Veranstaltungen

Startups are often creators of important inventions, yet they lack resources of their own and must prioritize their commercialization effort into a few promising technologies. We explore how corporate venture capitalists (CVCs) influence a startup’s likelihood of selling its patents. We find that, compared with startups backed solely by independent venture capitalists (IVCs), those backed by corporate venture capitalists (CVCs) are less likely to sell their patents. We further consider various factors, including complementary assets provided by corporate investors, competitive dynamics between corporate investors and startups, as well as other patent-, investor-, and startup-specific characteristics that further corroborate the theoretical mechanisms on how corporate investors influence a startup’s likelihood of selling its patents. Our results provide novel insights into how the resources and preferences of corporate investors can play important roles in a startup’s intellectual property (IP) management strategy and its participation in the market for technology.

Ansprechpartner: Daehyun Kim

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

While the state of technology at country level has been compared by using patent information, the difference of innovation process, how the inventions are converted into actual products and services, are under investigated. Historically, the relationship between technology and the market has been analyzed by using technology-industry concordance matrix, but the granularity of market information is fined by industrial classification system. In this study, we use both patent and web contents information to estimate key word level detail innovation conversion model, and compare those across three countries, China, Japan and the United States. First, we apply dual attention model to extract product/service information out of web page information (Motohashi and Zhu, 2023). Then, using the textual information of both patent abstracts and product/service keywords, we develop the machine learning model to predict product/service from some particular type of technology. Finally, we compare the actual product/service vector and predicted product/service vector by cosine similarity to see the innovation transformation process is different by countries.

Ansprechpartner: Daehyun Kim

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

When firms engage in lobbying, the primary outcome they seek is beneficial regulatory change. However, prior literature suggests that there may also be a secondary outcome to lobbying—the leakage of knowledge to competitors. In this paper, I explore the leakage of technological knowledge when firms lobby for technology-related regulations and the strategic response of competitors to the leakage. Based on field interviews and existing studies, I hypothesize that when a firm is involved in lobbying, competitors are able to learn about the lobbying firm’s technologies through the lobbyists and policies for which the firm advocates. In addition, I abductively explore whether leakage is more likely with external or internal lobbyists. I build a unique dataset on U.S. firms that engaged in lobbying on technology policy and the patents applied for by those firms. I utilize an embedding approach on the text of patents to directly detect knowledge leakage in the strategic response of competitors. The results show that when a firm increase lobbying efforts, its patents are associated with a greater number of imitative patents generated by competitors. Moreover, post-hoc analyses suggest that internal lobbying is closely linked to leakage because firm employees posses and share detailed information, but there is no evidence that external lobbyists opportunistically leak. Overall, this paper documents a technological risk inherent to coordinating a firm’s innovation activity with its political strategy.

Ansprechpartner: Daehyun Kim

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

Although researchers in Pasteur’s Quadrant are thought to be more productive, we lack comprehensive data on who they are, where they work, and how they impact both science and technology. We provide a new and publicly-available dataset that links the careers of scientists and U.S. inventors, providing large sample evidence that those working in “Pasteur’s Quadrant” produce more novel and highly cited science and technology. We describe the crosswalk of these individuals’ fields of science and classes of technology, institutional and organizational affiliations, consistently higher geographic concentration than pure scientists or pure inventors, and shift towards wealthy and western counties within the U.S..

Ansprechpartnerin: Marina Chugunova

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

GenAI Acceptance in Professional Services: The Case of Management Consulting
Presenter: Dennis Nesemeier (TUM)
Discussant: Mainak Ghosh (MPI)

The increasing relevance of Generative Artificial Intelligence (GenAI) in professional services and its impact on client services and operations raises the question: “What factors influence the acceptance of GenAI?” I explore the intricate factors influencing the acceptance of GenAI, specifically focusing on management consulting. Therefore, I use the Unified Theory of Acceptance and Use of Technology (UTAUT) as a theoretical framework, which I adapt and extend while employing a mixed-methods approach. Twenty semi-structured interviews and a quantitative survey of 121 consultants reveal insights into consultants’ perceptions and interactions with GenAI. The findings indicate the relevance of performance and effort expectations, social influence, facilitating conditions, and concerns about trustworthiness. Highlighting the complexity of human-technology dynamics, some consultants view GenAI as an opportunity to gain a competitive advantage in their career progression, while others report feeling ashamed when disclosing their use of it. The study broadens the scope of technology acceptance research, introduces specific adaptations of the theory to fit the GenAI context better, and provides practical managerial recommendations.

How to Measure and Improve the Quality of Crowd-Sourced Data Annotation?
Presenter: Chiara Belletti (ISTO)
Discussant: Leonard Hanschur (TUM)

Micro-tasking platforms enable the collection of data used to train machine learning algorithms and artificial intelligence. However, a classical Principal-Agent problem may limit the quality of the data produced by micro-taskers as firms do not always monitor the quality of the work done with sufficient frequency. We develop a structural model of equilibrium demand and supply of effort to measure quality and monitoring behavior. We estimate the parameters of this model using proprietary data from a leading micro-tasking platform. We find that metrics relying on observed task rejection severely underestimate the quality/effort with which data annotation, collection generation tasks are performed, exposing AI applications to noise and bias. We discuss several mitigation strategies. We find that increasing the pay of micro-taskers along with more frequent monitoring could help improve the quality of the data. Finally, we discuss incentive schemes to induce higher quality work by relying on counter-factual simulations. We show that charging penalties for workers with a rejected task could induce higher effort and require less monitoring from the firms.

