2025 Theses Doctoral

Three Papers on the International Political Economy of AI: Statistical and Game-Theoretic Analyses

Xiao, Wenquan

This three-paper dissertation merges theory and practice to explore the internationalpolitical economy (IPE) of emerging technologies, with a focus on artificial intelligence (AI).

Through three interconnected papers, it examines how states and firms interact in a dynamic cycle of action and reaction around technological development. The first two papers explore complementary aspects of this relationship: how states strategically invest in emerging technologies, and how other states react to these investments and technological progress through economic coercion. The third paper then reveals a crucial causal mechanism underlying both state investment decisions and coercive responses—how states gauge each other’s technological capabilities through the actions of firms. Together, these papers explore how states optimize investment strategies, impose economic sanctions, and gauge technological capabilities based on firm actions.

The first paper addresses the core problem of balancing state investments against the risk of foreign restrictions in strategic sectors. It highlights how sectoral characteristics, such as capital intensity and reliance on global networks, influence the relative importance of state investment and foreign access. It then offers a game theoretic framework to explain how governments determine their optimal investment levels based on the trade-off between the two factors. To test the empirical viability of the theory, I use comparative case studies of the semiconductor and AI sectors in China and the space technology and robotics sectors in the U.S.

The second paper investigates the timing of economic sanctions through the framework of “calculated coercion.” Using a dynamic game theory model, this paper argues that states often impose sanctions not when their leverage is strongest but when it is at the brink of erosion. This theory explains why the U.S. imposed semiconductor export controls as China neared technological independence, why Russia weaponized energy exports as Europe diversified its energy sources, and why China has so far refrained from weaponizing its manufacturing leverage. These findings reveal that, paradoxically, attempts at pursuing economic independence could trigger economic coercion.

The third paper turns the attention to the role of firms, focusing on how states gauge each others’ capabilities through the capabilities of firms. Using the case of Google DeepMind’s AlphaGo AI model release in 2016, the paper employs the synthetic control method and a unique dataset of over 11,000 Chinese military publications to show how AlphaGo’s release served as a “Sputnik moment” for China, spurring a significant increase in military AI publications by the People’s Liberation Army (PLA). Qualitative analysis confirm the hypothesis that the Chinese military used DeepMind’s capabilities as a signal for U.S. AI power. Together, these three papers contribute to the understanding of the IPE dynamics of AI, revealing how states navigate emerging technologies, global competition, and geopolitical tensions.