RESEARCH INTERESTS
Overall, I am interested in how interdependence (in goals/criteria, task/knowledge) in innovation and evaluation process shapes organizational actions and performance. I study this question in two contexts: product development empowered by A/B tests, and firm innovation strategies with deep learning as an enabling technology.
PEER-REVIEWED PUBLICATION
Selected On-Going Work:
I examine whether and how the availability of key infrastructure technologies affects firms’ cumulative invention efforts around the enabling technologies that they created in the past.
The empirical context is deep learning (2003-2020), wherein an unexpected breakthrough in GPU architectures (2008) made a critical infrastructure technology——high power processing units—widely available to developers of deep learning. Using a difference-in-difference design, I found that the availability of infrastructure technology increases corporate cumulative inventions on past deep learning inventions by 44.6% on average. Mechanism analyses suggest that disclosure and learning largely drive the observed relationship: the increase in corporate cumulative inventions concentrates in scientific publications rather than patents, and in inventions citing external forward citations rather than direct self-citations. Existing theories emphasizing the tradeoff between value-creation prospects or competition concerns cannot fully explain these empirical results. Instead, I propose a novel view anchored in innovation interdependence between application sector incumbents and enabling technology firms: when the availability of infrastructure technology brings application sector incumbents on board, enabling technology firms are further incentivized to cumulatively innovate—a cascading Domino effect that overtime increases the payoff to cumulative invention, disclosure, and learning.
Summary:
Within complex organizations, different functional departments pursue different yet interdependent goals. While previous research has emphasized the challenges posed by goal interdependencies in organizational search, it has not sufficiently delved into the differential impacts of positive (i.e., goal complementarity) versus negative goal interdependencies (i.e., goal conflict) on search. This study examines how positive and negative cross-departmental goal interdependencies influence department-level product innovation search. We posit that positive goal interdependencies trigger serendipity-driven search to capitalize on previously unrecognized cross-departmental complementarity, while negative goal interdependencies induce conflict-driven search to resolve cross-departmental goal conflict. Empirical support is found from data on over 15,000 product development A/B tests in a large internet company. Our study contributes to the broad literature on search within complex organizations and behavioral strategy.
Drawing from status hierarchy research, we also test whether status differentiation within the team moderates the relationship between team failure and teams’ tendency to add high-status members. We predict that teams with lower status differentiation will respond to failure by adding high-status members, whereas this effect will be weaker in teams with higher status
differentiation. We analyzed longitudinal archival data containing 25,035 performance episodes of 5,283 product development teams in a large IT company and found support for the contingency effect of status differentiation. This research helps advance research on team failure
and social hierarchy.
Overall, I am interested in how interdependence (in goals/criteria, task/knowledge) in innovation and evaluation process shapes organizational actions and performance. I study this question in two contexts: product development empowered by A/B tests, and firm innovation strategies with deep learning as an enabling technology.
PEER-REVIEWED PUBLICATION
- Xirong (Subrina) Shen, H. Li, and P. Tolbert. "Converging Tides Lift All Boats: Consensus in Evaluation Criteria Boosts Investments in Firms in Nascent Technology Sectors", Organization Science (2021)
- Xirong (Subrina) Shen, H. Kim, and J. Li. "Funding Ventures Similar To One of Us: How Status Dynamics Impact Similarity Bias in Heterogeneous Investment Teams", Strategic Management Journal, In Press (2022)
Selected On-Going Work:
- Triggering a Domino Effect: Infrastructure Technology and Corporate Cumulative Invention Trajectory of Deep Learning
I examine whether and how the availability of key infrastructure technologies affects firms’ cumulative invention efforts around the enabling technologies that they created in the past.
The empirical context is deep learning (2003-2020), wherein an unexpected breakthrough in GPU architectures (2008) made a critical infrastructure technology——high power processing units—widely available to developers of deep learning. Using a difference-in-difference design, I found that the availability of infrastructure technology increases corporate cumulative inventions on past deep learning inventions by 44.6% on average. Mechanism analyses suggest that disclosure and learning largely drive the observed relationship: the increase in corporate cumulative inventions concentrates in scientific publications rather than patents, and in inventions citing external forward citations rather than direct self-citations. Existing theories emphasizing the tradeoff between value-creation prospects or competition concerns cannot fully explain these empirical results. Instead, I propose a novel view anchored in innovation interdependence between application sector incumbents and enabling technology firms: when the availability of infrastructure technology brings application sector incumbents on board, enabling technology firms are further incentivized to cumulatively innovate—a cascading Domino effect that overtime increases the payoff to cumulative invention, disclosure, and learning.
- (Title omitted for blind review) Topic: Goal Interdependencies and Product Innovation Search
Summary:
Within complex organizations, different functional departments pursue different yet interdependent goals. While previous research has emphasized the challenges posed by goal interdependencies in organizational search, it has not sufficiently delved into the differential impacts of positive (i.e., goal complementarity) versus negative goal interdependencies (i.e., goal conflict) on search. This study examines how positive and negative cross-departmental goal interdependencies influence department-level product innovation search. We posit that positive goal interdependencies trigger serendipity-driven search to capitalize on previously unrecognized cross-departmental complementarity, while negative goal interdependencies induce conflict-driven search to resolve cross-departmental goal conflict. Empirical support is found from data on over 15,000 product development A/B tests in a large internet company. Our study contributes to the broad literature on search within complex organizations and behavioral strategy.
- (Title omitted for blind review) Topic: Status Dynamic and Turnarounds in Product Innovation Teams
Drawing from status hierarchy research, we also test whether status differentiation within the team moderates the relationship between team failure and teams’ tendency to add high-status members. We predict that teams with lower status differentiation will respond to failure by adding high-status members, whereas this effect will be weaker in teams with higher status
differentiation. We analyzed longitudinal archival data containing 25,035 performance episodes of 5,283 product development teams in a large IT company and found support for the contingency effect of status differentiation. This research helps advance research on team failure
and social hierarchy.