Work in Progress
Working papers are linked or available upon request. All my publications (including non-peer reviewed papers) are listed on my CV.
Under Review
Shooting for the Right Stars? Three Critical Challenges in Pursuit of Commercial Markets in NewSpace
The rise of NewSpace is widely described as a pivotal shift from government-led space exploration to a commercially-driven sector fueled by private investment. Yet, Earth observation (EO) satellite operators remain heavily dependent on government funding while struggling to develop viable commercial markets. Through a hypothesis-generating case study of the EO segment, we identify three interconnected challenges constraining commercial development. First, government (co-)dependence incentivizes upstream technological advances aligned with government rather than commercial priorities, keeping data prices too high to enable thorough downstream exploration. Second, limited commercial demand encourages EO ventures to pre-maturely integrate, creating barriers for entry in the downstream. Third, high prices and vertical integration prevent innovation ecosystem formation necessary for discovering new commercial applications, perpetuating limited demand. Using multi-source data triangulation combining quantitative analysis of satellite data and firm-level financial data with qualitative analysis of procurement documents and firm publications, we show that despite upstream advances, commercial revenue remains limited while government dependence increased during commercialization efforts. Our analysis contributes to debates about procurement as innovation policy, nascent markets, and innovation ecosystems while providing managerial and policy recommendation for realizing NewSpace’s commercial potential. Overall, our effort is to encourage further research and dialogue to foster sustainable, inclusive, and accountable futures for space commerce and governance.
When Stakeholder Engagement Isn’t Enough: When and How Home-Country Reputation Undermines Foreign Infrastructure Projects
How do authoritarian regimes manage dissent during political liberalization? Recent scholarship suggests modern autocracies increasingly adopt legalistic control over overt violence, yet systematic evidence documenting this shift during transitions remains scarce. Using original event-level data from Myanmar’s liberalization in the early 2010s, I examine how state responses to protest evolved during elite-controlled political opening. Legal reforms expanding assembly rights generated a fourfold increase in protest activity, confirming shifts in political opportunity structures. However, state repression adapted rather than disappeared; declining from informal violence toward selective legalistic prosecution that disproportionately targeted pro-democracy organizations. Extended analysis through 2020 shows this pattern persisted even during the politically sensitive 2015 election, with excessive force against protesters remaining virtually absent. These findings challenge assumptions that liberalizing regimes gradually reduce repression, demonstrating instead how transitional governments strategically employ legal mechanisms to maintain control while projecting democratic legitimacy. The study contributes to theories of repressive adaptation in hybrid regimes and provides rare systematic documentation of authoritarian control strategies during elite-managed transitions.
Working Papers
Dissertation
From Burden to Breakthrough: Off-the-Shelf AI/ML for Unstructured Data in Strategy Research
Strategy and management researchers increasingly rely on unstructured data such as news articles, website texts, and regulatory filings to study innovation, adaptation, and strategic behavior. Yet transforming such data into structured formats has remained labor-intensive, costly, and dependent on advanced programming skills, limiting broader adoption. This paper conceptualizes and evaluates the use of off-the-shelf artificial intelligence and machine learning (AI/ML) models, including large language models (LLMs) and topic modeling techniques, as scalable and cost-effective solutions for classifying and coding unstructured text. To assess their practicality, I apply off-the-shelf models to two domains that frequently require extensive unstructured data processing: bibliometric analysis of articles published in top management journals (2004-2024) and protest event analysis (PEA). In bibliometric analysis, the LLM reliably classified full-text academic articles. The analysis shows that up to 75 percent of recent articles rely on unstructured data, yet most still depend on manual coding, highlighting the potential value of accessible automation. In PEA, the LLM matched or outperformed trained human coders in both classification and nuanced coding tasks, in some cases achieving more than twice the accuracy with far less effort. Topic modeling approaches performed considerably worse. Rather than advocating full replacement of human or customized AI/ML pipelines, the paper clarifies where off-the-shelf AI/ML performs well, where trade-offs emerge, and how they can augment existing methods. A companion website provides tutorials, code, and best practices to help researchers apply these tools in their own studies.
