Abstract: Trade policy increasingly targets firms and products, and evaluating these interventions requires a model with granular substitution patterns. Using an event-study design, I show that textual product similarity predicts reallocation following a ban. Motivated by this evidence, I develop a general-equilibrium trade model with oligopolistic competition. Products differ in attribute mix and quality; substitution varies continuously with product similarity. I apply it to China's Great Firewall (GFW). Removing the GFW raises Chinese consumers' app-based leisure utility but shifts profits from Chinese to foreign companies. In the benchmark calibration, removal lowers Chinese welfare by 0.146% and raises U.S. welfare by 0.311%.

Abstract: This paper studies how e-commerce reshapes the spatial distribution of retail activity, manufacturing, and inter-regional trade. We develop a general-equilibrium model featuring two opposing forces of e-commerce: online search that weakens demand-side agglomeration, and platform logistics that lower trade costs and strengthen supply-side agglomeration. To quantify these forces, we exploit variation from both realized and planned-but-unbuilt Amazon distribution facilities as a quasi-natural experiment. We find that facility expansion causally increases seller entry, sales, and inter-regional trade, with stronger effects for heavy goods and high-transport-cost regions, and positive spillovers to nearby manufacturing employment. A calibrated model shows that Amazon's expansion between 2012 and 2022 raises average state-level welfare by 0.94 percent, but also widens regional disparities in both real income and employment.

Abstract: International trade in digital services, including mobile app trade (e.g., Facebook in Frankfurt, Instagram in Istanbul), is large and rapidly growing. Analyzing 2015--2020 data on 28,000 mobile apps, we find that AI increases an app's user base threefold, a massive effect. AI's handmaiden is data, including cross-border flows of personal data. Concerned with privacy and national security, policymakers have responded with thousands of conflicting data regulations, ranging from guaranteeing free trade in data (USMCA) to data export bans. We are the first to show that these regulations matter: AI's massive effect is halved in countries with strict data flow restrictions.

Abstract: Between 2000 and 2015, China's aggregate income quadrupled, its provincial income inequality fell by a third, and its share of employment in agriculture fell by a half. Internal migration is central to this transformation, with the number of internal migrant workers reaching 300 million by 2015. Combining rich data on migration with a spatial general equilibrium model of China's economy, we quantify the size and the impact of internal migration cost reductions in China between 2000 and 2015. During the 15-year period, China's internal migration costs fell by forty-five percent. In addition to contributing substantially to growth, these migration cost changes account for the majority of the reallocation of workers out of agriculture and the drop in regional inequality. We compare the effect of migration policy changes with other important economic changes, including changes in trade costs, capital market distortions, average cost of capital, and productivity. While each contributes meaningfully to growth, migration policy changes are central to China's structural change and regional income convergence. We also find that the recent slow-down in aggregate economic growth between 2010 and 2015 is associated with a smaller reduction in inter-provincial migration costs and a larger role of capital accumulation. 

Abstract: Artificial Intelligence is a powerful new technology that will likely have large impacts on the size, direction, and composition of international trade flows. Yet almost nothing is known empirically about this. One AI-enabled set of services that can be tracked resides in the palm of our hands: the Mobile Apps used by half the world's population. To analyze the impact of AI on international trade in mobile App services, we merge 2017--2020 data on international downloads of mobile Apps with data on the AI patents held by each App's parent company. We show the following (all results are within detailed App categories such as "social networking"). (1) Trade: The Apps of exporters deploying AI are downloaded three times more often than those of exporters who do not deploy AI.  (2) Variety Effects: AI deployment triples the number of exported App varieties. (3) Creative Destruction: AI deployment increases both the entry of new Apps and the exit of existing Apps. Compared to a world with no AI, AI deployment increases the welfare gains from creative destruction by 1.6 percentage points.