Domestic Lending and the Pandemic: How Does Banks' Exposure to Covid-19 Abroad Affect Their Lending in the United States?

Finance and Economics Discussion Series Federal Reserve Board, Washington, D.C. ISSN 1936-2854 (Print) ISSN 2767-3898 (Online) Domestic Lending and the Pandemic: How Does Banks’ Exposure to Covid-19 Abroad Affect Their Lending in the United States? Judit Temesvary and Andrew Wei 2021-056 Please cite this paper as: Temesvary, Judit, and Andrew Wei (2021). “Domestic Lending and the Pandemic: How Does Banks’ Exposure to Covid-19 Abroad Affect Their Lending in the United States?,” Finance and Economics Discussion Series 2021-056r1. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/FEDS.2021.056r1. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

Domestic Lending and the Pandemic: How Does Banks’ Exposure to Covid-19 Abroad Affect Their Lending in the United States? By Judit Temesvary and Andrew Wei1 Board of Governors of the Federal Reserve System Washington, DC 20551 October 2021 Abstract: Shortly after the onset of the pandemic, U.S. banks cut their term lending to businesses–but little is known about how much, and why, banks’ choice to ration credit contributed to this contraction. Afforded by a unique combination of several highly granular bank regulatory datasets, we identify the role of banks’ exposure to Covid-related restrictions abroad – a balance sheet “shock” that affects only banks’ credit supply, but not their US borrowers’ demand for loans. We find that US banks with higher foreign Covid exposure cut their lending to US firms, and tightened terms on such loans, significantly more. Banks having become less risk tolerant, as well as foreign borrowers defaulting and drawing down on their cross-border credit lines, were potent mechanisms through which foreign Covid exposure reduced banks’ domestic lending. Keywords: Cross-border exposure, bank lending, bank capital, bank balance sheet liquidity JEL codes: F34; F65; G15; G21 1 Judit Temesvary and Andrew Wei are both at the Board of Governors of the Federal Reserve System. We thank Bill Basset, Ricardo Correa, Itay Goldstein, Camelia Minoiu, Ann Owen, and workshop participants at the Federal Reserve Board for helpful comments and suggestions. The views expressed are the authors’ and should not be interpreted as representing the views of the Federal Open Market Committee, its principals, the Board of Governors of the Federal Reserve System, or any other person associated with the Federal Reserve System. Corresponding author: Judit Temesvary, judit.temesvary@frb.gov; 202-379-6229.

1. Introduction The onset of the pandemic had profound effects on US banks and the availability of bank credit in the United States. A quickly expanding literature (Kapan and Minoiu, 2021; Li et al, 2020; Chodorow-Reich et al, 2021; Berger et al, 2021) has given us three main messages. First, after the onset of the pandemic, banks continued to serve the liquidity needs of their large corporate borrowers via credit lines (Li et al, 2020; Chodorow-Reich et al, 2021; Kapan and Minoiu, 2021). Second, at the same time banks substantially tightened terms on credit to other corporate borrowers. They did so especially in their term lending (Kapan and Minoiu, 2021) and via shorterterm credit lines, particularly to smaller firms (Chodorow-Reich et al, 2021) and to known borrowers (Berger et al, 2021) – making such firms particularly liquidity-constrained (Chodorow- Reich et al, 2021). Third, reduced lender risk tolerance served as a driver of banks’ choice to cut loans (Kapan and Minoiu, 2021), while they remained amply liquid and capitalized (Li et al, 2020), despite sharp stock declines (Acharya et al, 2021). A very important question that has remained unanswered thus far is: why did banks contract their lending? Specifically, how have the pandemic and Covid-19-related restrictions affected banks’ supply of credit? The fact that this question has remained unanswered thus far is not surprising, given that trying to disentangle the credit supply vs. demand-side effects of the pandemic – or any crisis – is fraught with difficulties. Afforded by the novel combination of several highly granular bank regulatory datasets, we tackle this identification issue by focusing on the role of US banks’ exposure to Covid-related restrictions in foreign countries via their direct crossborder lending (henceforth, foreign Covid exposure) – “shocks” that affect banks’ US balance sheet directly, but not their US-based borrowers. We find that (especially worse-capitalized banks) with heavier foreign Covid exposure cut their domestic lending and tightened lending standards 1

much more. We point to three mechanisms through which higher exposure to economic restrictions has reduced banks’ domestic lending: lower risk tolerance by banks, high drawdowns by foreign borrowers on their cross-border credit lines, and higher foreign defaults. Our uniquely detailed data also allow us to ensure that these results are robust to controls for banks’ “global-ness” and for firms’ Covid exposure in the US, including adding state-level restrictions, bank*firm*maturity /credit rating and industry*quarter fixed effects and cutting by size and industry Covid sensitivity. To our knowledge, our identification strategy and results are unique in the Covid banking literature. More broadly, we contribute to the historical strand of papers that have attempted to delineate credit supply and demand effects during crises. Our contribution this way is largely due to our ability to overcome the high data needs that tackling several related identification issues requires. First, the pandemic (and other crises) and the associated economic restrictions affected borrowers and banks simultaneously (Baltik et al, 2020; Hasan et al, 2021). Therefore, to identify credit supply-side effects, one needs to focus on an exogenous “shock” that affected US banks (and thus their credit supply) only, but not US corporates (and thus their credit demand) – rendering the identification of supply-side drivers in a domestic context infeasible. Second, to successfully identify the supply-side effects of Covid (and other large shocks), one needs loan, or at least bankborrower, level data, to control for changes in demand at the firm level (Khwaja and Mian, 2008). Afforded by the novel combination of three highly granular confidential banking datasets (the loan-level FR Y14, the FFIEC 009 on banks’ foreign exposures, and the Senior Loan Officer Opinion Survey (SLOOS) on bank lending standards and reasons), in this paper we implement a unique identification strategy that effectively meets both these requirements. First, for a “shock” that affects banks only, but not their US borrowers, we focus on banks’ exposure to Covid-related effects abroad, through their (on-balance-sheet) cross-border lending. Arguably, for US banks 2

that lend across borders, these shocks impact balance sheets directly, like how US restrictions do, but without the confounding effects on domestic borrowers. Indeed, starting around the onset of the pandemic, governments in countries to which US banks lent the most imposed strict measures to curb the spread of the disease, closing down large segments of their economies and instituting stay-at-home orders (Hale et al, 2021; Figure A1), leading to higher corporate loan defaults (Hasan et al, 2021) and, as we show, also to higher drawdowns and charge-offs on loans to foreigners. Claims abroad make up around 30 percent of larger US banks’ assets; therefore, these foreign exposures are economically meaningful. Accessing highly detailed regulatory data on individual US banks’ cross-border claims at the bank*country*sector level (from the FFIEC 009 Country Exposure reports), we construct for each bank a foreign Covid exposure measure (our “shock”) as the portfolio-weighted average of Covid-related economic restrictions (from Hale et al, 2021), Covid cases and deaths, and corporate bankruptcies across all foreign countries whose borrowers the bank lends directly to. As an example, we utilize the fact that Covid-related “effects” in Germany – such as economic restrictions there, and the resulting credit line drawdowns by German clients, or losses incurred on direct loans to German borrowers – affect those US banks that have held cross-border claims (e.g. loans) in Germany, but not these banks’ US-based borrowers (such as a small firm in Minnesota) –thus serving as an exogenous balance sheet shock. Our approach has several benefits. By focusing on the domestic (US) lending effects of banks’ foreign exposures that leaves US borrowers unaffected, we can isolate the pandemic’s effects on the supply of credit. In other words, we rely on the geographic separation of the foreign “shock” (a bank’s Covid exposure abroad) and its domestic lending – allowing us to argue that the economic fallout from Covid in foreign countries are highly unlikely to affect the borrowing decisions of firms in the United States. By focusing on exposures via cross-border claims –that 3

are held on the US (parent) bank’s balance sheet directly, similar to how domestic (US) loans are held – we can examine direct balance sheet shocks, as opposed to shocks transmitted indirectly via foreign affiliates. To measure changes in lending to US firms in detail, we study loan-level data from the FR Y14 (the US “credit registry”), on both the intensive and extensive margins. To further aid the identification of credit supply effects, we bring in SLOOS micro data on changes in banks’ lending standards, an established proxy of credit supply changes (Bassett et al, 2014). Second, afforded by the loan-level Y14 data, we include a wide range of extensive and time-varying fixed effects and firm traits (such as credit quality), to account for the confounding credit demand-side effects of the pandemic. Returning to our example above, we are able to explicitly control for Covid-related restrictions in Minnesota, the location of our example borrowing firm – accounting for Covid’s effect on firms’ credit demand. Similarly, our use of bank*firm*maturity and bank*firm*risk rating fixed effects means comparing (foreign Covid exposure-induced) changes in lending within each bank-firm pair, abstracting away from differential effects by relationship type (Berger et al, 2021). The detailed Y14 data also enables us to study effects on both the intensive (lending volumes) and extensive (number of loans) margins. We find that US banks with heavier foreign Covid exposures cut their lending via term loans, and tightened their lending standards, to firms in the United States and lower bank capitalization intensified this effect. The magnitudes are economically significant. A one percentage point increase in a bank’s exposure reduced that bank’s lending and the growth in its number of loans to US firms by 6-7 percentage points –equivalent to a 7.9-billion-dollar decline. Furthermore, the effect of a one percentage point higher foreign Covid exposure is more than twice as large for a low vs. well-capitalized bank (at the 10th and 90th percentiles, respectively). 4

We also dig deeper to uncover the mechanisms through which foreign Covid exposure led banks to cut back on their US lending. Using data from banks’ 009 Country Exposure Reports, we show that foreign borrowers who experienced stricter Covid-related economic restrictions drew down significantly more on the cross-border credit lines that US banks pre-committed to them. This suggests that banks’ foreign commitments via credit lines may have restricted banks’ ability to serve domestic borrowing needs. In addition, bringing in survey micro data on banks’ reasons for tightening lending standards, we find that US banks with heavier foreign Covid exposures also cited reduced risk tolerance more – a factor that Kapan and Minoiu (2021) found to be associated with bigger domestic credit cuts. More exposed banks also cited a deteriorated capital position as a reason for tightening, suggesting that foreign Covid exposure limited credit to US borrowers in part by driving up banks’ risk aversion as they grew concerned about their capital positions. Consistently, banks with heavy foreign Covid exposure saw higher charge-offs on foreign loans. Our results have important policy implications. Specifically, our finding on the domestic credit crunching effect of credit line drawdowns by foreign borrowers highlights the importance of carefully monitoring banks’ commitments abroad. More broadly, our results on the crisisinduced contraction in credit supply suggest that balance sheet shocks can have important spillover effects even when capital and liquidity are abundant, if such shocks make banks more risk averse and concerned about capital, amid reputational concerns. We find strong credit supply effects despite aggressive policy actions globally to address the fallout from Covid (Demirguc-Kunt et al, 2021), suggesting that credit outcomes may have been even worse, absent accommodative policies. The paper proceeds as follows. Section 2 describes our hypotheses and in the context of the related literature. Section 3 presents the econometric methodology and Section 4 details the data. Section 5 presents the empirical results and Section 6 summarizes and concludes the paper. 5

