Knowledge that Transforms

Abstract Research Summary Organizations often struggle to identify promising innovations that balance novelty and feasibility in multidisciplinary domains, yet how does evaluator expertise heterogeneity shape these assessments? This study examines how evaluator expertise influences innovation evaluation through a field experiment with National Aeronautics and Space Administration's (NASA) Astrobee Robotic Arm Challenge, involving 354 evaluators assessing 101 solutions. Domain‐spanning evaluators assign higher novelty ratings while maintaining similar feasibility ratings compared to domain‐specific evaluators. Domain‐adjacent evaluators show higher ratings on both dimensions. Human‐LLM analysis of 3007 evaluator comments reveals a two‐stage process: feasibility filtering (evaluating minimum viability) followed by integrative assessment (evaluating enhancement potential). Different expertise types serve complementary functions: domain‐spanning evaluators recognize enhancement potential while maintaining rigorous standards; domain‐adjacent evaluators show openness to novel approaches; domain‐specific evaluators ensure technical rigor. These findings suggest effective innovation evaluation depends on strategically combining complementary expertise types rather than identifying optimal individual evaluators. Managerial Summary Organizations often struggle to identify innovations that are both novel and feasible, risking missed breakthroughs or wasted resources. We study how evaluator expertise shapes innovation assessments in a field experiment with NASA, in which 354 evaluators reviewed 101 robotic arm designs. Evaluators with expertise spanning multiple domains recognize more novel yet feasible ideas. Those with single‐domain expertise provide essential technical gatekeeping but overlook cross‐domain improvements, while adjacent‐field experts are more open but less rigorous. Organizations can strengthen innovation selection by strategically combining these complementary expertise types—using domain‐specific experts for initial feasibility screening and domain‐spanning experts to identify integrative opportunities—rather than seeking one “ideal” evaluator.

Abstract Research Summary Co‐using tasks—using the same task to produce more than one product—promises economies of scope. However, task co‐use also ties products together, changing a firm's task network by introducing cross‐product interdependencies. In light of these interdependencies, we identify an unrecognized downside of task co‐use for product reliability: Cross‐product interdependencies complicate task design and increase the risk of malfunctions. Hence, task co‐use may carry a reliability penalty. Second, we propose that the reliability penalty of task co‐use may shrink with experience as cross‐product interdependencies allow for cross‐improvements and therefore potentially higher improvement rates of task co‐using products. We study the US automotive industry and find our predictions supported, indicating that task co‐use is a critical design parameter with important consequences for reliability and product improvement. Managerial Summary Using the same tasks to produce multiple products can help firms lower costs and achieve economies of scope. However, this practice also links products more tightly together, creating interdependencies across product lines. These interdependencies can make tasks harder to design and manage, increasing the likelihood of errors or malfunctions. As a result, task co‐use may reduce product reliability—an important but often overlooked drawback. At the same time, co‐used tasks can enable portfolio‐wide improvements— improvements across products. As firms gain experience, problems identified in one product can lead to improvements that benefit others using the same tasks. Over time, these cross‐improvements accelerate product improvement for task co‐using products and can reduce the initial reliability disadvantage. Evidence from the U.S. automotive industry supports these arguments. The findings show that task co‐use is a key design decision that involves meaningful trade‐offs between cost savings, reliability, and improvement. For managers, this highlights the importance of carefully evaluating when and how tasks should be co‐used across products, as these choices have lasting implications for product quality and improvement.

Abstract Research Summary We track the enactment of targeted regulation of abortion providers (TRAP) laws in the United States and analyze 4.98 million person‐year mobility records for 535,568 biomedical scientists from 1990 to 2018. Our estimations reveal a 0.8–1.6 percentage‐point increase in scientists' relocation probability after states enacted abortion‐restrictive laws, with substantially stronger effects among junior scientists (1.6–3.9 percentage points). Anti‐abortion states also became less likely to be chosen as relocation destinations, particularly by higher‐quality scientists. These responses appear driven by ideological misalignment and research‐related concerns in fields affected by abortion‐related regulation. We further find that states that have adopted TRAP laws experienced declines in scientific research quality, federal research funding, and patenting in relevant technological fields and among local firms. Managerial Summary Social policies influence the location choices of creative talent, with further implications for organizational and regional innovations. Research scientists are an important part of the regional innovation ecosystem, whose work is intricately related to innovation activities in firms. In a large‐scale study of 4.98 million records of half a million biomedical scientists from 1990 to 2018, we reveal that restrictive policies, such as restrictions on abortion providers, deter high‐quality scientific talent from relocating to certain regions. Our analysis highlights that ideological (mis)alignment and research‐opportunity‐related concerns, rather than practical concerns like access to abortion services, heighten the effect of abortion‐restriction laws on the observed mobility patterns, underscoring the broader implications of policy choices in the sociopolitical environment on where scientific talent chooses to live and work.

