@article{xu2022reshaping,
title={Reshaping global trade: the immediate and long-run effects of bank failures},
author={Xu, Chenzi},
journal={Quarterly Journal of Economics},
volume={137},
number={4},
pages={2107--2161},
year={2022},
publisher={Oxford University Press}
}

I show that a disruption to the financial sector can reshape the patterns of global trade for decades. I study the first modern global banking crisis originating in London in 1866 and collect archival loan records that link multinational banks headquartered there to their financing abroad. Countries exposed to bank failures in London immediately exported significantly less and did not recover their lost growth relative to unexposed countries. Their market shares within each destination remained significantly lower for four decades. Decomposing the persistent market-share losses shows that they primarily stem from lack of extensive margin growth, as the financing shock caused importers to source more from new trade partners. Exporters producing more substitutable goods, those with little access to alternative forms of credit, and those trading with more distant partners experienced more persistent losses, consistent with the existence of sunk costs and the importance of finance for intermediating trade.

@article{frydman2023banking,
title={Banking Crises in Historical Perspective},
author={Frydman, Carola and Xu, Chenzi},
journal={Annual Review of Financial Economics},
year={2023},
volume={15},
pages={265--290},
}

This article surveys the recent empirical literature on historical banking crises, defined as events taking place before 1980. Advances in data collection and identification have provided new insights into the causes and consequences of crises both immediately and over the long run. We highlight three overarching threads that emerge from the literature: First, leverage in the financial system is a systematic precursor to crises; second, crises have sizable negative effects on the real economy; and third, government interventions can ameliorate these effects. Contrasting historical episodes reveals that the process of crisis formation and evolution varies significantly across time and space. Thus, we also highlight specific institutions, regulations, and historical contexts that give rise to these divergent experiences. We conclude by identifying important gaps in the literature and discussing avenues for future research.

@article{xu2024real,
title={Real effects of supplying safe private money},
author={Xu, Chenzi and Yang, He},
journal={Journal of Financial Economics}
, year={2024},
volume={157},
pages={103868},
}

Privately issued money often bears default risk, which creates transaction frictions when used as a medium of exchange. The late 19th century US provides a unique context to evaluate the real effects of supplying a new type of money that is safe from default. We measure the local change in "monetary" transaction frictions with a market access approach derived from general equilibrium trade theory. Consistent with theories hypothesizing that lowering transaction frictions benefits the traded and inputs-intensive sectors, we find an increase in traded goods production, in the share of manufacturing output and employment, and in innovation.

@article{coppola2025currency,
title={Currency development through liquidity provision},
author={Coppola, Antonio and Krishnamurthy, Arvind and Xu, Chenzi},
journal={American Economic Association Papers and Proceedings},
year={2025},
volume={115},
}

Drawing on the experiences of the historical Eurodollar market and recent Chinese dollar bond issuances traded outside U.S. jurisdiction at negative spreads to Treasurys, we examine the conditions under which a parallel offshore dollar financial system that circumvents Western sanctions may emerge. We propose a model in which currency use is driven by liquidity provision and safe bond supply. We characterize three equilibrium regimes: high convenience yields emerge in both the initial sanctions-driven region and the final liquidity-driven region, separated by an intermediate region. Transitions between equilibria depend on safe-asset supply and liquidity technologies, in addition to endogenous dynamic complementarities.

@techreport{coppola2023liquidity,
title={Liquidity, debt denomination, and currency dominance},
author={Coppola, Antonio and Krishnamurthy, Arvind and Xu, Chenzi},
year={2023},
type={NBER Working Paper 30984},
}

The international monetary system of the last four centuries has experienced the rise, persistence, and fall of specific currencies as the dominant unit of denomination in global debt contracts. We provide a liquidity-based theory to explain this pattern. Firms issue debt that can be extinguished by trading their revenues for financial assets of the same denomination. When asset markets differ in their liquidity, as modeled via endogenous search frictions, firms optimally choose to denominate their debt in the unit of the asset that is most liquid. Equilibria with a single dominant currency emerge from a positive feedback cycle whereby issuing in the more liquid denomination endogenously raises the benefits of that denomination. This feedback mechanism has historically been seeded by governments that created the largest pool of liquid assets in the same denomination. Once dominance is established, a country's costs of investing in the ability to create liquid assets, such as by increasing fiscal capacity, are lower while the incentives to do so are higher, thereby entrenching dominance. We explain the historical experiences of the Dutch florin, the British pound sterling, the US dollar, and the transitions between them. Our theory highlights normative features of liquidity provision in the international monetary system through the lens of the Bretton Woods arrangement, and we discuss the implications of modern policy tools such as central bank swap lines. We rationalize the current dollar-dominant international financial architecture and provide predictions about the potential rise of the Chinese renminbi.