2 Hypothesis development and literature review We consider our work as a primarily “domestic lending” paper; we utilize the cross-border dimension of Covid exposures primarily to ensure the exogeneity of our direct balance sheet “shock”. As such, our study is closest related to the evolving literature on the pandemic’s US lending effects. We hypothesize that US banks restricted their supply of credit to US-based corporate borrowers (both in volume and standards), and banks with heavier exposure to Covid’s economic effects (via cross-border claims abroad) did so substantially more (Hypothesis #1). We base this conjecture on a common result from the growing Covid literature: new corporate bank credit contracted during the pandemic. Specifically, US banks served the liquidity needs of their corporate clients by accommodating drawdowns on existing credit lines (Kapan and Minoiu, 2021; Li et al, 2020), and, as a result, banks cut (and tightened standards on) new term loans (Kapan and Minoiu, 2021), making small firms particularly credit constrained (Chodorow-Reich et at, 2021). We link these documented pandemic-era reductions in the supply of bank credit to disruptions in banks’ willingness/ability to lend (specifically, banks’ foreign Covid exposure). Our work is related with another strand of the banking literature which shows that balance sheet disruptions/shocks can cause banks to rebalance their asset portfolio and ration lending. Several notable papers in this strand have documented national and international spillover effects from asset losses in one region to other areas (Kleimeier et al, 2013 among others). In recent work, Hasan et al (2021) document Covid’s negative effect on global syndicated lending via corporate defaults across regions. Earlier (non-Covid) papers found spillover effects from sovereign downgrades (Schertler and Moch, 2021), nuclear tests ( Khwaia and Mian, 2008), and regional 6

floods (Choudhary and Jain, 2017).2 Our focus on international asset exposures being a source of bank “shocks” relates to Peek and Rosengren (1997) who find that stock market-induced losses at Japanese banks reduced lending by their US affiliates.3 Unlike Peek and Rosengren (1997) and subsequent papers, we focus on direct balance sheet effects resulting from cross-border exposure to foreign shocks, rather than international spillovers into affiliate activities via internal capital markets. We show that US banks with heavier foreign Covid exposures cut their US lending more. Next, we explore the role of balance sheet constraints in propagating the effect of foreign Covid exposure into domestic lending. We assert that banks differ in how much foreign Covid exposure causes them to ration credit to US borrowers. Specifically, we point to capitalization as a determinant of the extent of transmission effects into US lending. We hypothesize that lowercapitalized banks saw stronger lending effects from foreign Covid exposure (Hypothesis #2). Recognizing that due to the decade-long, post-GFC buildup of capital and liquidity and the influx of deposits from fiscal policy interventions, banks went into the crisis with ample liquidity (Li et al, 2020), we base this conjecture on two reasons. First, lower capitalized banks are perceived as “riskier” by external funding markets and thus pay higher borrowing costs (Bernanke and Gertler, 1995; Bernanke et al, 1999; Halvorsen and Jacobsen, 2016). Indeed, in the pandemic, banks with lower capital ratios saw larger increases in their CDS spreads (BCBS, 2021) and larger Covid-19related declines in stock prices (Acharya et al, 2021), implying higher reputational effects for these 2 More broadly, our hypothesized effects of banks’ Covid-19 exposure are consistent with papers on natural disasters (Cortes and Strahan, 2017; Berg and Schrader, 2016; Hosono et al, 2016) and on pandemics (Gong et al, 2020; Houle et al, 2015; Leoni, 2011; Zhang et al, 2020; Lagoarde-Sego and Leoni, 2013). 3 Focusing on the transmission of a liability-side shock into lending, papers have found strong lending effects from funding shocks that Peruvian banks suffered due to the 1998 Russian debt crisis (Schnabl, 2012) and that European branches suffered due to the European sovereign debt crisis (Correa et al, 2021). More broadly, the literature has found that banks’ international exposure brings not only benefits (additional funding sources (Cetorelli and Goldberg. 2012); higher-yield investments (Temesvary, 2014), and shock absorption (Cetorelli and Goldberg, 2011)), but also risks (Frame et al, 2020; Karolyi et al, 2018) and to spillovers from abroad (Brauning and Ivashina, 2018, Hale et al, 2020). 7

banks. Second, despite having ample liquidity and capital and despite regulatory calls to draw on their buffers to lend, large banks were reluctant to lower their risk-weighted capital (Abboud et al, 2021), perhaps due to concerns about an adverse market reaction. We find that higher foreign Covid exposure led to loan cuts especially at lower-capitalized banks. Third, we explore several mechanisms through which foreign Covid exposure affected domestic lending. Using SLOOS micro data, we examine if foreign Covid exposure reduced banks’ risk tolerance (Hypothesis #3a). This conjecture further explores Kapan and Minoiu’s (2021) result that US banks that tightened their corporate loan standards reported reduced risk tolerance as an important reason for doing so. Next, we examine if foreign borrowers who faced economic restrictions drew down more on the cross-border credit lines that US banks precommitted to them (Hypothesis #3b), serving as a mechanism through which foreign Covid exposure led to lending cuts.4 Lastly, we study if foreign economic restrictions causing banks to face higher defaults on their corporate loans (Hypothesis #3c) was a path to domestic lending cuts. The idea is that foreign borrowers who faced stricter economic restrictions not only drew down their credit lines more, but also suffered more bankruptcies – leading to loan and investment losses to banks (Ari et al, 2020; Park and Shin, 2021) and lower credit supply (Serrano, 2021). We find that all three mechanisms were at play in connecting banks’ foreign Covid exposure to credit cuts. 4 We study the effect of foreign Covid exposure on existing bank-firm relationships. The emerging literature on relationship lending and credit supply during Covid is mixed. Hasan et al (2021) show that relationships lower the credit effects of Covid-19, but Berger et al (2021) find that clients with closer banking relationships suffered deeper credit cuts and worse terms. Given our focus on lending in already existing relationships, Berger et al (2021)’s results in the context of our work suggests that the credit cuts may be partly due to banks cutting loans especially to firms with existing relationships – which can be costly to replace (James, 1987; Slovin et al, 1993). Foreign exposures can lead to domestic lending cuts also via large movements in the value of investments (such as the stock market; Zhang, Hu and Ji, 2020; Acharya, Engle and Steffen, 2021). 8

3 Econometric methodology 3.1. Foreign Covid Exposure and Domestic Lending (Hypotheses #1 and #2) Our main explanatory variable is bank i’s foreign Covid-19 exposure in quarter t, denoted by . 𝑋𝑋𝑖𝑖,𝑡𝑡 We take the weighted average of country-specific restrictions proxies (such as the government stringency index) across all country n’s that bank i lends to at time t. 𝑥𝑥𝑛𝑛,𝑡𝑡 1. 𝑁𝑁 𝑋𝑋𝑖𝑖,𝑡𝑡 = ∑𝑛𝑛=1𝛽𝛽𝑖𝑖,𝑛𝑛,𝑡𝑡𝑥𝑥𝑛𝑛,𝑡𝑡 To construct the country-specific weights , we use the fraction of bank i's cross-border claims 𝛽𝛽𝑖𝑖,𝑛𝑛,𝑡𝑡 in country n in quarter t-1 in bank i's total cross-border claims in t-1. 2. 𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝐶𝐶𝑖𝑖,𝑛𝑛,𝑡𝑡−1 𝛽𝛽𝑖𝑖,𝑛𝑛,𝑡𝑡 = ∑ 𝑁𝑁 𝑛𝑛=1𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝐶𝐶𝑖𝑖,𝑛𝑛,𝑡𝑡−1 In our benchmark specification, we estimate (at the bank-firm-loan maturity or bank-firm-credit rating level) , which is the quarterly change in the natural logarithm of total lending 𝛥𝛥ln(Y)𝑖𝑖,𝑗𝑗,𝑡𝑡 volume (or the number of loans). In some estimations, denotes loan rates and spreads. 𝛥𝛥ln(Y)𝑖𝑖,𝑗𝑗,𝑡𝑡 3. 2 � 𝛥𝛥 𝑙𝑙𝑙𝑙(𝑌𝑌)𝑖𝑖,𝑗𝑗,𝑡𝑡,𝑔𝑔=𝛼𝛼1+∑𝑘𝑘=1[𝛼𝛼2,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘+𝛼𝛼3,𝑘𝑘 𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘+𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 × 𝛼𝛼4,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘+ + 𝛼𝛼5,𝑘𝑘 � 𝐹𝐹𝑖𝑖𝐹𝐹𝐹𝐹 � +𝛼𝛼6,𝑘𝑘 � 𝐵𝐵𝐵𝐵𝑙𝑙𝑘𝑘 � � + 𝛼𝛼7,𝑘𝑘 � 𝐹𝐹𝑖𝑖𝐹𝐹𝐹𝐹 � + 𝑗𝑗,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑗𝑗,𝑡𝑡 𝐶𝐶𝐶𝐶𝑙𝑙𝑡𝑡𝐹𝐹𝐶𝐶𝑙𝑙𝐶𝐶 𝐶𝐶𝐶𝐶𝑙𝑙𝑡𝑡𝐹𝐹𝐶𝐶𝑙𝑙𝐶𝐶 𝐶𝐶𝐶𝐶𝑙𝑙𝑡𝑡𝐹𝐹𝐶𝐶𝑙𝑙𝐶𝐶 +𝛼𝛼8,𝑘𝑘 � 𝐵𝐵𝐵𝐵𝑙𝑙𝑘𝑘 � ]+ � 𝐹𝐹𝑖𝑖𝑥𝑥𝐹𝐹𝐹𝐹 � +𝜀𝜀𝑖𝑖,𝑗𝑗,𝑡𝑡,𝑔𝑔 𝐶𝐶𝐶𝐶𝑙𝑙𝑡𝑡𝐹𝐹𝐶𝐶𝑙𝑙𝐶𝐶 𝑖𝑖,𝑡𝑡 𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹𝐸𝐸𝑡𝑡𝐶𝐶 𝑖𝑖,𝑗𝑗,𝑔𝑔 where i, j, and t index banks, firms, and quarters respectively, and g indexes either loan maturity or credit rating category. Firm Controls and Bank Controls are firm and bank-specific balance sheet control variables, respectively, and Fixed Effects has bank, bank*firm, bank*firm*maturity or bank*firm*credit rating fixed effects. We interact each explanatory variable with bank capital ratio C, and we include two lags of all the right-hand-side variables. As per Hypothesis #1, we 9

expect greater foreign Covid exposure to translate into lower US-based lending: . 2 ∑𝑘𝑘=1 𝛼𝛼2,𝑘𝑘 < 0 Hypothesis #2 suggests that this effect is larger for worse-capitalized banks: . 2 ∑𝑘𝑘=1 𝛼𝛼4,𝑘𝑘 > 0 In our study of foreign exposure’ effects on lending standards, our dependent variable is the quarterly change in bank i's standards on lending to large and middle-market firms from the SLOOS micro data, denoted by , where higher values mean easier standards. 𝛥𝛥 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆_𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 4. 𝛥𝛥 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆_𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 = 2 𝐹𝐹𝑖𝑖𝑥𝑥𝐹𝐹𝐹𝐹 𝛽𝛽1 +∑𝑘𝑘=1[𝛽𝛽2,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘 +𝛽𝛽3,𝑘𝑘 𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 +𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 ×𝛽𝛽4,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘]+� � +µ𝑖𝑖,𝑡𝑡 𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹𝐸𝐸𝑡𝑡𝐶𝐶 𝑖𝑖,𝑡𝑡 We conjecture that greater foreign Covid exposure means tighter C&I lending standards: 2 ∑𝑘𝑘=1 (Hypothesis #1), and especially so for worse-capitalized banks: . 2 𝛽𝛽2,𝑘𝑘 < 0 ∑𝑘𝑘=1 𝛽𝛽4,𝑘𝑘 > 0 3.2. Mechanisms (Hypothesis #3) We examine the relationship between bank risk tolerance and foreign Covid exposure by estimating Equation (4) for the set of banks that reported tighter C&I standards. We replace the dependent variable with three reasons banks cited for tightening C&I loan 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆_𝑅𝑅𝑆𝑆𝑅𝑅𝑆𝑆𝑖𝑖,𝑡𝑡 standards: reduced risk tolerance, unfavorable economic outlook, and deteriorated capital position. As higher values mean banks chose the reason as more important, higher foreign exposure translates into reduced risk tolerance, and especially so for lower-capitalized banks (Hypothesis #3a) if the sum of coefficients on and are positive and negative, respectively. 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘 𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 ×𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘 Next, we examine (at the bank-country level) if stricter economic restrictions in 𝑥𝑥𝑛𝑛,𝑡𝑡 country n caused bank i to experience higher cross-border credit line drawdowns on its pre-committed credit lines to borrowers in that country n (Hypothesis #3b), a𝛥𝛥s 𝐶𝐶fo𝑆𝑆ll𝐶𝐶ow𝐶𝐶s𝐶𝐶:𝑆𝑆 𝑖𝑖,𝑛𝑛,𝑡𝑡 10