Abstract Research Summary Knowledge accumulation requires that we understand whether and when relationships identified in any research setting generalize to others—that is, suggesting domains where results hold (or not). Strategy scholars carefully identify how theoretical mechanisms operate in their chosen research contexts, but attend less to whether their findings apply in other contexts. Accordingly, we recommend reframing empirical contexts, typically described in terms of nominal categories (e.g., industries, countries, or time periods), by highlighting contextual attributes of the nominal settings (e.g., industry concentration or technological modularity), which in turn reflect more abstract conceptual categories (e.g., uncertainty, interdependence, and variance) across potential research contexts. This approach can help integrate prior findings and suggest future study contexts to better enhance our understanding of the research landscape. Managerial Summary Academic research in strategic management tends to derive findings in very specific nominal settings—particular industries, years, and regions. Since strategy practitioners operate across a wide variety of industries and regions, they need to assess whether available research findings are applicable in their own settings. We suggest that understanding whether and when research findings apply to unstudied settings can be facilitated by categorizing research settings using more abstract conceptual constructs (such as environment uncertainty, variance across firms, or interdependence between firms), rather than by the traditional emphasis on nominal settings. We discuss a variety of research setting attributes (such as industry concentration and technological modularity) that can aid the translation of extant research findings to settings where practitioners are operating.

Abstract Research Summary Academic spin‐off (ASO) performance has been studied in relation to either specific university‐level or regional‐level characteristics. However, ASOs originate from universities, which are embedded in regional ecosystems. This nested structure can create an attribution problem when either level is studied in isolation. Consequently, the relative importance of these two different levels for ASO performance has remained ambiguous. To address this ambiguity, we rely on multi‐level modeling and use a novel, hand‐collected dataset of 3164 ASOs founded between 2010 and 2019 from 212 universities nested within 99 European regions. We find that the region effect matters for about 27% for Return on Assets and 16% for Sales, whereas the university effect is negligible. Our study contributes to research at the nexus of academic entrepreneurship and variance decomposition in strategy. Managerial Summary Academic spin‐offs (ASOs) bring innovations from the university to the market, thereby potentially generating new sales, employment, and value. However, once formed, their performance prospects vary significantly, and understanding this variance is important for entrepreneurs and policymakers alike. Our findings from a new European dataset reveal that the region effect matters for ASO performance, but the university effect is negligible. This evidence does not imply that universities lack importance; rather, it suggests that universities in a region may have a collective impact that diffuses into regional resources and networks. Our evidence highlights the importance of fostering a supportive regional ecosystem.

Abstract Research Summary CEO promises are powerful but understudied communication tools. We develop a dual‐mechanism framework theorizing that while CEO promises elevate stakeholder expectations, they simultaneously constrain strategic flexibility. We argue that CEO promise‐making is shaped by two competing pressures: making more promises when the need to manage expectations upward is heightened and fewer promises when the need to preserve flexibility is increased. Further, we predict that under heightened uncertainty, CEOs preserve flexibility through strategic ambiguity (i.e., issuing promises with extended or vague time horizons and lower specificity). Leveraging Large Language Models (LLMs) to analyze over 69,000 earnings‐call transcripts from S&P 1500 firms (2010–2022), we identify 74,017 CEO promises and find support for our predictions. We contribute an original, publicly available dataset of CEO promises. Managerial Summary When CEOs publicly promise positive future results, they shape how investors and analysts view the company. These promises are a double‐edged sword: they can boost investor and stakeholder confidence, but they also lock the company into a pre‐determined path and make later shortfalls or pivots reputationally costly. Analyzing more than 69,000 earnings calls (2010–2022), we find that CEOs make more promises when needing to prove their legitimacy, such as during early tenure, when facing gender bias, or after missing earnings targets. However, in uncertain environments, CEOs often pull back from promise‐making or employ “strategic ambiguity” by making promises with vaguer timelines and details to maintain flexibility. Our analysis suggests that failing to deliver on these public commitments significantly increases the likelihood of CEO dismissal.