Political Loyalty as a Liability: How Corporate Political Connections Affect Survival Amid Democratic Backsliding
How do firms’ political strategies perform when democracy collapses? This paper investigates the effectiveness of corporate political hedging—cultivating ties across rival political camps—under conditions of democratic breakdown. While prior research finds that hedging helps firms navigate regime transitions and electoral turnover in democratizing and consolidated democracies, we know little about its value when democratic progress reverses. I exploit Myanmar’s 2021 military coup as a natural experiment to examine how hedging shaped firm survival following the country’s abrupt return to dictatorship. Linking the universe of firms and their owners to all politicians from the preceding democratic period (2010–2021), I construct a novel dataset from nearly one million leaked registry documents using machine learning and extensive human validation. Difference-in-differences analyses reveal that hedged firms were 20–30 percentage points more likely to survive two years after the coup than both unconnected and pro-democracy-only firms—and even outperformed firms exclusively tied to the military. Follow-up experimental evidence with Burmese activists suggests a mechanism: focused boycott behavior disproportionately targeted firms with exclusive military ties, while hedgers avoided activist punishment and investor flight. The study extends theories of nonmarket strategy and political risk by showing that under democratic breakdown, political hedging primarily protects firms not through balanced access to power, but by reducing exposure to non-governmental retaliation.
Digital Natives: Conceptualizing and Measuring Innovation on the Internet,
Introducing the Internet Dynamism Index (IDX)
Standard innovation measures are operationalized using observable artifacts like patent applications or R&D spending, which are by-products of ‘good’ institutions or resources allocated to innovation-related activity. Unlike conventional innovators, however, internet innovators engage much less with local resources and institutions; and in some countries, these are practically absent altogether. To address this ‘observation problem,’ I propose the Internet Dynamism Index (IDX), a novel comparative measure of internet innovation. I leverage domain name zone file data to construct the general population of all websites, randomly sample a large subset of domains each month, scrape their current and historic content, and apply various classification algorithms to identify key dimensions of Schumpeterian innovation, geographic origin, and more. After aggregation and standardization, the IDX generates ordinal rankings at the country and city levels that diverge significantly from those produced by conventional measures, also suggesting internet innovation provides unique opportunities for growth in regions lacking traditional comparative advantage in innovation. I validate the IDX through regressions showing distinct patterns of association with predictors of innovation and its correlation with economic performance. Using an instrumental variable approach in the context of Russia’s invasion of Ukraine, I show that internet innovation performance is more resilient to crises than conventional innovation. These findings highlight the economic distinctiveness of internet innovation and its potential to foster sustainable growth, particularly in regions that are unable to maintain or entirely lack conventional innovation. I offer strategic insights for managers and policymakers to optimize location decisions and innovation policies.
Others
How to Detect AI-Generated Answers to Open-Ended Questions?
(Previous title: “AIDetection: A Generative AI Detection Tool for Educators Using Syntactic Matching of Common ASCII Characters As Potential ‘AI Traces’ Within Users’ Internet Browser”)
This paper introduces a novel method to detect AI-generated answers to open-ended questions in surveys, survey experiments, and other research and educational settings. Building on new concerns from AI- rather than human-generated answers, the study presents both a browser-based tool and an accompanying R package which implement a transparent, heuristic-based approach for identifying linguistic and syntactic artifacts commonly produced by large language models such as ChatGPT, Claude, Gemini, Llama, DeepSeek, and others. Unlike proprietary AI-detection systems that rely on opaque machine learning classifiers, this approach focuses on interpretable textual features that can distinguish human from AI-generated responses in many contexts. The paper outlines the detection logic, validates it, and proposes a data processing workflow that researchers can readily implement. It also discusses research applications for quality control in online surveys, crowdsourced experiments, and educational assessments, highlighting implications for reproducibility and ethical data collection in the age of generative AI.