5. 𝛥𝛥𝐶𝐶𝑆𝑆𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑖𝑖,𝑛𝑛,𝑡𝑡 = 2 𝐹𝐹𝑖𝑖𝑥𝑥𝐹𝐹𝐹𝐹 𝛾𝛾1 +∑𝑘𝑘=1[𝛾𝛾2,𝑘𝑘 𝑥𝑥𝑖𝑖,𝑛𝑛,𝑡𝑡−𝑘𝑘 +𝛾𝛾3,𝑘𝑘 𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 +𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 ×𝛾𝛾4,𝑘𝑘 𝑥𝑥𝑖𝑖,𝑛𝑛,𝑡𝑡−𝑘𝑘]+� � +𝜈𝜈𝑖𝑖,𝑛𝑛,𝑡𝑡 𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹𝐸𝐸𝑡𝑡𝐶𝐶 𝑖𝑖,𝑛𝑛,𝑡𝑡 with bank and bank*country Fixed Effects. By Hypothesis #3b, and . 2 2 ∑𝑘𝑘=1 𝛾𝛾2,𝑘𝑘 > 0 ∑𝑘𝑘=1 𝛾𝛾4,𝑘𝑘 < 0 Lastly, we examine if higher foreign Covid exposure reduces domestic lending by causing the bank to face more defaults on foreign loans (Hypothesis #3c) in two ways. First, we run banklevel regressions in which we relate bank i’s foreign Covid exposure to losses it faces on its 𝑋𝑋𝑖𝑖,𝑡𝑡 foreign corporate loans, denoted by . 𝐶𝐶𝐶𝐶𝐶𝐶𝑅𝑅𝐶𝐶𝐸𝐸_𝑆𝑆𝐹𝐹𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 6. 𝐶𝐶𝐶𝐶𝐶𝐶𝑅𝑅𝐶𝐶𝐸𝐸_𝑆𝑆𝐹𝐹𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 = 2 𝐹𝐹𝑖𝑖𝑥𝑥𝐹𝐹𝐹𝐹 𝛿𝛿1 +∑𝑘𝑘=1[𝛿𝛿2,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘 +𝛿𝛿3,𝑘𝑘 𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 +𝐶𝐶𝑖𝑖,𝑡𝑡−𝑘𝑘 ×𝛿𝛿4,𝑘𝑘 𝑋𝑋𝑖𝑖,𝑡𝑡−𝑘𝑘]+� � +𝜔𝜔𝑖𝑖,𝑡𝑡 𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹𝐸𝐸𝑡𝑡𝐶𝐶 𝑖𝑖,𝑡𝑡 With bank and year-quarter Fixed Effects. By Hypothesis #3c, and . 2 2 ∑𝑘𝑘=1 𝛿𝛿2,𝑘𝑘 > 0 ∑𝑘𝑘=1 𝛿𝛿4,𝑘𝑘 < 0 Second, we examine if country n’s restrictions lead to higher (total or corporate) Bankruptcies there, with Hypothesis #3c implying a positive 𝑥𝑥re 𝑛𝑛 l ,𝑡𝑡 ationship. We then calculate a weighted average exposure to bankruptcies , akin to for each bank and run regressions as in Equation (3). 𝐵𝐵𝑖𝑖,𝑡𝑡 𝑋𝑋𝑖𝑖,𝑡𝑡, By Hypothesis #3c, the coefficients on and are negative and positive, respectively. 𝐵𝐵𝑖𝑖,𝑡𝑡 𝐶𝐶𝑖𝑖,𝑡𝑡 × 𝑋𝑋𝑖𝑖,𝑡𝑡 4 Data 4.1 Measures of US-based corporate lending: Changes in C&I loans and lending standards We measure changes in US-based corporate lending in two ways: via changes in loan volumes and number of loans (at the bank-firm-maturity or bank-firm-credit rating levels) and via changes in 11

C&I lending standards (at the bank level). For the former, we collect data on banks’ US-based loan originations from the FR Y14 database. For the latter, we utilize micro data from the SLOOS. In more detail, the (FR) Y14 database is a highly detailed regulatory database (the closest to a credit registry available for the United States) that provides quarterly data on all corporate loans made by the largest US bank holding companies.5 US banks report loan originations with commitments over 1 million dollars with quarterly frequency, covering about three-fourths of all US commercial and industrial lending. Our sample covers 33 large US banks, for which we have data on loans to 138,975 unique firms. During our sample period, less than 10 percent of firms borrowed from more than one bank in each quarter. For our dependent variables, we focus on the dollar volume and number of US-based loan originations over 2020. We are interested in how a bank’s foreign Covid exposure affects the way in which the intensity of its lending relationships evolves over time. Therefore, to capture the intensity of lending relationships, we aggregate loanlevel data at the bank-firm-loan maturity or bank-firm-credit rating level. To capture the evolution of these relationships, we use as our dependent variables the quarterly changes in the dollar volume and the number of loans, for the given bank-firm-maturity or bank-firm-credit rating bucket. Corporate lending declined in 2020 at a quarterly average rate of 1.5 percent within bank-firmmaturity/credit rating pairs (Table 1). The number of loans issued each quarter was little changed. We measure changes in C&I lending standards using micro (bank-level) data from banks’ quarterly responses to the SLOOS. Specifically, we use banks’ responses to the following question: 5 The respondent panel is comprised US BHCs, US IHCs of foreign banking organizations, and covered SLHCs with $100 billion or more in total consolidated assets, as based on: (i) the average of the firm's total consolidated assets in the four most recent quarters as reported quarterly on the firm's Consolidated Financial Statements for Holding Companies (FR Y9C); or (ii) if the firm has not filed an FR Y9C for each of the most recent four quarters, then the average of the firm's total consolidated assets in the most recent consecutive quarters as reported quarterly on the firm's FR Y9Cs. Participation in reporting is mandatory. For further details, please refer to the reporting form at https://www.federalreserve.gov/apps/reportforms/reportdetail.aspx?sOoYJ+5BzDZGWnsSjRJKDwRxOb5Kb1hL. 12

“Over the past three months, how have your bank’s credit standards for approving applications for C&I loans or credit lines—other than those to be used to finance mergers and acquisitions—to large and middle-market firms changed?”. We focus on standards for large firms, which (defined by SLOOS as those with annual sales over 50 million dollars) make up nearly 90 percent of our sample. Higher values of responses indicate easing standards and lower values show that the respondent bank has tightened C&I standards compared to the prior quarter. On average, banks reported having left their C&I lending standards unchanged from one quarter to the next over 2020, with responses ranging from having tightened standards significantly to having eased them somewhat. From the SLOOS, we also include banks’ responses for select reasons as to “why” they tightened C&I loan standards. On average, banks reported that an uncertain economic outlook is a very important reason for having tightened standards, reduced risk tolerance is a somewhat important reason, and deterioration in their capital position was cited as not an important reason. 4.2 Measures of foreign Covid-19 exposure Our primary proxy for a bank’s foreign Covid exposure is the weighted average of the countryspecific government response Stringency Index from the Oxford COVID-19 Government Response Tracker database (Hale et al., 2021).6 This index incorporates several sub-indices: Measures related to Containment and closure (School closing; Workplace closing; Cancellation of public events; Restrictions on gathering size; Closing of public transport; Stay-at-home requirements; Restrictions on internal movement; Restrictions on international travel) and Health systems (Public information campaign). As such, increasing values of this index over time (Figure A1) testify to increasing government intervention in response to the pandemic, corresponding to 6 The historical series of the data, including the Stringency Index and its subcomponents, are available at: https://github.com/OxCGRT/covid-policy-tracker/raw/master/data/timeseries/OxCGRT_timeseries_all.xlsx 13

more restrictive economic actions. We also use additional measures of the fallout from Covid, in alternative specifications. We examine the number of new Covid-19 cases and new Covid-19related deaths, scaled by population, for the countries a bank holds claims in. Importantly, therefore, these measures are independent of the steps that the US government and states have taken in response to the US Covid epidemic (which we include additional controls for). This separation of foreign exposure and domestic lending effects is a notable identification advantage of our estimation setup. Furthermore, these measures capture interference that can translate into higher credit needs by firms abroad – or even defaults and losses for the bank. Indeed, we later show that across countries, higher values of the Stringency Index indeed translate into higher crossborder credit line drawdowns (measured from the FFIEC 009, Table 7) and ultimately, higher bankruptcies (from OECD Statistics, Table 9A) and charge-offs (from the Y9-C, Table 8). To understand how exposed each bank is to Covid-related restrictions abroad, we need to have a full picture of the extent of individual banks’ foreign activities. For this purpose, we utilize a rarely accessed regulatory database on U.S. banks’ cross-border and foreign affiliate claims from the FFIEC 009 Data Report form.7 This dataset shows claims which, in addition to loans, include bonds, stocks, and guarantees, enabling us to capture a bank’s cross-border exposure via a wide range of foreign investments. Banks report on this supervisory form if they have 30 million dollars or more in claims on residents of foreign countries.8 We construct the bank and country-specific weights (Equation (1)) using claims on both ultimate and immediate counterparty risk bases.9 𝛽𝛽𝑖𝑖,𝑗𝑗,𝑡𝑡 7 For more information on this regulatory reporting form, see https://www.ffiec.gov/forms009_009a.htm. 8 Cross-border claims and foreign affiliate claims are reported separately for each foreign country-bank-time (i.e., year-quarter) combination. In additional specifications, for each bilateral bank-foreign country pair, we use crossborder claims data delineated by target sector of investment (financial sector and non-financial private sector). 9 Lending calculated on an immediate counterparty basis captures the actual amount of claims the bank invests in a foreign country, while lending calculated on an ultimate risk basis is adjusted for transfer of risk exposure. This implies that the ultimate risk amount may differ from the actual (immediate counterparty) amount extended to the host country. The ultimate risk amounts reflect the claims for the repayment of which the given host country is responsible. For 14

The banks in our sample have substantial holdings abroad: In the fourth quarter of 2019, before the onset of the Covid crisis, foreign claims made up 30 percent of the average sample bank’s assets. Not only the scale, but the scope of US banks’ foreign exposure is notable: US banks are well diversified across foreign countries. Any one country sees an average of only 0.9 percent of a US bank’s cross-border portfolio and the average bank holds cross-border claims in as many as 93 countries; only about one-fourth of our observations come from banks that hold claims in 33 or fewer foreign countries. As a result, the weighted average foreign Covid exposure that we construct by combining the FFIEC 009 data (for weights) with the Government Stringency Index (Equation (1)) varies substantially in the cross-section: with a mean of 57 and standard deviation of near 20, the index ranges from 23 (at the 10th percentile) to 69 (at the 90th percentile; Table 1). We differentiate the lending effect of foreign Covid exposure by bank capitalization. In our main specifications, we use banks’ Tier1 capital ratio: a bank’s core capital relative to its riskweighted assets. This key regulatory capital ratio remained high near 13 percent in our sample; the largest US banks were well capitalized on average even during the crisis (Li et al, 2020).10 4.3 Bank and firm-specific control variables In addition to the detailed fixed effects, we include in our specifications a set of measures for balance sheet and financial health at both the bank and firm levels. Total Assets capture the scale of operations.11 Return on Assets is a direct and well-established measure of profitability and is instance, if Country A issues guarantees for the loans that the U.S. banks made to Country B, then Country A’s ultimate risk exposure would exceed the immediate counterparty claims in that country. Similarly, Country B’s reported ultimate risk claims would be less than the immediate counterparty claims the bank acquired there. 10 In alternative specifications, we use the common equity Tier1 (CET1) capital ratio, which excludes preferred shares and non-controlling interests from Tier1 capital. This ratio remained at 12.4 percent of risk-weighted assets in 2020. 11 For borrowing firms, Total assets (firm size) can proxy for international exposure: the extent to which they are exposed to the effects of the economic fallout from foreign governments’ Covid-related restrictions. Hence, in some specifications, we delineate firms by size, examining those below and above the sample median asset size separately. 15

hence a potentially important driver of a bank’s ability to supply credit, and a firm’s need for financing. Bank Leverage Ratio is a measure of a bank’s capital relative to its total assets, and hence proxies the bank’s ability to withstand economic shocks. For firms, this variable captures the extent of a bank’s liabilities relative to its assets, and hence measures vulnerability to shocks. We collect bank-level control variables from a merger-adjusted version of the quarterly Y9C data and firm-level controls from the Y14 dataset. Table 1 shows definitions and summary statistics. In select specifications, we include the Covid government restrictions Stringency Index for the US state in which the borrowing firm is headquartered (from the Oxford COVID-19 database), to control for restriction effects on credit demand. We also add a bank’s share of foreign assets in some specifications, to control for more global banks being more affected by foreign restrictions. 5 Results We structure our results as follows. In Tables 2-5, we show evidence that higher foreign Covid exposure caused banks to cut their US (term) lending and tighten standards on such loans – results that are robust to alternative measures of bank Covid exposure and capital, and to controlling for bank “globalness” and borrowers’ Covid exposure (Tables A1-A8). Then, in Tables 6-9 we explore the mechanisms through which foreign Covid exposure reduced bank lending to US firms. 5.1 Foreign Covid Exposure and Domestic Lending (Hypotheses #1 and #2) 5.1.1 Loan Volumes and Number of Loans We start by examining how a bank’s foreign Covid exposure has affected its corporate lending in the United States. We proxy foreign Covid exposure with each bank’s portfolio-weighted exposure 16

