Working Papers

Abstract: We document novel evidence that semiconductors are a critical innovation input whose price declines and technological advancements drive other sectors' R&D, making semiconductor sanction policies amid U.S.—China technology competition highly consequential for growth and welfare. To evaluate counterfactual regimes, we develop and estimate a two-country endogenous growth model in which semiconductor firms innovate strategically along a quality ladder and other firms use semiconductors in innovation. Export controls reallocate chips from China to the U.S., raising U.S. innovation in other sectors while reducing Chinese counterpart. Restrictions on semiconductor technology diffusion raise the returns to U.S. technological leadership and stimulate U.S. semiconductor innovation. Although market scale loss ultimately dampens long-run growth in both countries, these trade and technology restrictions generate net welfare gains for the U.S. They also amplify U.S. gains while nearly eliminating China’s gains from an AI boom. These policies induce greater U.S. innovation when the pre-sanction U.S.—China technology gap is narrow.
  1. Firm-Level Origins of Tariff Pass-Through: Importer Heterogeneity and Welfare Decomposition [Draft] [SSRN], with Chengyuan He, Chang Liu, and Soo Kyung Woo  
     Best Paper Award, Korea-China Economic Development Forum, 2026Abstract: Using confidential U.S. Census data, we show that tariff pass-through to U.S. import prices during the U.S.–China trade dispute was incomplete within continuing trade relationships, even though it was complete in aggregate. We develop an empirical decomposition showing that complete aggregate pass-through reflects import reallocation toward higher-priced new relationships and importers. We then derive a welfare decomposition that applies to a broad class of heterogeneous-importer models. The formula identifies the key statistics needed to estimate the welfare effects of tariffs and shows how importer heterogeneity and sourcing adjustments matter. Quantitatively, ignoring these margins understates the welfare losses from tariffs. Our methods can be used to study other forms of pass-through and heterogeneous responses to common cost shocks.

Abstract: I provide new facts on firm type, innovation quality and innovation frequency across countries. In developed countries, firm type, innovation quality, and innovation frequency are mostly determined at birth. In developing countries, there are fewer innovative and radically innovative firms. Additionally, these firms have lower radical innovation frequency but higher incremental innovation frequency. The cross-country difference is mainly observed in high external financial dependence industries. To explain these facts, I build a semi-endogenous growth model with financial frictions and firm heterogeneity and demonstrate that collateral constraints combined with limited liability conditions impede innovative firm creation and distort firms' choice between radical and incremental innovation in developing countries. I offer new policy implications for countries at different development stages.  

Abstract: Green innovation has slowed across OECD economies since 2010 despite growing climate concern and policy support. Using U.S. patent and firm data, we show that the slowdown is concentrated among firms with information and communication technology (ICT) capabilities. We develop and estimate a general-equilibrium model with heterogeneous innovation efficiency, technology choice, and firm dynamics. The model attributes part of the slowdown to weaker ICT-enabled green innovation and reveals a policy trade-off: ICT-targeted green innovation subsidies are more effective when dirty innovation is costly, whereas uniform green innovation subsidies can dominate when dirty innovation is cheap because of general-equilibrium crowd-out.

Abstract: Recent advancements in digitisation and the green transition have amplified the role of ICT (information and communications technology) in productivity growth. To optimise ICT's potential for future growth, it is essential to identify and address barriers to its progress. My study examines the historical disparity in aggregate productivity growth across countries since the ICT revolution. I find that firms in less developed financial markets, particularly small ones in industries with high ICT intensity and high external financial dependence, have experienced relatively slower growth in productivity and patents. To rationalize these findings, I build a model featuring endogenous firm productivity growth through innovation investment, capital investment, and financial frictions. Financial frictions reduce aggregate productivity growth through two channels: innovation investment and misallocation, with the former being the dominant factor. The model also provides conditions under which rapid capital and output growth can coexist with slow productivity growth due to firms' non-linear trade-offs in allocating constrained investment between capital and productivity.