@article{quincy2025branching,
title={Branching out: Capital mobility and long-run growth},
author={Quincy, Sarah and Xu, Chenzi},
year={2025},
type={NBER Working Paper 34457},
}

We study the long-run effects of the first wave of U.S. banking market integration on capital mobility and manufacturing productivity. Using newly digitized bank and branch balance sheet data matched to state and county panels, we provide direct evidence that branching produced lasting productivity gains without aggregate capital deepening by leveraging internal capital markets to improve the geographic allocation of capital. Our novel "deposit market access" measure shows that bank funding grew most in capital-constrained counties within branching states. Both market access and border discontinuity designs indicate that branching's organizational structure reduced capital allocation frictions to generate persistent growth.

@article{quincy2025big,
title={Regulation in crisis: Bank failures and entry barriers},
author={Quincy, Sarah and Xu, Chenzi},
year={2026},
type={Working Paper},
}

This paper uncovers and rationalizes a significant, previously overlooked shift in U.S. bank entry regulation during the interwar period. Using a newly constructed granular database of state banking laws, we document a striking countercyclical pattern: states systematically tightened entry barriers during the 1920s boom and loosened them during the Great Depression. The post-crisis reversal was driven by the severity of local banking distress: places more exposed to failures were more likely to relax entry restrictions. We argue that the crisis destroyed the incumbent rents that had sustained political support for high barriers, thereby transforming defenders of local banking monopolies into advocates for the entry deregulation they had previously opposed.

@techreport{matray2026indpolicy,
title={Industrial policy and capital misallocation in exporting},
author={Matray, Adrien and Mueller, Karsten and Xu, Chenzi and Kabir, Poorya},
year={2026},
type={NBER Working Paper 32019},
}

Trade financing through Export Credit Agencies is a key tool of modern industrial policy. We provide a theory to evaluate its welfare effects and study its causal impact on trade, firm investment, and capital misallocation by using the effective shutdown of the Export-Import Bank of the United States (EXIM) from 2015--2019 as a natural experiment. First, we show that EXIM has large causal effects: comparing industries exposed and un-exposed to the shutdown, we find that exposed industries experience a product-level export reduction of approximately $4.49 for each $1 lost in EXIM financing. EXIM-dependent firms also experience substantial contractions in revenues, investment, and employment. Second, shutting down EXIM increases capital misallocation because firms with high marginal revenue product of capital (MRPK) disproportionately contract while low-MRPK firms are largely unaffected. Countervailing forces, namely terms-of-trade adjustments and freeing capital for domestic producers, do not appear to offset these welfare losses empirically. Our results indicate that even in advanced economies with developed financial markets, industrial policy that lowers financing constraints for exporters can raise output, improve capital allocation, and generate welfare gains.

@techreport{matray2026geopolitics,
title={The geopolitics of industrial policy: Evidence from export credits},
author={Matray, Adrien and Mueller, Karsten and Xu, Chenzi and Shen, Zekai},
year={2026},
type={Working Paper},
}

Tariffs and sanctions dominate debates over geoeconomic fragmentation, yet governments shape trade just as actively through a large and understudied export-side instrument: officially supported export credits, one of the most widely used tools of modern industrial policy. We construct the first datasets to characterize this market globally, and we provide the following evidence that geopolitical factors play an important role for understanding global financial and trade flows: (i) export credits account for over 60% of developing countries' external debt and often over half of their imports; (ii) geopolitical alignment is a strong predictor of export credit allocation, both within bilateral relationships over time and in response to country risk shocks; (iii) quasi-random rotation into the UN Security Council, which temporarily increases countries' geopolitical importance, leads to an inflow of export credits from permanent members; (iv) in recent years, China has provided more export credits than all OECD countries combined; and (v) the importance of geopolitical factors in export credits among OECD economies has increased over time and is now comparable to that of geopolitical competitors, particularly China. Taken together, these findings establish export credits as a central but overlooked instrument of trade policy, industrial policy, and economic statecraft.

@article{beaumont2026reconciling,
title={Reconciling Treatment Effects Estimates across Levels of Aggregation},
author={Beaumont, Paul and Matray, Adrien and Tu, Yachi and Xu, Chenzi},
year={2026},
type={Working Paper},
}

Empirical researchers often estimate treatment effects at different levels of aggregation to address confounding shocks or to detect violations of the stable unit treatment value assumption. Using eight widely used datasets on trade, employment, syndicated loans, patents, and stock holdings, we document that disaggregated data share common properties that complicate comparisons across aggregation levels. We show that, under standard specifications, these properties can lead treatment effect estimates to differ across levels of aggregation even in the absence of shocks or SUTVA violations. These gaps arise mechanically from implicit weighting across levels of aggregation, entry and exit, and Jensen's inequality. We characterize the conditions under which they become quantitatively important and propose simple approaches that are not sensitive to these sources of bias. Using practical examples, we show how aligning estimates across levels of aggregation can help researchers separately identify confounding shocks, spillovers, and extensive-margin changes at little additional cost.