Abstract Research Summary We study how the entry of a rival platform affects the strategies of the incumbent's complementors. The latter face a trade‐off: While the entry threatens their benefits from indirect network effects, it also allows them to escape intense within‐platform competition. Studying Epic Games' entry into the PC video game market—until then dominated by Steam—we show that this trade‐off does not resolve uniformly, driving heterogeneity in strategic reactions. Complementors with weaker strategic resources (independent developers) were more likely to multihome and became less responsive to the incumbent's attempts to orchestrate collective action through platform‐wide sales promotions. In contrast, complementors more reliant on indirect network effects (multiplayer developers) were less likely to multihome and became more responsive to orchestration attempts. Managerial Summary As competition between digital platforms intensifies, complementors—firms that provide complementary products or services—must adjust how they engage with platform owners. The entry of a rival platform creates both opportunities and risks: it offers an alternative with less intense competition but also fragments the user base that underpins network benefits. We find that these opposing effects shape complementors' behavior differently. Those with fewer strategic resources are more likely to join the new platform and become less willing to follow the incumbent's coordinated initiatives. In contrast, complementors whose products rely strongly on network effects tend to remain with the incumbent and cooperate more closely with its orchestration efforts. For managers, this highlights that platform competition not only shifts market dynamics but also reshapes the motivations and strategies of heterogeneous complementors.

Abstract Research Summary Entrepreneurship accelerators are increasingly promoted as structural interventions to close gender gaps, yet studies have not established a differential impact of participation for women. This prior evidence—drawn from high‐tech, male‐dominated settings—may overlook how outcomes differ in more feminized domains such as social innovation. Using unique multi‐level data from 1417 ventures applying to 33 accelerators, we examine whether average effects conceal variation across institutional environments and program designs. We find that in more gender‐egalitarian countries, women‐led ventures performed better than peers after participating, especially in programs aiming to support women. In less egalitarian environments, however, participation offered negative or no such advantages even in gender supportive programs. These results suggest that well‐intended interventions to advance women entrepreneurs are context‐dependent and may sometimes reinforce the very disparities they intended to diminish. Managerial Summary Social innovation accelerators are often championed as a way to help women entrepreneurs thrive. But our analysis of 1417 ventures applying to 33 programs shows that results vary—and context matters. In more egalitarian countries, participation is associated with improved performance for women‐led ventures, especially when programs are explicitly designed to support them. Surprisingly, the pattern flips in less egalitarian environments: women founders who join even women‐focused programs often see no performance gains—or even declines—relative to peers. These findings suggest that well‐intentioned support can backfire if it doesn't fit the local institutional environment. For program designers and leaders, this is a call to move beyond one‐size‐fits‐all approaches and carefully align equity goals with the realities of the ecosystems they operate in.

Abstract Research Summary Firm technological research has the potential to spawn multiple applications. Despite recognizing such potential, past literature disagrees on the process through which firms discover and grow new applications out of their past technological research. I examine this question in the context of deep learning, taking a question‐driven approach. Difference‐in‐difference analysis suggests that firms radically increased cumulative investments in past deep learning research upon signals indicating elevated application potential of deep learning. Furthermore, rather than investing in proprietary efforts, firms disclosed their cumulative development trajectories to engage external innovation efforts from which they learn and build. Grounded in these findings, I propose that the discovery and growth of new applications of past research entails unfolding innovation interdependence which motivates firms to co‐evolve with external innovators. Managerial Summary Firm technological research has the potential to spawn multiple applications. This article examines how firms cumulatively invest in their past technological research to grow new applications in the context of deep learning. Employing a difference‐in‐difference approach, analysis suggests that firms radically increased cumulative investments in deep learning after a shock elevating the application potential of their past deep‐learning research. Furthermore, firms publicly disclosed their cumulative development trajectories to attract innovation efforts from application sectors while actively learning from the attracted efforts to innovate further. These findings suggest that firms engaged, leveraged and co‐evolved with external innovation efforts to discover and grow new applications of their past research.

Abstract Research Summary Complementing the role of AI in facilitating search and identifying combinations of high value in inventive activity, we argue that AI fundamentally alters the innovation landscape by unlocking new combinations that were previously infeasible. This effect arises because AI acts as a powerful shared layer due to its predictive capabilities and its ability to transmit solutions across domains, thereby creating a bridge between previously unconnected elements. Utilizing a matched sample of patents, we show that inventions incorporating AI exhibit a greater degree of novel recombinations compared to those without AI, and that our proposed bridging mechanism is consistent with these novel recombinations. Our study contributes by identifying a new mechanism by which recombinations emerge in inventive activity while also highlighting the role of enabling technologies such as AI in facilitating such recombinations. Managerial Summary How does AI impact inventive activity? We examine the question by studying the patenting activity of US firms over the period 2005–2023. We argue that AI fundamentally alters the innovation landscape by unlocking new combinations that were previously infeasible. This effect arises because AI acts as a powerful shared layer due to its predictive capabilities and its ability to transmit shared solutions, thereby creating a bridge between previously unconnected technological elements. Our analysis demonstrates that inventions that build on AI involve novel recombinations to a greater degree compared to inventions that don’t, and that AI bridges and connects knowledge domains that were hitherto disparate. These findings indicate that AI is more than just an invention of a new method of invention, and that it fundamentally reshapes the nature of inventive activity.