Why do developing countries borrow predominantly in foreign currency rather than domestic currency (i.e., the “original sin”)? Why is most external debt denominated in the U.S. dollar rather than other advanced economies' currencies? Since the currency denomination shapes emerging markets' balance-sheet exposure to the hegemon's monetary spillovers, understanding its determinants has direct policy implications for debt sustainability and financial market stability. This paper brings the trade literature—the Melitz (2003) model and Eaton-Kortum (2002) model—to bear on whether models that explain bilateral trade flows also account for the currency denomination of external debt. To rationalize the empirical fact that most external debt is denominated in U.S. dollar, I introduce non-negligible fixed entry costs of borrowing in an additional currency. From a finance perspective, I also incorporate an optimal portfolio choice model to assess consistency with the trade mechanism. The key insights highlight the importance of trade frictions in the international borrowing markets, which include (i) the fixed-cost margin generates threshold behavior in currency adoption (extensive margin), and (ii) the heterogeneous creditors generate strategic complementarities, so small cost-push shocks need not necessarily overturn currency denomination shares.

This research project develops a novel historical dataset on prices of tradable goods and non-tradable services (occupation-level wages), returns on financial assets (bank deposit rates across maturities), as well as firms' balance sheets, across U.S. counties and international borders from 1850 to 2000. Using advanced machine learning methods, I examine printed advertisements from millions of digitized historical newspapers in order to shed light on (i) price stickiness and firms' markup structures at both domestic and international scales, (ii) the heterogeneous and distributional effects of macroeconomic policies—such as tariffs and monetary shocks—on realized inflation experienced by households from different income groups and demographic compositions across geographies, and (iii) the differential factor prices for capital and labor (e.g., interest/mortgage rates across maturities, and offered wages across service occupations) in different regions. The resulting “Global Price Initiative” database provides researchers with a highly detailed, product-level dataset that supports transparency, replicability, and versatility in macroeconomic and international research on prices.

After the bursting of real estate bubbles in 1991, Japanese banks continued lending to unviable firms to conceal problem loans. We revisit Japan's experience and propose a new mechanism via which banks' loan-evergreening policy undermines allocative efficiency across industries by focusing on construction and real estate loans. Namely, banks' continuing support for construction and real estate firms encourages labor hoarding in unviable construction projects. Since construction projects predominantly use low-skilled workers, banks' loan-evergreening policy in these troubled sectors may depress other low-skilled industries. Based on the industry-level data in each of Japan's 47 prefectures from 1992 to 1996, we document empirical facts consistent with this hypothesis. On average, low-skilled industries experienced disproportionately slower output and employment growth and more sluggish growth in the number of new establishments in prefectures where the share of bank loans to local construction/real estate sectors increased more after construction boom ended.