled workforce reduces automation innovation. The effect is strongest in regions with a low unemployment rate. Moreover, the effect is driven by the response of large firms in manufacturing industries, suggesting that internal demand for automation innovation is an important mechanism.
with D. Harhoff, S. Sorg and G. Von Graevenitz
We study the blocking effect of patents for follow-on innovation. We argue that follow-on innovation requires freedom to operate (FTO), which firms usually obtain through licensing. Where licensing fails, follow-on innovation is blocked unless firms gain FTO through patent invalidation. Using large-scale data from post-grant oppositions at the European Patent Office, we find that patent invalidation causes an increase in follow-on innovation by 16% on average. This effect is the strongest for patents of low and very high value in our sample. Invalidation of low-value patents predominantly increases marginal follow-on innovation. Here, transaction costs likely exceeded the joint surplus of licensing, causing bargaining failure. Invalidation of high-value patents also increases follow-on innovation, especially in the patentees' own industries and technology fields. We explain this result with rent dissipation effects, which make patentees less willing to license out valuable technologies to (potential) competitors.
Where AI is a Game Changer – Evidence from Chess Computerswith H. Piezunka
AI has been conceptualized as either displacing or complementing humans in task fulfillment. We propose a third role: helping decision-makers learn. We suggest that AI-enabled simulations can train decision-makers in activities where conventional training methods have fallen short. Specifically, we suggest that they help decision-makers learn strategic interactions by providing intelligent responses at scale. We present evidence from chess computers, the first widespread incarnation of AI. Chess computers diffused in Western countries after 1977 but became available in the (former) Soviet Union only after the fall of the Iron Curtain. We illustrate a strategically important heterogeneous treatment effect: actors at a competitive disadvantage benefit particularly from AI-enabled simulations. We discuss implications for research on AI in management and strategy, on learning, and on competitive strategy.
with Y. Lefouili
This paper examines court selection by plaintiffs in patent litigation. We build a forum shopping model that provides a set of predictions regarding plaintiffs’ court preferences, and the way these preferences depend on the market proximity between the plaintiff and the defendant. Then, using a rich dataset of patent litigation at German regional courts between 2003 and 2008, we estimate the determinants of court selection with alternative-specific conditional logit models. In line with our theoretical predictions, our empirical results show that plaintiffs prefer courts that have shorter proceedings, especially when they compete against the defendants they face. Further, we find negative effects of the plaintiff’s, as well as the defendant’s, distance to court on the plaintiff’s court selection. Our empirical analysis also allows us to infer whether plaintiffs perceive a given court as more or less pro-patentee than another one.