to the economic fallout from government restrictions related to the pandemic in the foreign countries it lends to. We measure lending as quarterly percent changes in the volume of new loans (the intensive margin) and the number of newly originated loans (the extensive margin) at the bank-firm level within a given maturity/credit rating bucket, and as quarterly changes in C&I lending standards (at the bank level). As discussed in Section 2, we expect foreign Covid exposure to reduce US-based lending and tighten standards, and especially so for lower-capitalized banks. In our benchmark specifications shown in Table 2, we focus on changes in banks’ USbased lending flows (Columns 1-5, the intensive margin) and changes in the number of loans (Columns 6-10, the extensive margin) separately, on lending data that is pooled by loan maturity. Panel A shows the foreign Covid exposure measure weighted by a bank’s bilateral cross-border lending to each country on an ultimate risk basis, and Panel B shows results using as weights a bank’s bilateral cross-border claims calculated on an immediate counterparty basis. Table 2 shows consistent evidence that foreign Covid exposure has a negative effect on banks’ US-based lending (first row), and especially so for lower-capitalized banks (second row) on the intensive margin (Columns 1-5), and, consistent with Kapan and Minoiu (2021), on the extensive margin as well (Columns 6-10). The significant negative lending effect prevails as we add more stringent sets of fixed effects, including at the bank (Columns 1 and 6), bank-firm (Columns 2 and 7), and bank-firm-maturity (Columns 3 and 8) levels.12 The lending effects are economically significant. Evaluated at the sample-average capital ratio, a one percentage point higher foreign Covid exposure (measured via Stringency) reduces lending flows and the growth in the number of loans by 6-7 percentage points—equivalent to a 7.9 billion dollar decline in loans. 12 Robustness to the inclusion of maturity fixed effects ensures that our results are not driven by the confounding effects of a bank’s potential reallocation of credit from longer-term to shorter-term lending. 17

Figure 1 plots the marginal effects on the intensive margin (left panel, for Column 3) and the extensive margin (right panel, for Column 8), to illustrate how lower bank capital amplifies the negative lending effects of foreign Covid exposure. A one percentage point higher foreign Covid exposure reduces US lending by more than twice as much for a low-capitalized bank (at the 10th percentile) than for a well-capitalized bank (at the 90th percentile). We delineate the sample into shorter-term (with maturity below one year, Columns 4-5) and longer-term (with maturity over one year, Columns 9-10) loans, motivated by earlier results that the transmission of shocks affects different loan maturities differentially (Black and Rosen, 2008; Temesvary et al, 2018; Morais et al, 2019). The negative effect of foreign Covid exposure operates through longer-term lending, which makes up the vast majority of our sample. Foreign Covid exposure has no significant effect on shorter-term loans – consistent with such shorter-term loans being generally more volatile and driven by idiosyncratic factors. Altogether, we do not find that banks are reallocating funds from longer-term to shorter-term lending in response to Covid. An important issue to address is that banks might cut loans if firms’ borrowing ability worsens due to the US-based effects of the pandemic. In fact, there is evidence of a wave of credit downgrades in the second quarter of 2020 (the “fallen angels”). Or, banks might reallocate credit to higher-rated borrowers. In Table 3 we present results derived from data for 11 distinct credit rating categories. Categorizing by credit rating lets include fixed effects to explicitly control for credit quality on the demand side, and for (changes in) lender risk preference on the supply side. In Table 3, we continue to find strong evidence that foreign Covid exposure reduces US lending (both on the intensive and extensive margins) and more so for lower-capitalized banks, even when we hold credit quality “constant” by including bank*firm*credit rating fixed effects. A one percentage point higher foreign Covid exposure lowers lending and the growth in the number 18

of loans by 5-8 percentage points (a 7.9-billion-dollar lending decline). The effect at the 10th percentile of capital is 2 to 4 times larger than at the 90th percentile. Our results are significant for both speculative-grade (BB or below, Columns 4 and 9) and investment-grade (above BB, Columns 5 and 10) loans and hold on both the intensive (Columns 1-5) and extensive (Columns 6-10) margins. The results are robust to measuring pandemic effects via Cases and Deaths (Table A1) and to using CET1 ratios to proxy funding resilience (Tables A2 and A3). Could it be the scale of a bank’s global activities (and the various risks such exposure brings), rather than its foreign Covid exposure, that made a bank more vulnerable to balance sheet shocks during Covid? This may be the case if globally more active banks were systematically more affected by the pandemic, or if the effect of foreign Covid exposure depends on the extent of banks’ foreign activities. Though our use of bank fixed effects controls for the role of time-invariant bank features such as international openness, in Table A4 we interact each regressor with the share of foreign assets. We continue to find that higher foreign Covid exposure lowers US lending, both on the intensive (Columns 1-4) and extensive (Columns 5-8) margins, even when we bank*firm*maturity and bank*firm*credit rating fixed effects ((Columns 3 and 7, and 4 and 8). Next, we examine whether the spillover effects of foreign Covid-19 exposure differ for US corporate term loans or credit lines (Table 4). Our benchmark results are driven by term lending (Columns 1-5); we find no spillover effects into credit lines (Columns 6-10). In other words, banks continued to serve cash flow needs for existing customers but did not extend loans to new clients. 5.1.2 Loan interest rates and spreads and foreign Covid exposure There is evidence that the Covid crisis affected pricing terms, in addition to bank loan volumes (Berger et al, 2021; Kapan and Minoiu, 2021). Afforded by the rich Y14 dataset, we explore the 19

relationship between banks’ foreign Covid exposure and the levels and spreads of interest rates that banks charge on their newly issued loans to US borrowers (Table A5, Columns 1-4 and Columns 5-8, respectively). If banks with higher foreign Covid exposure tightened loan pricing terms, we should see positive coefficients on Stringency. Negative coefficients on the interaction terms would reflect stronger effects for lower capitalized banks. We do not find a relationship between a bank’s foreign Covid exposure and the interest rate it charges on its new loans. However, in our more stringent specifications, we find that a bank’s greater foreign Covid exposure leads to higher loan spreads, and this effect is larger for lower capitalized banks (Table A5, Columns 7-8). 5.1.3 C&I Lending Standards and foreign Covid-19 exposure In Table 5, we home in on the effect of foreign Covid exposure on changes in banks’ SLOOS C&I lending standards, an established measure of credit supply conditions. Using this survey micro data, we find that banks with heavier foreign Covid exposures tightened C&I standards significantly more (first now) and lower capitalized banks did so even more (second row). 5.2 Accounting for Covid-19’s Effects on Borrowing Firms The pandemic hit economies around the world nearly simultaneously; when foreign governments responded to the pandemic with strict restrictions, most US states also did so. There are two related issues for our identification: (1) Covid-related restrictions imposed in the US might have lowered US firms’ credit demand and (2) foreign Covid restrictions may affect large US firms directly.13 Specifically, the first concern is that restrictions by U.S. states also inflicted losses on U.S. firms operating within their jurisdictions, limiting those firms’ credit demand and their ability to borrow from banks. We address this concern in three ways. First, we explicitly include government 13 Bloom, Fletcher and Yeh (2021) provide evidence of the negative economic impact of Covid on firms in the US. 20

stringency indices calculated for the US state of borrowing firms’ headquarters (Table A6). Even after controlling for state-level economic restrictions in the US, we find that more foreign Covid exposed banks cut their lending more, and especially so for lower capitalized banks. The results hold on the intensive (Columns 1-5) and the extensive (Columns 6-10) margins and when we control for maturities and for credit ratings. State-level restrictions (Firm Stringency) and its interaction with the capital ratio come in insignificantly throughout (third and fourth rows). A second way we examine the confounding effect of firms’ exposure to the pandemic is by separating firms in industries more affected by Covid (such as hotel and retail) from those in less affected industries (Table A7). We run specifications with bank*firm*maturity or bank*firm*credit rating fixed effects for firms in Covid sensitive (Columns 1, 3, 5 and 7) and insensitive (Columns 2, 4, 6 and 8) industries, as defined by Kaplan et al (2020). On both the intensive (Columns 1-4) and extensive (Columns 5-8) margins, we find strong results for Covid insensitive industries also, alleviating concerns about Covid-induced reductions in credit demand. In Table A8, we run specifications with first industry*year:quarter fixed effects, effectively comparing banks with different foreign Covid exposures lending to firms in the same industry and same quarter – the closest approximation of the Khwaja-Mian (2008) identification strategy that we can do. We continue to find a significant negative coefficient on bank foreign Covid exposure. The second concern relating to firms’ exposure to the pandemic is that large, globally active firms can be directly exposed to the same foreign restrictions-related economic fallout, the effect of which we study on banks. To address this concern, in Table A9 we examine borrowers by firm size (Chodorow-Reich et al, 2021), separating our sample into small (below the median sample asset size) and large (above the median size) firms. Our results are significant across firm sizes, alleviating concerns that the effect on borrowing firms of our “shock” might drive our results. 21

5.2 Mechanisms After establishing the robustly negative relationship between foreign Covid exposure and domestic lending, we now turn to disentangling the mechanisms through which this causal effect prevails. First, afforded by our access to bank-level SLOOS responses, we study banks’ reported reasons for having tightened C&I loan standards (Table 6). We examine how foreign Covid exposure affected the extent banks cited a deteriorated capital position (Columns 1-2), an unfavorable/uncertain economic outlook (Columns 3-4) and reduced risk tolerance (Columns 5-6) as reasons for tightening. We find that a heavier foreign Covid exposure is strongly related to banks citing a reduction in risk tolerance as a reason for tightening C&I standards (first row), and especially so for lower capitalized banks (second row). In addition, banks with heavier foreign Covid exposure cited a deterioration in their capital position as a reason for tightening standards (especially the lower capitalized ones; Column 1, first two rows), but also cited that a worsening economic outlook was not an important reason (negative coefficients in the first row of Columns 3-4). Together, these results suggest that heavier foreign Covid exposure caused (especially low capitalized) banks to cut loans in part by making lenders more risk averse, as they grew concerned about their capital positions while remaining unconcerned about the economic outlook. Second, we study if a mechanism through which foreign Covid exposure reduced domestic lending may have been by causing foreign borrowers to draw down on their cross-border credit lines with US banks. We tackle this question by studying the relationship between Covid-related government response Stringency abroad and drawdowns by foreign borrowers on the cross-border credit lines that US bank had committed to them. We are uniquely able to study this relationship at the bank-country level, afforded by the highly detailed FFIEC 009 dataset. Indeed, Table 7 shows that more Stringency in a country leads to higher drawdowns by residents of that country 22

on their credit lines with US banks. A unit increase in Stringency raises credit line drawdowns by 2 to 4 percent (Columns 1-2, first row), a finding that is robust to bank*country fixed effects. We conclude that the strain on balance sheet liquidity that US banks experience from foreign borrowers’ credit line drawdowns may be a channel through which restrictions abroad cause banks to cut domestic lending. Such effects can be potent even when banks have abundant liquidity, if they are reluctant to cut into buffers (Abboud et al, 2021; BCBS, 2021) for reputational concerns. Third, we study if a channel through which banks’ foreign exposure led to their lower US lending is by foreign restrictions abroad causing losses on banks’ books. We do so in two ways. In Table 8, we examine how each bank’s foreign Covid exposure relates to charge-offs on its foreign corporate lending. We find conclusive evidence that a bank’s higher exposure to foreign economic restrictions corresponds to six to ten percent higher charge-offs on its foreign C&I portfolio in the subsequent quarters (first row) and this effect is 0.4 to 0.7 percent larger for lowercapitalized banks (second row). The effects, robust to bank and year:quarter fixed effects, are even larger when we calculate foreign Covid exposure on an immediate counterparty basis (Panel B). Next, we examine the relationship between foreign bankruptcies and domestic lending cuts. First, we run country-level regressions of foreign bankruptcies (from the OECD) and government response Stringency (Table 9, Panel A). We find that a five-unit increase in a country’s Stringency subsequently translates into a 2.5 percentage point higher quarterly growth in total bankruptcies (Column 1) and a one percentage point higher growth in corporate bankruptcies (Column 2). Having established this relationship, next we relate each bank’s weighted-average exposure (via foreign lending) to bankruptcies abroad to changes in their US lending (Table 9, Panel B). Results suggest that a bank’s higher exposure to bankruptcies abroad corresponds to subsequently lower US lending (Column 1, first row), especially for lower capitalized banks (Columns 2-3, second 23

row). Overall, we conclude from Tables 8 and 9 that the balance sheet effects of losses that banks incur on their cross-border lending to borrowers who face strict restrictions at home is a mechanism through which banks’ foreign Covid exposure leads to lower domestic lending to firms. 5.3 Additional specifications 5.3.1 Exposure to OECD vs non-OECD countries Are the spillover effects of foreign Covid exposure stronger from developed countries or economically less developed regions? We study the role of the source region of exposure by calculating two foreign exposure measures for each bank: one that captures its exposure to Covid restrictions in OECD countries, and another one for its exposure in non-OECD countries. In results available by request, we find that the spillover effects we document above reflect banks’ Covid exposure in OECD countries, and there are no spillover effects from non-OECD countries. 5.3.2 Exposure to foreign financial vs non-financial sectors We explore if the spillover effects of a bank’s foreign Covid exposure into its US lending depend on the sector of exposure in foreign countries. Afforded by the rich FFIEC 009 data, we calculate two foreign exposure measures: one that captures its exposure to Covid using weights based on the bank’s bilateral cross-border claims on the financial sector in foreign countries, and another one that captures exposure to foreign non-financial sectors. In results available by request, we find that the spillover effects we document earlier reflect banks’ Covid exposure through both foreign financial and non-financial sectors. The spillover results are consistently significant across the delineation of loans (by maturity or by credit rating) and the intensive and extensive margins. 24