Breaking the Ice: Can Initially Active Peers Improve Platform Engagement and Persistence?
Presenter: Svenja Friess (MPI)
Discussant: Ambre Nicolle (ISTO)

Online knowledge exchange has flourished in recent years, yet struggles with low user engagement remain. This study investigates the role of early peer interactions in sustaining engagement on digital platforms. Analyzing novel data from 12,000+ professionals upskilling in an online business program, we exploit quasi-random variations in initial peer activity levels per cohort to estimate their impact on future engagement and platform persistence. Results reveal that a high initial share of active peers giving likes reduces platform persistence by 3%, while a high share of early active commenting peers exhibits no correlation with future engagement or persistence. However, when looking at directed interactions, we find that receiving early comments and likes significantly boosts future engagement and platform persistence. Employing cutting-edge Natural Language Processing techniques, we classify comment characteristics to shed light on underlying mechanisms. These results indicate that receiving early positive comments has a stronger positive association with future engagement and persistence than when early comments are more negative. Our findings provide insights for digital platform designers to effectively leverage early and directed peer interactions, enhancing user experience and platform value.

Commercialization is a crucial phase in the innovation process and its associated costs significantly influence R&D decisions. Yet our understanding of how commercialization costs impact various stages of innovation remains underdeveloped. In this study, I investigate the effects of commercialization costs on early and late stages of the innovation process in a quasi-experimental setting. Specifically, I leverage sudden policy shifts in the US Food and Drug Administration (FDA) that increased commercialization costs for drugs in certain therapeutic areas. Employing a difference-in-differences methodology, I trace the impacts of these elevated costs on discovery and clinical trial advancement of 3,357 drug candidates between 1997 and 2015. My research places emphasis on the contrasting roles that startups and established firms have in innovation. My findings reveal that while commercialization costs diminish the late-stage efforts in commercializing innovations, especially by established firms, they stimulate an environment conducive to early-stage entrepreneurial drug discovery efforts. Furthermore, I find that the disruption that commercialization costs create in markets for technology drives these opposite findings: new discoveries remain without buyers in technology markets, failing to complete their development process as commercialized products.

Ansprechperson: Elisabeth Hofmeister

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

Based on comparative research on recent AI regulation and Law between China and EU, especially “The Interim Administrative Measures on Generative AI Services” in China and the “EU Artificial Intelligence Act”, also based on recent AI case studies related to intellectual property, and ethics issues (including fairness and safety). This paper put forward some suggestions about restrictions and supplemental regulations etc. in the related laws (including patent law, copyright law). And gives considerations for promoting the sustainable development of AI industry by promoting ethical and sustainable innovation.

Prof. Dr. Xiang YU is a Member of Academia Europaea, Director of Sino-European Institute for Intellectual Property of Huazhong University of Science and Technology, and Vice President of Hubei Normal University.

Ansprechpartner: Daehyun Kim

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

We study the welfare implications of the rise of common ownership in the United States from 1995 to 2021. We build a general equilibrium model with a hedonic demand system in which firms compete in a network game of oligopoly. Firms are connected through two large networks: the first reflects ownership overlap, the second product market rivalry. In our model, common ownership of competing firms induces unilateral incentives to soften competition and the magnitude of the common ownership effect depends on how much the two networks overlap. We estimate our model for the universe of U.S. public corporations using a combination of firm financials, investor holdings, and text-based product similarity data. We perform counterfactual calculations to evaluate how the efficiency and the distributional impact of common ownership have evolved over time. According to our estimates the welfare cost of common ownership, measured as the ratio of deadweight loss to total surplus, has increased about ninefold between 1995 and 2021. Under various corporate governance models the deadweight loss of common ownership ranges between 3.5% and 13.2% of total surplus in 2021. The rise of common ownership has also resulted in a significant reallocation of surplus from consumers to producers.

Ansprechpartnerin: Marina Chugunova

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

We examine how reviewer–applicant social ties (department and university affiliation, co-author/co-applicant relationships, research field similarity) influence reviewer evaluations, based on Japanese research grant administrative data (2005–2016). All relationships between social ties and scores are positively correlated, even after accounting for unobservable applicant characteristics and proposal quality. Regarding bias and information advantage effects, upward deviation from department match negatively correlates with applicants’ future research outputs, implying bias. Upward deviation from research field similarity or university match positively correlates with future productivity, indicating that information advantage predicts applicants’ future productivity. 

Ansprechpartnerin: Marina Chugunova

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.

The international mobility of researchers has been central to the agenda of policy-makers for several decades. Despite the growing presence of mobility grants within public funding agencies' portfolios, empirical evidence on their effects remains scant. In this paper, we contribute to the literature by studying the Marie Curie fellowships, the flagship program of the EU, providing competitive grants to early-stage researchers to spend a research period abroad. Based on data for the universe of applicants to the Seventh Framework Programme (2007-2013), we exploit the discontinuity in grant assignment to uncover causal effects on individual researchers. Results show that grants are indeed conducive to higher chances of experiencing mobility towards the scientists' country of choice. Conversely, we do not find systematic evidence that grants on average lead to increases in publication quantity or quality, nor improved career progression. Finally, we document interesting heterogeneous effects: grants supporting extra-European mobility, as opposed to those supporting mobility within Europe, generally yield more positive effects across most outcomes. This suggests that grants are most effective when targeting mobility flows subject to larger frictions. (co-authors: Stefano Baruffaldi, Yevgeniya Shevtsova)

Ansprechpartner: David Heller

Eintragung in den Einladungsverteiler und mehr Informationen auf der Seminarseite.