Surviving Dictatorship: Organizational Responses to Democratic Backsliding
This study examines how sudden democratic regime breakdown—an increasingly common outcome of democratic backsliding—affects the organizational environment and how organizations respond to these changes. While existing theory suggests that extreme environmental uncertainty leads organizations to either exit or fall into paralysis, I argue that organizations in dictatorships often deploy a wider range of strategic responses, including operating in exile or underground. Drawing on Oliver’s (1991) seminal work on responses to institutional pressures, I theorize how these response choices are shaped by organizational resources, stakeholder preferences, and perceptions of the regime’s coercive capacity. I explore the plausibility of my theory using a mixed-methods approach that triangulates pre-coup survey data, topic modeling of organizational texts, and in-depth interviews to track sixty-six civil society organizations in Myanmar before and after the 2021 military coup. By examining how organizations navigate contradictory institutional pressures in contexts of autocratic restoration, this research offers implications for organizational resilience that extend beyond Myanmar.
Protest, Repression, and the Limits of Early Liberalization
How do authoritarian regimes manage dissent during political liberalization? Recent scholarship suggests modern autocracies increasingly adopt legalistic control over overt violence, yet systematic evidence documenting this shift during transitions remains scarce. Using original event-level data from Myanmar’s liberalization in the early 2010s, I examine how state responses to protest evolved during elite-controlled political opening. Legal reforms expanding assembly rights generated a fourfold increase in protest activity, confirming shifts in political opportunity structures. However, state repression adapted rather than disappeared; declining from informal violence toward selective legalistic prosecution that disproportionately targeted pro-democracy organizations. Extended analysis through 2020 shows this pattern persisted even during the politically sensitive 2015 election, with excessive force against protesters remaining virtually absent. These findings challenge assumptions that liberalizing regimes gradually reduce repression, demonstrating instead how transitional governments strategically employ legal mechanisms to maintain control while projecting democratic legitimacy. The study contributes to theories of repressive adaptation in hybrid regimes and provides rare systematic documentation of authoritarian control strategies during elite-managed transitions.
Collective Memory and Social Group Mobilization: Insights from Youths in Myanmar
Why are youths often at the forefront of street protests, even if the state represses them the most? I investigate this question in the context of transitional Myanmar. I argue that many factors commonly associated with youth activism elsewhere were largely absent when Myanmar began liberalizing in 2011, providing opportunities for generating new theoretical insights. To demonstrate that youths were indeed among the first to mobilize after political opportunity structures changed, and that they were the most active social group in the years after, I examine protest event data from Myanmar. Based on in-depth interviews that I conducted with a diverse set of youth activists in 2017 and 2018, as well as survey experiments conducted via face-to-face interviews with a representative sample of Yangon residents in 2019, I argue that collective memory of past youth-led struggles, particularly of the 8888 Uprising, has been instrumental in the Myanmar case. I find that collective memory has instilled a sense of moral obligation in those self-identifying as youth to ‘finish’ what previous youth generations have started. Activists can appeal to this in order to mobilize youths, making collective memory politically salient. For autocrats, collective memory is difficult to repress as it can be shared in the safety of one’s home; it may even become ‘stickier’ when the state tries to repress public sharing. While derived from Myanmar, my theory has broader applicability to contexts where the state suppresses the open commemoration of focal events, fueling the intergenerational transmission of collective memory that can potentially mobilize future protests. It may also help explain why and how past atrocities can backfire on authoritarian leaders much later in time, especially when opportunities for collective action shift.
Does Economic Salience Reduce Support for Radical Movements?
This paper examines how economic salience shapes support for radical political movements, focusing on the case of Hong Kong. While many observers attribute the city’s recent social unrest to economic grievances—especially among younger generations—such accounts remain largely untested. Drawing on a survey experiment embedded in a representative household survey conducted in early 2020, I randomly exposed respondents to varying degrees of economic salience before asking about their support for Hong Kong independence in a hypothetical, central government–approved referendum. Framing the question in this way isolates attitudes toward independence itself, apart from concerns about legality or conflict. Contrary to the prevailing economic grievance narrative, I find that greater economic salience increases support for independence rather than dampening it. These findings challenge conventional explanations linking economic hardship to moderation and suggest that material concerns may, under certain conditions, heighten support for radical political change.
Contact
Email: andybu@umich.edu
Links: Google Scholar · ORCID · LinkedIn