6 Conclusion In this paper, we study US banks’ exposure to the economic fallout from Covid-related economic restrictions has affected their supply to credit to US firms. Afforded by a novel combination of several highly granular banking datasets, we employ an identification strategy in which we focus on large US banks that lend abroad. We study the domestic lending effects of these banks’ exposure to Covid-related economic restrictions abroad through their cross-border lending – a shock we argue affects banks’ credit supply, without affecting borrowing US firms’ credit demand. We show that US banks with higher exposures in foreign regions with stricter Covidrelated restrictions cut their US lending (and tightened lending standards) substantially more, and this effect is particularly strong for lower-capitalized banks. The results are robust to a wide range of controls for borrowing firms’ simultaneous Covid exposure, including explicitly including US state-level Covid restrictions and industry*quarter fixed effects. We also show that banks having become less risk tolerant, as well as foreign borrowers defaulting and drawing down on their crossborder credit lines, were potent mechanisms through which foreign Covid exposure reduced credit. Our results have important policy implications. Specifically, our finding on the domestic credit crunching effect of credit line drawdowns by foreign borrowers highlight the importance of carefully monitoring banks’ commitments abroad. More broadly, our findings on the crisisinduced contraction of credit supply suggest that balance sheet shocks can have important spillover effects even when bank capital and liquidity are abundant, if such shocks make banks more risk averse and concerned about capital, amid reputational concerns. 25

References Alice Abboud, Elizabeth Duncan, Akos Horvath, Diana Iercosan,Bert Loudis, Francis Martinez, Timothy Mooney, Ben Ranish, Ke Wang, Missaka Warusawitharana, and Carlo Wix (2021). “Covid-19 as a stress test: Assessing the bank regulatory framework,” FEDS Working Paper 2021-024. Acharya, Viral, Robert F. Engle and Sascha Steffen (2021). “Why did bank stocks crash during Covid-19?” NBER Working Paper 28559. Ari, Anil, Sophia Chen, and Lev Ratnovski (2020). “COVID-19 and non-performing loans: lessons from past crises,” Available at SSRN. Bartik, Alexander W., Marianne Bertrand, Zoë B. Cullen, Edward L. Glaeser, Michael Luca and Christopher T. Stanton (2020). “How Are Small Businesses Adjusting to COVID-19? Early Evidence from a Survey,” NBER Working Paper 26989. Bassett, William F., Mary Beth Chosak, John C. Driscoll, and Egon Zakrajšek (2014). “Changes in bank lending standards and the macroeconomy,” Journal of Monetary Economics 62, pp. 23-40. Basel Committee on Banking Supervision (2021). “Early lessons from the Covid-19 pandemic on the Basel reforms,” (July). Berg, Gunhild and Jan Schrader (2016). “Access to credit, natural disasters, and relationship lending,” Journal of Financial Intermediation 21(4), pp. 549-568. Berger, Allen N., Christa H.S. Bouwman, Lars Norden, Raluca Roman, Gregory Udell and Teng Wang (2021). “Is a Friend in Need a Friend Indeed? How Relationship Borrowers Fare during the COVID-19 Crisis,” Available at SSRN. Bernanke, B.S. and M. Gertler (1995). “Inside the Black Box: The Credit Channel of Monetary Policy Transmission,” Journal of Economic Perspectives 9, pp. 27‐48. Bernanke, B. S., Mark Gertler and Simon Gilchrist (1996). “The Financial Accelerator and the Flight to Quality,” Review of Economics and Statistics 78(1), pp. 1–15. Black, Lamont K., Rosen, Richard J. (2008). “The Effect of Monetary Policy on the Availability of Credit: How the Credit Channel Works,” Board of Governors of the Federal Reserve System, Washington DC. Bloom, Nicholas, Robert S. Fletcher and Ethan Yeh (2021). The Impact of COVID-19 on US Firms,” NBER Working Paper 28314. Brauning, Falk and Victoria Ivashina (2018). “U.S. Monetary Policy and Emerging Market Credit Cycles,” NBER Working Paper 25185. 26

Cetorelli, Nicola and Linda S. Goldberg (2011). “Global Banks and International Shock Transmission: Evidence from the Crisis,” IMF Economic Review 59, pp. 41‐76. ___ (2012). “Banking globalization and monetary transmission,” The Journal of Finance 67(5), pp. 1811–1843. Chodorow-Reich, Gabriel, Olivier Darmouni, Stephan Luck, and Matthew C. Plosser (forthcoming). “Bank Liquidity Provision across the Firm Size Distribution,” Journal of Financial Economics. Choudhary, M. A., and Anil Jain (2017). “Finance and inequality: The distributional impacts of bank credit rationing,” Federal Reserve IFDP 1211. Correa, Ricardo, Horacio Sapriza and Andrei Zlate (forthcoming). “Wholesale funding runs, global banks’ supply of liquidity insurance, and corporate investment,” Journal of International Economics. Cortés, Kristle Romero and Philip E.Strahan (2017). “Tracing out capital flows: How financially integrated banks respond to natural disasters,” Journal of Financial Economics 125(1), pp. 182-199. Demirgüç-Kunt, Asli, Alvaro Pedraza, and Claudia Ruiz-Ortega (forthcoming). “Banking sector performance during the covid-19 crisis,” Journal of Banking & Finance. Frame, Scott, Atanas Mihov and Leandro Sanz (2020). “Foreign Investment, Regulatory Arbitrage, and the Risk of U.S. Banking Organizations,” Journal of Financial and Quantitative Analysis 55(3), pp. 955-988. Gong, Huiwen, Robert Hassink, Juntao Tan and Dacang Huang (2020). “Regional Resilience in Times of a Pandemic Crisis: The Case of COVID-19 in China,” Journal of Economic and Human Geography 111(3), pp. 497-512. Hale, Galina, Tumer Kapan and Camelia Minoiu (2020). “Shock Transmission Through Cross- Border Bank Lending: Credit and Real Effects,” The Review of Financial Studies 33(10), pp. 4839-4882. Hale, Thomas, Noam Angrist, Rafael Goldshmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily Cameron-Blake, Laura Hallas, Saptarshi Majumdar and Helen Tatlow (2021). “A global panel database of pandemic policies: Oxford COVID-19 Government Response Tracker,” Nature Human Behaviour 5, pp. 529–538. Halvorsen, Jorn and Dag Henning Jacobsen (2016). “The bank lending channel empirically revisited,” Journal of Financial Stability 27, pp. 95-105. Hasan, Iftekhar, Panagiotis Politsidis and Zenu Sharma (2021). “Global syndicated lending during the COVID-19 pandemic,” Journal of Banking and Finance https://doi.org/10.1016/j.jbankfin.2021.106121 27

Houle, Brian, Samuel J Clark, Kathleen Kahn, Stephen Tollman and Alicia Ely Yamin (2015). “The impacts of maternal mortality and cause of death on children’s risk of dying in rural South Africa: evidence from a population-based surveillance study (1992-2013),” Reproductive Health S7. James, Christopher (1987). “Some evidence on the uniqueness of bank loans,” Journal of financial economics 19(2), pp. 217-235. Kapan, Tumer and Camelia Minoiu (2021). “Liquidity Insurance vs. Credit Provision: Evidence from the COVID-19 Crisis,” Available at SSRN. Kaplan, Moll and Violante (2020). “The great lockdown and the big stimulus:Tracing the pandemic possibility frontier for the U.S.” NBER Working Paper 27794. Karolyi, Andrew, John Sedunov and Alvaro Taboada (2018). “Cross-Border Bank Flows and Systemic Risk,” Working paper. Kashyap, Anil.K., and Jeremy C. Stein (2000). “What Do A Million Observations on Banks Say About the Transmission of Monetary Policy?” American Economic Review 90, pp. 407‐ 428. Khwaja, A. I., and A. Mian (2008). “Tracing the impact of bank liquidity shocks: Evidence from an emerging market,” American Economic Review 98(4), pp. 1413-42. Kleimeier, Stefanie, Harald Sander and Sylvia Heuchemer (2013). “Financial crises and crossborder banking: New evidence,” Journal of International Money and Finance 32, pp. 884-915. Lagoarde-Segota, Thomas and Patrick L. Leoni (2013). “Pandemics of the poor and banking stability,” Journal of Banking & Finance 37(11), pp. 4574-4583. Leoni, Patrick L. (2011). “HIV/AIDS and banking stability in developing countries,” Bulletin of Economic Research. Li, Lei, Philip E Strahan and Song Zhang (2020). “Banks as Lenders of First Resort: Evidence from the COVID-19 Crisis,” The Review of Corporate Finance Studies 9(3), pp. 472– 500. Morais, Bernardo, Jose-Luis Peydro, Jessica Roldan-Pena and Claudia Ruiz-Ortega (2019). “The International Bank Lending Channel of Monetary Policy Rates and QE: Credit Supply, Reach-for-Yield, and Real Effects,” The Journal of Finance 74(1), pp. 55-90. Park, Cyn-Young, and Kwanho Shin (forthcoming). “COVID-19, nonperforming loans, and cross-border bank lending,” Journal of Banking & Finance. Peek, Joe, and Eric S. Rosengren (1997). “The international transmission of financial shocks: The case of Japan,” The American Economic Review 87(4), pp. 495-505. 28

Schertler, Andrea, and Nils Moch (2021). “Bank Foreign Assets, Government Support and International Spillover Effects of Sovereign Rating Events on Bank Stock Prices,” Journal of Banking & Finance, pp. 1061-87. Serrano, Antonio Sánchez (2021). “The impact of non-performing loans on bank lending in Europe: an empirical analysi,” The North American Journal of Economics and Finance 55. Slovin, Myron B., Marie E. Sushka, and John A. Polonchek (1993). “The Value of Bank Durability: Borrowers as Bank Stakeholders,” Journal of Finance 48(1), pp. 247-67. Schnabl, Philipp (2012). “The international transmission of bank liquidity shocks: Evidence from an emerging market,” The Journal of Finance 67(3), pp. 897-932. Temesvary, Judit (2014). “The Determinants of U.S. Banks' International Activities,” Journal of Banking and Finance 44, pp. 233-247. Temesvary, Judit, Steven Ongena and Ann Owen (2018). “A global lending channel unplugged? Does US monetary policy affect cross-border and affiliate lending by global US banks?” Journal of International Economics 112, pp. 50-69. Zhang, Dayong, Min Hu, and Qiang Ji (2020). “Financial markets under the global pandemic of COVID-19,” Finance Research Letters 36, pp. 1015-28. 29

Figure 1. Marginal effect of Foreign Covid-19 Exposure on Quarterly Change in Domestic Bank Lending at different Tier1 Capital Ratios

