Stephen Michael Impink researches the impact of technology and innovation on
management and organizations.
Over the past four years, I have developed a body of research on entrepreneurship that serves as a base for my future research agenda, which focuses on examining the relationship between larger technology firms and high-tech startups. High-tech startups need lots of data, computational power, and computer science expertise to develop their product and scale. This relationship with larger technology firms, specifically cloud services providers, is often a conduit through which startups receive their initial digital resources. My prior work experience at Microsoft in Seattle and Singapore, where I managed emerging market monetization strategies with device manufacturers, informs my research questions and enables me to delve deeply into the mechanisms underscoring these relationships.
The faculty and students at NYU Stern have provided me with all the resources necessary to build out my research agenda, develop my identity as a scholar, and take the initial steps along this path. Throughout my PhD, I have focused on using novel data to answer my research questions. Over the last four years, I have managed a large-scale annual survey of AI-producing startups to access data to provide an initial glimpse at this nascent industry. Additionally, I partner with several large technology firms and two accelerator programs to access novel data in pursuit of this research agenda. Beyond data-sharing agreements, my research projects have received financial or technical support from the Brookings Institution, Kauffman Foundation, Microsoft Corporation, GitHub, MassChallenge, and the NYU Fubon Center, and my papers have been published in Research Policy, Nature Humanities and Social Sciences Communications, the American Economic Association Papers and Proceedings, and the Proceedings of the AAAI/ACM Conference on AI, Ethics and Society. Moreover, I have several advanced working papers, including a manuscript with Raffaella Sadun and Andrea Prat that we recently resubmitted to Management Science.
High-tech entrepreneurship
Data and processes enabling firms to organize, reconfigure and use data more efficiently and effectively are crucial to digital product development. Larger technology firms have abundant digital resources, and they engage with startups through establishing cloud services and IT supplier relationships while often competing downstream in product markets. In my job market paper, “Outsourcing IT and Technological Differentiation: Evidence from Digital Startups,” I examine how this cloud services supplier relationship can impact the breadth of innovation. Through this relationship, startups receive valuable platform-related resources that affect technology adoption, such as technical expertise, software, training data, and compatibility guidance.
Using panel data on app-developing startups, I find that startups using cloud platforms adopt larger product development technology bundles, consisting of frameworks and developer tools core to coding digital product applications. But these technology bundles become more similar to others on cloud platforms to fit with the cloud platform’s underlying technology and reduce the coordination costs associated with using a larger number of interdependent technologies. To differentiate their products, startups adopt larger data analytics technology bundles that are increasingly dissimilar from others on cloud platforms, producing more robust and unique data resources. This project enables me to construct a novel data set from many data providers, and this high-dimensional panel data allows me to use numerous quantitative research methods to address endogeneity issues and support a causal interpretation. Moreover, my findings contribute to a broader literature on innovation by explaining that firms adopt more similar bundles of technologies when constrained by the need for those technologies to fit with their suppliers’ platform and the other technologies in their bundle for their digital products to work effectively.
The rest of my dissertation focuses on the sharing relationship between larger technology firms and high-tech startups. In the second chapter of my dissertation, I use data from the AI startup survey to examine how these shared resources and a startup’s perceived fit with a larger technology firm affect its plan to scale or be acquired. In the third chapter of my dissertation, co-authored with Rob Seamans, Josh Lerner, and Nataliya Wright, I use exclusive data from a large technology firm’s corporate accelerator program to examine the impact of sharing specialized technical resources instead of the more general resources that typical accelerator programs share.
As part of this research agenda, I have several projects with Rob Seamans and Jim Bessen devoted to examining high-tech AI-producing startups, whose products are expected to spur macroeconomic growth. We will continue to run this survey annually to build a panel of survey data and have collected information from almost 1000 AI startups. We published research on using proprietary training data in AI development from different sources in Research Policy and on AI startups’ ethical product development in a top computer science outlet, Proceedings of the AAAI/ACM Conference on AI, Ethics and Society. Next, we have a Brookings Institution Working Paper examining how data-sharing relationships between technology firms and startups could lead to potential positive spillovers for product development which we are preparing to submit to a mainstream strategy journal. Third, we have a working paper exploring how increased data regulation impacts data-centric AI startups. As a follow-up to these papers, we plan to examine the governance mechanism that startups could use to enforce their ethical AI principles and develop more ethical products.
While completing my PhD, I have assembled panel data that will enable me to answer several important research questions on startups that use data as inputs in production. I have partnered with GitHub to examine which open-source technologies startups use in AI production and whether and how using specific technologies or frameworks could impact startup performance. This project extends from my dissertation work, enabling me to determine if a relationship with a larger technology firm relates to adopting open source technologies or using more proprietary solutions. Lastly, in a project using exclusive data from a different accelerator program with two other students, Nataliya Wright and Aticus Peterson, I examine how connecting startups with mentors from larger firms affects entrepreneurial growth.
Digital information flow
I work with Raffaella Sadun and Andrea Prat on a secondary research track to examine information flow within firms with novel communication meta-data at scale across many firms. We source this data through a partnership with a large multinational email provider. We discuss the value of communications metadata for modern digital firms in a paper published in the American Economic Association Paper and Proceedings. As a follow-on paper, we use a theoretical information flow model and event study design to examine how information flow changes during a CEO transition across 104 firms and map these changes to stock performance (resubmitted to Management Science). Using similar communications metadata at the city level, I published research with co-authors in Nature Social Sciences and Humanities Communications on how Covid-19 lockdowns and increased remote work impact information flow.