@techreport{mueller2026longrun,
title = {The Very Long-Run Costs of Financial Crises},
author = {Mueller, Karsten and Xu, Chenzi and Lehbib, Mohamed},
year = {2026},
type = {Working Paper},
}

We provide new empirical evidence of the long-run effects of financial crises on the real economy by examining output dynamics over unprecedented time horizons. To do so, we introduce the Global Macroeconomic Database, which covers 243 countries and extends back to the origins of modern data collection. Using this extended historical coverage, we find that banking crises are associated with statistically significant contractions in real GDP that persist for five decades following the initial shock. These estimated effects are considerably larger than those found in previous studies, which have typically been constrained by shorter time horizons and less comprehensive data. We find a similarly prolonged disruption in both human and physical capital accumulation while fiscal revenues deteriorate, spending increases, and debt accumulates. These disruptions are particularly pronounced during globally synchronized crises and in emerging economies. The effects are consistent with the view that recoveries are slower when alternative balance sheets cannot absorb financial losses, suggesting that financial constraints are a central mechanism. The magnitudes of these persistent output losses imply that the true economic costs of financial instability have been substantially underestimated in the literature.

@article{bilal2025climate,
title={Climate change and financial stability: 1850--2024},
author={Bilal, Adrien and Ilhan, Emir and Kanzig, Diego and Mueller, Karsten and Xu, Chenzi},
year={2025},
type={Working Paper},
}

This paper shows that rising global temperature erodes financial stability. We combine macro-financial data covering more than 180 countries over 150 years with a time series approach based on natural climate variability. Our estimates imply that a permanent 1 degree C increase in global temperature increases the risk of a sovereign debt crisis by 20 p.p. and a banking crisis by 15 p.p. Sovereign debt rises by 55%, debt-to-GDP ratios increase by 22 p.p., and government expenditures soar due to the underlying economic losses. Crisis risk is concentrated in emerging economies. In contrast, a comparable analysis using local temperature shocks detects mostly negligible effects.

@techreport{mueller2025global,
title = {The Global Macro Database: A New Historical Dataset of Macroeconomic Statistics},
author = {Mueller, Karsten and Xu, Chenzi and Lehbib, Mohamed and Chen, Ziliang},
year = {2025},
type = {NBER Working Paper 33714},
}

The Global Macro Database is an open-source, continuously updated dataset of macroeconomic statistics that unifies and extends existing resources. By harmonizing and integrating data from 32 major contemporary sources--including the IMF, World Bank,and OECD--with historical records from 78 additional datasets, we construct comprehensive annual time series for 46 variables across 243 countries. This database covers global macroeconomic trends from the origins of modern data collection to projected estimates for 2030. Using this extensive database, we study the long-run output losses of financial crises and global temperature shocks, two applications in which historical time series are a crucial input. Our findings show that financial crises are associated with statistically detectable contractions in real GDP for five decades into the future, which are considerably larger than previously estimated. Temperature shocks also predict real GDP contractions up to 30 years ahead, especially in emerging economies.

@techreport{jiang2025dollar,
title = {Dollar upheaval: This time is different},
author = {Jiang Zhengyang and Krishnamurthy, Arvind and Lustig, Hanno and Richmond, Robert and Xu, Chenzi},
year = {2025},
type = {Working Paper},
}

What can we learn from the high-frequency responses in bond and currency markets to the recent tariff announcement about the status of the U.S. dollar as the global reserve currency? The dollar depreciated by 6.5% after April 1 in spite of rising U.S. interest rates and market volatility, which is highly unusual. The willingness of foreign investors to pay extra for the safety of dollar safe assets, including but not limited to U.S. Treasury, has declined. These asset market responses suggest that investors started to question the role of the dollar as the reserve currency.

What explains the differences in how often countries default on their external debt and how quickly they re-access external credit markets afterwards? While some countries borrow again within a few years, often only to default again, others remain excluded for decades. In this paper we first document this pattern of "serial" default in the London market during the period from 1820 to 1939 using the frequency, duration, and scale of default. We provide a model of investor learning about sovereigns' likelihood of default based on the publicly observed circumstances of default, which generates predictions for their bond prices and terms of future issuance. Second, we test these hypotheses using a newly created dataset of text-based measures of the circumstances of borrowing and default. These qualitative measures come from the universe of historical financial newspapers published in Britain, and they allow us to more fully capture the information set that was available to investors.

We introduce novel data on the universe of multinational banking activity during the first age of globalization from 1870 1914 and show that these financial connections significantly increased trade volumes. First, we describe the data and show that the distribution of countries `exporting' banks is very skewed: the top four exporters are responsible for almost 80% of all multinational banks. Second, we show that there is a significant positive relationship between a multinational banking connection and exports using a standard gravity framework. We also employ a near-neighbors approach to disentangle the causal effect of bank entry from the possibility that banks simply anticipated exports growth.