Table 1. Variable Definitions and Summary Statistics. VARIABLES Definition Source N mean SD p10 p25 p50 p75 p90 Dependent variables: Quarterly change in the natural log of total C&I Quarterly Change in the lending over 1 million between a bank and firm in a FR Y-14 428,255 -0.014 0.227 -0.0596 -0.0127 0 0 0.00336 Log of Lending quarter. Quarterly Change in the Quarterly change in the natural log of total number Log of the Number of of C&I loans over 1 million between a firm and bank FR Y-14 428,255 -0.00324 0.139 0 0 0 0 0 Loans in a quarter. Natural log of total C&I lending over 1 million Ln[Lending] FR Y-14 604,647 15.78 1.543 14 14.47 15.42 16.96 18.1 between a bank and firm in some quarter. Natural log of total number of C&I loans over 1 Ln[Number of Loans] FR Y-14 604,647 0.24 0.478 0 0 0 0.693 0.693 million between a firm and bank in a quarter. Individual bank responses from the quarterly Senior SLOOS C&I Standards Loan Officer Opinion Survey, to the question: "Over the past three months, how have your bank's SLOOS 107 2.776 0.756 2 2 3 3 4 Deteriorated Capital Position (1=not important; 50 1.2 0.404 1 1 1 1 2 3=very important reason) SLOOS Reasons for Unfavorable/uncertain Economic Outlook (1=not SLOOS 53 2.698 0.54 2 2 3 3 3 Tightening C&I Standards important; 3=very important reason) Reduced Risk Tolerance (1=not important; 3=very 50 1.82 0.629 1 1 2 2 3 important reason) Foreign Covid-19 exposure measures: Government Stringency An index of government response stringency from Hale et al. 132 57.14 11.66 40.11 49.19 58.01 64.83 72.88 (unweighted) Hale et al. [2020] (2021) An index of government response stringency from Hale et al. Foreign Covid Exposure Hale et al. [2020], weighted by ultimate risk (2021) and 132 56.75 19.38 22.64 43.08 66.44 68.51 71.99 [UR Weighted] exposure. FFIEC 009 An index of government response stringency from Hale et al. Foreign Covid Exposure Hale et al. [2020], weighted by immediate (2021) and 132 57.02 19.48 22.67 43.66 66.53 68.62 72.04 [IC Weighted] counterparty exposure. FFIEC 009 New cases per 1000 individuals in each quarter, Hale et al. Covid-19 Cases [UR averaged across all countries a bank lends to, (2021) and 132 14 13.74 0.578 3.045 8.149 21.13 38.37 Weighted] weighted by ultimate risk exposure. FFIEC 009 New cases per 1000 individuals in each quarter, Hale et al. Covid-19 Cases [IC averaged across all countries a bank lends to, (2021) and 132 14.04 13.7 0.575 2.954 8.15 21.14 38 Weighted] weighted by immediate counterparty exposure. FFIEC 009 New deaths per 1000 individuals in each quarter, Hale et al. Covid-19 Deaths [UR averaged across all countries a bank lends to, (2021) and 132 0.25 0.156 0.0163 0.0868 0.258 0.371 0.434 Weighted] weighted by ultimate risk exposure. FFIEC 009 New deaths per 1000 individuals in each quarter, Hale et al. Covid-19 Deaths [IC averaged across all countries a bank lends to, (2021) and 132 0.251 0.156 0.0163 0.0856 0.26 0.374 0.432 Weighted] weighted by immediate counterparty exposure. FFIEC 009

Table 1 continued. Variable Definitions and Summary Statistics. VARIABLES Definition Source N mean SD p10 p25 p50 p75 p90 Foreign Covid-19 exposure measures (continued): Number of corporate bankrupcties per OECD Bankruptcies (unweighted) 165 89.86 12.13 76.5 80.74 88.9 99.17 107.4 country, indexed to 2007 Statistics OECD Bankruptcies [UR Number of corporate bankrupcties per Statistics and 165 76.2 4.408 70.69 73.12 75.77 79.58 81.53 Weighted] country, weighted by ultimate risk exposure. FFIEC 009 Number of corporate bankrupcties per OECD Covid-19 Deaths [IC country, weighted by immediate counteroarty Statistics and 165 76.09 4.382 70.68 73.06 75.76 79.47 81.62 Weighted] exposure. FFIEC 009 Capitalization measures: Total Tier1 capital of a bank divided by total Tier 1 Capital Ratio FR Y9-C 819 12.83 4.718 10.13 11.01 12.37 13.95 16.66 risk weighted assets. Total common equity Tier1 capital of a bank CET1 Capital Ratio FR Y9-C 786 12.37 3.127 9.608 10.46 11.76 13.18 15.93 divided by total risk weighted assets. Bank-level variables: Drawdowns on Cross- Quarterly change in unused commitments (at FFIEC 009 20,755 -25.53 1,656 -3 0 0 0 1 border Credit Lines the bank-country level) Total Tier1 capital of a bank divided by Bank Leverage Ratio FR Y9-C 819 9.596 1.818 7.799 8.493 9.312 10.28 11.64 consolidated assets. Net income divided by total consolidated Bank ROA FR Y9-C 819 0.198 0.446 0.0405 0.149 0.236 0.317 0.406 assets. Ln[Bank Size] Natural log of bank total assets. FR Y9-C 819 16.74 1.447 15.35 15.66 16.39 17.34 18.94 Total liabilities of a firm divided by total Firm Leverage Ratio FR Y-14 460,318 0.61 0.26 0.232 0.426 0.636 0.811 0.969 assets. Operating income of a firm divided by total Firm ROA FR Y-14 454,583 0.145 0.318 -0.0334 0.022 0.073 0.161 0.337 assets. Ln[Firm Size] Natural log of total assets. FR Y-14 460,552 16.97 2.387 14.51 15.52 16.63 18.12 20.18 Nature log of net charge-offs on foreign C&I Ln[Net Charge-offs] FR Y9-C 12,471 0.933 3.543 0 0 0 0 0 loans Nature log of gross charge-offs on foreign Ln[Gross Charge-offs] 12,471 1.398 5.25 0 0 0 0 0 C&I loans FR Y9-C Share of Foreign Assets Total cross-border claims (aggregated across FFIEC 009 and 224 0.00213 0.000364 0.00183 0.002 0.00202 0.002 0.00269 [UR Weighted] countries) divided by total bank assets. FR 9Y-C Share of Foreign Assets Total cross-border claims (aggregated across FFIEC 009 and 224 0.00213 0.000358 0.00183 0.002 0.00202 0.002 0.00268 [IC Weighted] countries) divided by total bank assets. FR 9Y-C

