We examine the relationship between data analytics capabilities and innovation using detailed firm-level data. To measure innovation, we first utilize a survey to capture two types of firm practices, process improvement and new technology development for 331 firms. We then use patent data to further analyze new technology development for a broader sample of more than 2,000 publicly traded firms. We find that data analytics capabilities are more likely to be present and are more valuable in firms that are oriented around process improvement and that create new technologies by combining a diverse set of existing technologies than they are in firms that are focused on generating entirely new technologies. These results are consistent with the theory that data analytics are complementary to certain types of innovation because they enable firms to expand the search space of existing knowledge to combine into new technologies, as well as the theoretical arguments that data analytics support incremental process improvements. Data analytics appears less effective for developing entirely new technologies or creating combinations involving a few areas of knowledge, innovative approaches where there is either limited data or limited value in integrating diverse knowledge. Overall, our results suggest that firms that have historically focused on specific types of innovation—process innovation and innovation by diverse recombination—may receive the most benefits from using data analytics. This paper was accepted by Chris Forman, information systems.

Data-analytics technology can accelerate the innovation process by enabling existing knowledge to be identified, accessed, combined, and deployed to address new problem domains. However, like prior advances in information technology, the ability of firms to exploit these opportunities depends on a variety of complementary human capital and organizational capabilities. We focus on whether analytics is more valuable in firms where innovation within a firm has decentralized groups of inventors or centralized ones. Our analysis draws on prior work measuring firm-analytics capability using detailed employee-level data and matches these data to metrics on intrafirm inventor networks that reveal whether a firm’s innovation structure is centralized or decentralized. In a panel of 1,864 publicly traded firms from the years 1988–2013, we find that firms with a decentralized innovation structure have a greater demand for analytics skills and receive greater productivity benefits from their analytics capabilities, consistent with a complementarity between analytics and decentralized innovation. We also find that analytics helps decentralized structures to create new combinations and reuse of existing technologies, consistent with the ability of analytics to link knowledge across diverse domains and to integrate external knowledge into the firm. Furthermore, the effect primarily comes from the analytics capabilities of the noninventor employees as opposed to inventors themselves. These results show that the benefit of analytics on innovation depends on existing organizational structures. Similar to the IT–productivity paradox, these results can help explain a contemporary analytics–innovation paradox—the apparent slowdown in innovation despite the recent increase in analytics investments. This paper was accepted by Anandhi Bharadwaj, information systems.

Advances in artificial intelligence (AI) could potentially reduce the complexities and costs in drug discovery. We conceptualize an AI innovation capability that gauges a firm’s ability to develop, manage, and utilize AI resources for innovation. Using patents and job postings to measure AI innovation capability, we find that it can affect a firm’s discovery of new drug-target pairs for preclinical studies. The effect is particularly pronounced for developing new drugs whose mechanism of impact on a disease is known and for drugs at the medium level of chemical novelty. However, AI is less helpful in developing drugs when there is no existing therapy. AI is also less helpful for drugs that are either entirely novel or those that are incremental “follow-on” drugs. Examining AI skills, a key component of AI innovation capability, we find that the main effect of AI innovation capability comes from employees possessing the combination of AI skills and domain expertise in drug discovery as opposed to employees possessing AI skills only. Having the combination is key because developing and improving AI tools is an iterative process requiring synthesizing inputs from both AI and domain experts during both the development and the operational stages of the tool. Taken together, our study sheds light on both the advantages and the limitations of using AI in drug discovery and how to effectively manage AI resources for drug development.

Abstract. Nearly all personal relationships exhibit a multiplexity where people relate to one another in many different ways. Using a set of faculty CVs from multiple research institutions, we mined a hypergraph of researchers connected by co-occurring named entities (people, places and organizations). This results in an edge-sparse, link-dense structure with weighted connections that accurately encodes faculty department structure. We introduce a novel model that generates dyadic proposals of how well two nodes should be connected based on both the mass and distributional similarity of links through shared neighbors. Similar link prediction tasks have been primarily explored in unipartite settings, but for hypergraphs where hyper-edges out-number nodes 25-to-1, accounting for link similarity is crucial. Our model is tested by using its proposals to recover link strengths from four systematically lesioned versions of the graph. The model is also compared to other link prediction methods in a static setting. Our results show the model is able to recover a majority of link mass in various settings and that it out-performs other link prediction methods. Overall, the results support the descriptive fidelity of our text-mined, named entity hypergraph of multi-faceted relationships and underscore the importance of link similarity in analyzing link-dense multiplexitous relationships.