Table 2. Quarterly Change in Domestic Bank Lending across Firms and Credit Maturities for banks with different Tier1 Capital Ratios. Measure of U.S.-based lending: Quarterly Change in the Log of Lending Quarterly Change in the Log of the Number of Loans Included Maturities All All All ≤ 1 year > 1 year All All All ≤ 1 year > 1 year VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.191*** -0.193*** -0.184*** 0.156 -0.200*** -0.229*** -0.237*** -0.235*** 0.259 -0.247*** [0.0296] [0.0438] [0.0439] [0.506] [0.0444] [0.0236] [0.0358] [0.0360] [0.402] [0.0364] ∑ Foreign Covid Exposure * Capital 0.0104*** 0.0102*** 0.00961*** -0.0159 0.0106*** 0.0127*** 0.0132*** 0.0130*** -0.0133 0.0137*** {t-2 to t-1} [0.00160] [0.00239] [0.00240] [0.0252] [0.00242] [0.00127] [0.00194] [0.00194] [0.0198] [0.00197] ∑ Capital {t-2 to t-1} -1.030*** -0.986*** -1.030*** 2.742 -1.127*** -1.357*** -1.430*** -1.474*** 3.133 -1.577*** [0.217] [0.328] [0.330] [4.070] [0.329] [0.174] [0.268] [0.270] [3.337] [0.267] Observations 144,261 144,261 144,261 5,390 138,871 144,261 144,261 144,261 5,390 138,871 R-squared 0.002 0.483 0.528 0.641 0.516 0.003 0.443 0.473 0.562 0.468 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.185*** -0.184*** -0.177*** 0.209 -0.190*** -0.230*** -0.238*** -0.236*** 0.194 -0.246*** [0.0277] [0.0410] [0.0412] [0.426] [0.0417] [0.0219] [0.0332] [0.0334] [0.319] [0.0337] ∑ Foreign Covid Exposure * Capital 0.0103*** 0.00996*** 0.00947*** -0.0188 0.0103*** 0.0129*** 0.0134*** 0.0133*** -0.0106 0.0139*** {t-2 to t-1} [0.00154] [0.00230] [0.00232] [0.0231] [0.00235] [0.00122] [0.00185] [0.00187] [0.0167] [0.00188] ∑ Capital {t-2 to t-1} -0.945*** -0.894*** -0.956*** 2.816 -1.037*** -1.300*** -1.367*** -1.417*** 2.617 -1.514*** [0.209] [0.318] [0.319] [3.473] [0.319] [0.165] [0.256] [0.258] [2.702] [0.255] Observations 144,261 144,261 144,261 5,390 138,871 144,261 144,261 144,261 5,390 138,871 R-squared 0.002 0.483 0.528 0.641 0.516 0.003 0.443 0.473 0.562 0.468 Year-Quarter FE X X X X X X X X X X Bank FE X X Bank-Firm FE X X X X X X Bank-Firm-Maturity FE X X Notes: In Columns 1-5,thedependent variableis quarterly change in thenaturallogarithmofU.S.banks'domestic lendingacross firmsand loanmaturities [i.e.loan witha maturity less than oneyearand loan with amaturity more than oneyear]. In Columns 6-10,the dependent variable is the quarterly change in the naturallogarithmofthe numberofU.S.banks'domesticloans across firms and loan maturities.Allspecifications include thefollowing controls at thebank and firmlevelfor quarters t-2 and t-1: Ln[Total Assets], Return on Asset, and Leverage Ratio. Robust Standard errors (clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table 3. Quarterly Change in Domestic Bank Lending across Firms and Credit Ratings for banks with different Tier1 Capital Ratios. Measure of U.S.-based lending: Quarterly Change in the Log of Lending Quarterly Change in the Log of the Number of Loans Included Maturities All All All ≤ BB > BB All All All ≤ BB > BB VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.184*** -0.181*** -0.169*** -0.165** -0.226*** -0.205*** -0.201*** -0.196*** -0.225*** -0.171** [0.0304] [0.0494] [0.0492] [0.0712] [0.0821] [0.0235] [0.0396] [0.0396] [0.0554] [0.0688] ∑ Foreign Covid Exposure * Capital 0.0102*** 0.00970*** 0.00897*** 0.00959** 0.0114*** 0.0117*** 0.0115*** 0.0112*** 0.0129*** 0.0100*** {t-2 to t-1} [0.00166] [0.00275] [0.00273] [0.00394] [0.00437] [0.00128] [0.00217] [0.00216] [0.00303] [0.00366] ∑ Capital {t-2 to t-1} -1.048*** -1.071*** -0.898** -0.956* -1.242** -1.186*** -1.205*** -1.122*** -1.244*** -1.094** [0.217] [0.372] [0.365] [0.510] [0.572] [0.171] [0.297] [0.294] [0.399] [0.487] Observations 143,596 143,596 143,596 103,033 40,563 143,596 143,596 143,596 103,033 40,563 R-squared 0.002 0.531 0.557 0.545 0.534 0.003 0.489 0.513 0.502 0.502 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.174*** -0.167*** -0.157*** -0.156** -0.206*** -0.205*** -0.203*** -0.197*** -0.230*** -0.170*** [0.0288] [0.0467] [0.0464] [0.0679] [0.0758] [0.0223] [0.0373] [0.0372] [0.0523] [0.0633] ∑ Foreign Covid Exposure * Capital 0.00993*** 0.00922*** 0.00854*** 0.00933** 0.0106*** 0.0118*** 0.0117*** 0.0114*** 0.0133*** 0.0100*** {t-2 to t-1} [0.00161] [0.00265] [0.00262] [0.00383] [0.00410] [0.00125] [0.00211] [0.00209] [0.00294] [0.00344] ∑ Capital {t-2 to t-1} -0.955*** -0.962*** -0.794** -0.872* -1.064** -1.132*** -1.150*** -1.068*** -1.240*** -1.001** [0.210] [0.361] [0.354] [0.500] [0.536] [0.165] [0.285] [0.283] [0.386] [0.459] Observations 143,596 143,596 143,596 103,033 40,563 143,596 143,596 143,596 103,033 40,563 R-squared 0.002 0.531 0.557 0.545 0.534 0.003 0.49 0.513 0.503 0.502 Year-Quarter FE X X X X X X X X X X Bank FE X X Bank-Firm FE X X X X X X Bank-Firm-Credit Rating FE X X Notes:InColumns 1-5,thedependentvariableis quarterlychangeinthenaturallogarithmofU.S.banks'domestic lendingacross firms andcreditratings[i.e.AAA,AA,A,BBB,BB,B, CCC,CC,C,D,NotRated].InColumns 6-10,thedependentvariableis thequarterlychangeinthenaturallogarithmofthe numberof U.S.banks' domesticloans across firms andcredit ratings.Allspecifications includethefollowingcontrols atthebankandfirmlevelforquarters t-2andt-1:Ln[TotalAssets],ReturnonAsset,andLeverageRatio.RobustStandarderrors (clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table 4. Quarterly Change in Domestic Bank Lending across Firms and Credit Maturities for banks with different Tier1 Capital Ratios. Depemdent Variable Quarterly Change in the Log of Lending Quarterly Change in the Log of Lending Type of Credit: Term Loans Credit Lines Included Maturities All All All ≤ 1 year > 1 year All All All ≤ 1 year > 1 year VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.224*** -0.206** -0.203** 7.738 -0.199** -0.0570* -0.044 -0.0402 -0.201 -0.0407 (0.06) (0.09) (0.09) (20.79) (0.09) (0.03) (0.04) (0.04) (0.64) (0.04) ∑ Foreign Covid Exposure * Capital 0.0131*** 0.0117** 0.0117** -0.361 0.0112* 0.00311** 0.00165 0.0014 0.00978 0.00156 {t-2 to t-1} (0.00) (0.01) (0.01) (0.97) (0.01) (0.00) (0.00) (0.00) (0.03) (0.00) ∑ Capital {t-2 to t-1} -1.631*** -1.681** -1.670** 76.67 -1.696** -0.309 -0.158 -0.124 0.765 -0.0858 (0.43) (0.69) (0.69) (196.80) (0.69) (0.22) (0.32) (0.32) (2.99) (0.32) Observations 56,322 56,322 56,322 783 55,539 88,805 88,805 88,805 3,737 85,068 R-squared 0.01 0.57 0.59 0.56 0.60 0.00 0.48 0.49 0.65 0.49 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.245*** -0.212** -0.210** 3.97 -0.208** -0.0515* -0.0367 -0.0337 -0.182 -0.035 (0.07) (0.11) (0.11) (9.62) (0.11) (0.03) (0.04) (0.04) (0.47) (0.04) ∑ Foreign Covid Exposure * Capital 0.0142*** 0.0120* 0.0120* -0.182 0.0117* 0.00286* 0.00135 0.00115 0.009 0.00133 {t-2 to t-1} (0.00) (0.01) (0.01) (0.47) (0.01) (0.00) (0.00) (0.00) (0.02) (0.00) ∑ Capital {t-2 to t-1} -1.714*** -1.687** -1.687** 53.92 -1.735** -0.265 -0.113 -0.084 0.92 -0.0492 (0.46) (0.75) (0.75) (116.10) (0.76) (0.21) (0.30) (0.31) (2.78) (0.30) Observations 56,322 56,322 56,322 783 55,539 88,805 88,805 88,805 3,737 85,068 R-squared 0.01 0.57 0.59 0.57 0.60 0.00 0.48 0.49 0.65 0.49 Year-Quarter FE X X X X X X X X X X Bank FE X X Bank-Firm FE X X X X X X Bank-Firm-Maturity FE X X Notes:In Columns 1-5,the dependent variable is quarterly change in the naturallogarithmofU.S.banks'domestic term lending across firms and loan maturities [i.e.loan with amaturity less than oneyearandloan witha maturitymore thanone year].In Columns 6-10,thedependent variable is the quarterly change in the naturallogarithmofU.S.banks'domestic credit line commitments across firms and loan maturities.All specifications include the following controls at the bank and firm level for quarters t-2 and t-1: Ln[Total Assets], Return on Asset, and Leverage Ratio. Robust Standard errors [clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table 5. Quarterly Change in Banks' Lending Standards for Domestic Corporate Loans, for banks with different Capital Ratios. Measure of U.S.-based lending: C&I Lending Standards VARIABLES [1] [2] [3] [4] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.460** -1.138*** -0.241* -0.594** (0.081) (0.291) (0.076) (0.242) ∑ Foreign Covid Exposure * Capital 0.0256** 0.0627*** 0.0144 0.0352** {t-2 to t-1} (0.006) (0.018) (0.006) (0.017) ∑ Capital {t-2 to t-1} -1.726** -4.233*** -0.981 -2.400** (0.370) (1.224) (0.387) (1.135) Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.458** -1.130*** -0.231* -0.565*** (0.080) (0.269) (0.069) (0.214) ∑ Foreign Covid Exposure * Capital 0.0252** 0.0616*** 0.0134 0.0322** {t-2 to t-1} (0.005) (0.017) (0.005) (0.015) ∑ Capital {t-2 to t-1} -1.700** -4.162*** -0.913 -2.207** (0.358) (1.121) (0.350) (0.990) Observations 75 75 75 75 Year-Quarter FE X X X X Capital Ratio Tier1 Tier1 CET1 CET1 Estimation Method OLS probit OLS probit Notes: The dependent variable is individualbanks' responses to the question on the SeniorLoan Officer Opinion Survey, as follows: "Over the past three months, how have your bank's credit standards forapproving applications forC&Iloans orcredit lines—otherthan those to be used to finance mergers and acquisitions—to large and middle-market firms changed?",with highervalues indicating easing standards and lower values indicating tightening lending standards. Robust Standard errors (clustered at the quarter level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table 6. SLOOS Reasons for Banks' Tightening of Corporate Lending Standards, for banks with different Capital Ratios. Reason for Tightening Corporate Lending Deteriorated Capital Unfavorable/Uncertain Reduced Risk Tolerance Standards: Position Economic Outlook VARIABLES [1] [2] [3] [4] [5] [6] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} 1.518* 1.124 -2.564* -2.849* 1.498** 1.334 (0.468) (0.583) (0.717) (0.962) (0.211) (0.889) ∑ Foreign Covid Exposure * Capital -0.101** -0.0842 0.191** 0.207* -0.0939* -0.103 {t-2 to t-1} (0.023) (0.033) (0.035) (0.059) (0.024) (0.060) ∑ Capital {t-2 to t-1} 6.222** 5.157 -13.59** -14.16* 4.903 5.78 (1.382) (2.015) (2.404) (3.969) (2.092) (4.187) Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} 1.547* 1.113 -2.909** -2.851* 1.607* 1.475 (0.402) (0.479) (0.625) (0.914) (0.504) (0.928) ∑ Foreign Covid Exposure * Capital -0.104** -0.0826* 0.218** 0.207* -0.107 -0.12 {t-2 to t-1} (0.019) (0.026) (0.038) (0.059) (0.047) (0.064) ∑ Capital {t-2 to t-1} 6.387** 5.016 -15.40** -14.21* 5.911 7.013 (1.224) (1.722) (2.772) (4.129) (3.764) (4.500) Observations 28 28 28 28 28 28 Year-Quarter FE X X X X X X Capital Ratio Tier1 CET1 Tier1 CET1 Tier1 CET1 Notes: The dependent variable is individualbanks' responses to the question on the SeniorLoan OfficerOpinion Survey,as follows:"fyourbank has tightened oreased its credit standards or its terms forC&I loans or credit lines overthe past three months, how important have been the following possible reasons for the change?", with higher values indicating the given reason being more important and lowervalues indicating the reason being less important or unimportant.The sample includes responses from the subset of banks that reported tightening corporate lending standards, as analyzed in Table 5. Robust Standard errors (clustered at the quarter level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table 7: Government Response Stringency and Credit Line Drawdowns by Foreign Borrowers. Quarterly Drawdowns by Foreign Borrowers on their Cross-border Credit Lines at US Banks VARIABLES [1] [2] 2.294** 3.906*** Country-specific Stringency Index t-1 (0.990) (1.376) Observations 12,224 12,224 R-squared 0.002 0.088 Bank FE X -- Bank-Country FE X Note: The table shows bank-country-level regressions for the relationship between drawdowns by foreign borrowers on their credit lines at US banks and country-specific responses to the economic fallout resulting fromsovereigns'actions toprevent thespread of Covid-19, as captured by the Government Stringency Index. Standard errors in parentheses. "--" means that the given set of fixed effects is included within the more restrictive set of a fixed effects shown below. *** p<0.01, ** p<0.05, * p<0.1

Table 8. Charge-offs on Foreign C&I Loans, for banks with different Capital Ratios. Log of Gross Charge-offs Log of Net Charge-offs on Foreign C&I Loans on Foreign C&I Loans VARIABLES [1] [2] [3] [4] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} 10.39** 8.082* 6.482* 4.737 (4.776) (4.514) (3.207) (2.997) ∑ Foreign Covid Exposure * Capital -0.730** -0.718** -0.472** -0.444* {t-2 to t-1} (0.299) (0.341) (0.202) (0.226) ∑ Capital {t-2 to t-1} 37.5 15.33 26.26 8.228 (54.830) (51.410) (38.190) (36.950) Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} 13.97** 10.33* 8.826** 6.204* (5.280) (5.125) (3.567) (3.372) ∑ Foreign Covid Exposure * Capital -0.919** -0.870** -0.596** -0.544** {t-2 to t-1} (0.334) (0.396) (0.228) (0.261) ∑ Capital {t-2 to t-1} 34.01 20.36 24.14 11.66 (54.030) (52.770) (37.760) (38.100) Observations 81 81 81 81 Year-Quarter FE X X X X Bank FE X X X X Capital Ratio Tier1 CET1 Tier1 CET1 Notes: Thedependentvariableis individualbanks'charge-offs on foreignC&Iloans,fromthemergeradjusted version of the Y9-C. Columns 1-2 show results for gross charge-offs on such loans,and Columns 3-4showresults fornetcharge-offs.RobustStandard errors are inparentheses ***p<0.01, ** p<0.05, * p<0.1.

Table 9 Panel A: Government Response Stringency and Bankruptcies in Foreign Countries %Δ Corporate Dependent Variable: %Δ Total Bankruptcies Bankruptcies [1] [2] Country-specific Stringency 0.513** 0.118* Index t-1 [0.213] [0.0621] Constant -29.22** -11.58*** [11.06] [3.249] Observations 36 39 R-squared 0.145 0.089 Note: Thetableshows country-levelregressions ofthechangein totalbankruptcies (Column 1)and in corporate bankruptcies (Column 2)in response to the economic fallout resulting from sovereigns' actions to prevent the spread of Covid-19, as captured by the Government Stringency Index.Standard errors in parentheses.*** p<0.01, ** p<0.05, * p<0.1

Table 9 Panel B. Quarterly Change in Domestic Bank Lending across Firms and Credit Ratings for banks with different Tier1 Capital Ratios - Role of Exposure to Foreign Bankruptcies. Quarterly Change in the Log of Quarterly Change in the Log of Measure of U.S.-based lending: Lending the Number of Loans VARIABLES [1] [2] [3] [4] Panel A: Ultimate Risk Weighted ∑ Exposure to Foreign Bankruptcies {t-2 to t-1} -0.00832* -0.0013 -0.000794 -0.000454 (0.005) (0.001) (0.003) (0.001) ∑ Exposure to Foreign Bankruptcies * Capital 0.000388 0.000159** 0.000276* 6.43E-05 {t-2 to t-1} (0.000) (0.000) (0.000) (0.000) ∑ Capital {t-2 to t-1} -0.0975** -0.0749 -0.0042 -0.00175 (0.046) (0.050) (0.032) (0.035) Observations 214,464 214,464 214,464 214,464 R-squared 0.002 0.002 0.002 0.002 Panel B: Immediate Counterparty Weighted ∑ Exposure to Foreign Bankruptcies {t-2 to t-1} -0.00537 -0.00138 0.000769 -0.00143** (0.004) (0.001) (0.003) (0.001) ∑ Exposure to Foreign Bankruptcies * Capital 0.000196 0.000161** 0.000107 0.000150*** {t-2 to t-1} (0.000) (0.000) (0.000) (0.000) ∑ Capital {t-2 to t-1} -0.0827* -0.0669 -0.00878 -0.00786 (0.045) (0.047) (0.031) (0.033) Observations 214,464 214,464 214,464 214,464 R-squared 0.002 0.002 0.002 0.002 Bank FE X X X X Type of Bankruptcy Corporate Total Corporate Total Notes:In Columns 1-2,thedependent variableis quarterlychange inthe naturallogarithmofU.S.banks'domestic lending across firms and credit ratings [i.e.AAA,AA,A,BBB,BB,B,CCC, CC,C,D,Not Rated].In Columns 3-4, thedependentvariableis thequarterlychangeinthe naturallogarithmofthe numberofU.S.banks'domesticloans across firms and credit ratings.Exposure to Foreign Bankruptcies is the weighted average ofbankruptcies in the countries the bank lends to, where the weights are each country's share in the bank's foreign portfolio.Robust Standard errors (clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Figure A1. The government response Stringency Index over time in countries that U.S. banks have the highest exposure to.

Table A1. Quarterly Change in Domestic Bank Lending across Firms, Credit Maturities and Credit Ratings, for banks with different Tier1 Capital Ratios - using Cases and Deaths as Foreign Covid-19 Exposure Measure. Foreign Covid-19 exposure measure: Cases Deaths Quarterly Change in Log of Quarterly Change in Log of Quarterly Change in Log of Quarterly Change in Log of Measure of U.S.-based lending: Lending Volume Number of Loans Lending Volume Number of Loans VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Exposure {t-2 to t-1} -0.0356* -0.0408* -0.0429*** -0.0395*** -14.36*** -12.17** -19.67*** -16.15*** [0.0188] [0.0211] [0.0129] [0.0143] [4.279] [4.786] [3.291] [3.620] ∑ Exposure * Capital 0.00189*** 0.00174** 0.00165*** 0.00143** 0.892*** 0.761*** 1.242*** 1.033*** {t-2 to t-1} [0.000720] [0.000810] [0.000527] [0.000583] [0.258] [0.289] [0.201] [0.222] ∑ Capital {t-2 to t-1} -0.721** -0.511* -0.992*** -0.568** -0.203 -0.0957 -0.358** -0.155 [0.286] [0.307] [0.224] [0.232] [0.230] [0.260] [0.170] [0.190] Observations 144,261 143,596 144,261 143,596 144,261 143,596 144,261 143,596 R-squared 0.528 0.557 0.472 0.513 0.527 0.557 0.472 0.513 Panel B: Immediate Counterparty Weighted ∑ Exposure {t-2 to t-1} -0.0238 -0.0335 -0.0287** -0.0312** -8.248*** -8.131** -11.27*** -10.37*** [0.0185] [0.0210] [0.0125] [0.0143] [3.075] [3.695] [2.479] [2.832] ∑ Exposure * Capital 0.00200*** 0.00196** 0.00193*** 0.00176*** 0.535*** 0.529** 0.755*** 0.700*** {t-2 to t-1} [0.000762] [0.000875] [0.000560] [0.000631] [0.191] [0.230] [0.155] [0.178] ∑ Capital {t-2 to t-1} -0.779** -0.636* -1.128*** -0.764*** -0.387 -0.335 -0.688*** -0.519** [0.324] [0.363] [0.255] [0.278] [0.275] [0.315] [0.216] [0.243] Observations 144,261 143,596 144,261 143,596 144,261 143,596 144,261 143,596 R-squared 0.527 0.557 0.472 0.513 0.527 0.557 0.472 0.513 Year-Quarter FE X X X X X X X X Bank-Firm-Maturity FE X X X X Bank-Firm-Credit Rating FE X X X X Notes: InColumns 1-2and5-6,thedependentvariableis quarterlychangeinthenaturallogarithmofU.S.banks'domestic lendingacross firms.In Columns 3-4and7- 8,thedependentvariableis thequarterlychangeinthenaturallogarithmofthenumberofU.S.banks'domesticloans across firms andcredit ratings.In Columns 1,3,5 and7,thedependentvariableis pooledacross loanmaturities [i.e.loanwithamaturityless thanoneyearandloanwithamaturitymorethanoneyear],and inColumns 2,4,6and 8it is pooled across credit ratings [i.e.AAA,AA,A,BBB,BB,B,CCC,CC,C,D,Not Rated].Allspecifications include thefollowing controls at thebank and firmlevelforquarters t-2and t-1:Ln[TotalAssets],Return on Asset,and Leverage Ratio.Allstandard errors clustered at the bank-firmlevel. Robust Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A2. Quarterly Change in Domestic Bank Lending across Firms and Credit Maturities for banks with different Common Equity Tier1 Capital Ratios. Quarterly Change in the Log of Lending Quarterly Change in the Log of the Number of Loans Included Maturities All All All ≤ 1 year ≥ 1 year All All All ≤ 1 year ≥ 1 year VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.0996*** -0.102*** -0.0901*** -0.307 -0.0946*** -0.121*** -0.126*** -0.121*** 0.0278 -0.126*** [0.0230] [0.0339] [0.0335] [0.576] [0.0334] [0.0159] [0.0240] [0.0241] [0.262] [0.0245] ∑ Foreign Covid Exposure * Capital 0.00632*** 0.00603*** 0.00538*** 0.00654 0.00585*** 0.00865*** 0.00890*** 0.00864*** -0.00171 0.00899*** {t-2 to t-1} [0.00140] [0.00208] [0.00207] [0.0275] [0.00210] [0.000984] [0.00148] [0.00149] [0.0134] [0.00152] ∑ Capital {t-2 to t-1} -0.186 -0.0619 -0.101 -0.74 -0.138 -0.0529 -0.0224 -0.0601 1.985 -0.111 [0.155] [0.240] [0.238] [4.671] [0.233] [0.107] [0.167] [0.168] [2.305] [0.167] Observations 144,261 144,261 144,261 5,390 138,871 144,261 144,261 144,261 5,390 138,871 R-squared 0.001 0.482 0.527 0.641 0.516 0.003 0.442 0.472 0.562 0.467 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.0832*** -0.0825*** -0.0730** -0.281 -0.0735** -0.112*** -0.116*** -0.113*** -0.0173 -0.115*** [0.0215] [0.0315] [0.0312] [0.503] [0.0312] [0.0144] [0.0216] [0.0217] [0.207] [0.0220] ∑ Foreign Covid Exposure * Capital 0.00581*** 0.00538*** 0.00484** 0.00408 0.00513** 0.00862*** 0.00885*** 0.00863*** 0.000161 0.00890*** {t-2 to t-1} [0.00138] [0.00205] [0.00206] [0.0234] [0.00209] [0.000958] [0.00144] [0.00145] [0.0113] [0.00149] ∑ Capital {t-2 to t-1} -0.147 -0.0207 -0.0691 -0.723 -0.106 -0.0231 0.0077 -0.0339 1.795 -0.0882 [0.155] [0.240] [0.238] [4.444] [0.234] [0.106] [0.166] [0.168] [2.037] [0.166] Observations 144,261 144,261 144,261 5,390 138,871 144,261 144,261 144,261 5,390 138,871 R-squared 0.001 0.482 0.527 0.641 0.516 0.003 0.442 0.472 0.562 0.467 Year-Quarter FE X X X X X X X X X X Bank FE X X Bank-Firm FE X X X X X X Bank-Firm-Maturity FE X X Notes:In Columns 1-5,thedependentvariableis quarterly changein thenaturallogarithmofU.S.banks'domestic lending across firmsand loanmaturities [i.e.loan withamaturityless thanoneyearandloanwithamaturitymorethanoneyear].In Columns6-10,thedependentvariableis thequarterly changein thenaturallogarithmofthenumberofU.S.banks'domestic loansacrossfirmsandloanmaturities.Allspecifications includethefollowingcontrols atthebankand firmlevelforquarters t-2and t-1:Ln[TotalAssets],Return onAsset,andLeverage Ratio. Robust Standard errors (clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A3. Quarterly Change in Domestic Bank Lending across Firms and Credit Ratings for banks with different Common Equity Tier1 Capital Ratios. Quarterly Change in the Log of Lending Quarterly Change in the Log of the Number of Loans Included Maturities All All All ≤ BB > BB All All All ≤ BB > BB VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.0789*** -0.0819** -0.0873** -0.0574 -0.154** -0.103*** -0.102*** -0.105*** -0.118*** -0.0925 [0.0236] [0.0393] [0.0391] [0.0494] [0.0782] [0.0165] [0.0281] [0.0280] [0.0375] [0.0595] ∑ Foreign Covid Exposure * Capital 0.00505*** 0.00462* 0.00469** 0.00414 0.00815* 0.00753*** 0.00749*** 0.00757*** 0.00916*** 0.00674** {t-2 to t-1} [0.00144] [0.00238] [0.00236] [0.00315] [0.00436] [0.00104] [0.00175] [0.00173] [0.00241] [0.00336] ∑ Capital {t-2 to t-1} -0.236 -0.246 -0.246 -0.305 -0.448 -0.0741 -0.113 -0.117 -0.162 0.00546 [0.151] [0.268] [0.264] [0.351] [0.475] [0.106] [0.187] [0.185] [0.244] [0.337] Observations 143,596 143,596 143,596 103,033 40,563 143,596 143,596 143,596 103,033 40,563 R-squared 0.002 0.53 0.556 0.545 0.533 0.002 0.489 0.513 0.502 0.501 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.0635*** -0.0634* -0.0684* -0.0439 -0.118 -0.0954*** -0.0944*** -0.0971*** -0.112*** -0.0733 [0.0220] [0.0367] [0.0365] [0.0461] [0.0746] [0.0153] [0.0260] [0.0258] [0.0331] [0.0538] ∑ Foreign Covid Exposure * Capital 0.00456*** 0.00396* 0.00398* 0.00383 0.00641 0.00751*** 0.00747*** 0.00752*** 0.00931*** 0.00602* {t-2 to t-1} [0.00140] [0.00231] [0.00228] [0.00312] [0.00409] [0.00104] [0.00172] [0.00170] [0.00235] [0.00308] ∑ Capital {t-2 to t-1} -0.206 -0.21 -0.206 -0.295 -0.35 -0.0515 -0.0904 -0.0914 -0.18 0.084 [0.151] [0.269] [0.265] [0.357] [0.474] [0.106] [0.188] [0.186] [0.246] [0.330] Observations 143,596 143,596 143,596 103,033 40,563 143,596 143,596 143,596 103,033 40,563 R-squared 0.002 0.53 0.556 0.545 0.533 0.002 0.489 0.513 0.502 0.501 Year-Quarter FE X X X X X X X X X X Bank FE X X Bank-Firm FE X X X X X X Bank-Firm-Credit Rating FE X X Notes:InColumns1-5,thedependentvariableisquarterlychangeinthenaturallogarithmofU.S.banks'domestic lendingacrossfirmsandcreditratings[i.e.AAA,AA,A,BBB,BB,B,CCC,CC,C,D, NotRated].InColumns 6-10,thedependentvariableis thequarterlychangeinthenaturallogarithmof thenumber ofU.S. banks'domestic loans across firms and creditratings. Allspecifications includethefollowingcontrolsatthebankandfirmlevelforquarterst-2andt-1:Ln[TotalAssets],ReturnonAsset,andLeverageRatio.RobustStandarderrors(clusteredatthebank-firmlevel)arein parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A4. Quarterly Change in Domestic Bank Lending across Firms, Credit Maturities and Credit Ratings, for banks with different Tier1 Capital Ratios - Controlling for the Share of Banks' Foreign Assets. Quarterly Change in the Log of Quarterly Change in the Log of the Measure of U.S.-based lending: Lending Number of Loans Pooled across: Maturities Credit Ratings Maturities Credit Ratings VARIABLES [2] [4] [6] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.539** -0.619** -0.564*** -0.575*** (0.256) (0.298) (0.191) (0.222) ∑ Foreign Covid Exposure * Capital 0.0980* 0.114* 0.0780* 0.0853* {t-2 to t-1} (0.055) (0.066) (0.043) (0.050) ∑ Foreign Assets Share * Foreign Covid Exposure {t-2 to t-1} -2.052 -2.978 -25.22** -22.46 (15.450) (17.580) (11.900) (13.770) ∑ Foreign Assets Share * Foreign Covid Exposure * Capital -19.55 -22.69 -9.662 -11.89 {t-2 to t-1} (14.080) (16.740) (11.220) (13.050) ∑ Capital {t-2 to t-1} -4.514 -5.841 -3.304 -4.141 (3.430) (4.071) (2.727) (3.173) Observations 144,261 143,596 144,261 143,596 R-squared 0.528 0.557 0.474 0.514 Panel B: Immediate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.334** -0.361** -0.553*** -0.513*** (0.163) (0.179) (0.131) (0.151) ∑ Foreign Covid Exposure * Capital 0.0501*** 0.0526*** 0.0720*** 0.0675*** {t-2 to t-1} (0.017) (0.019) (0.013) (0.016) ∑ Foreign Assets Share * Foreign Covid Exposure {t-2 to t-1} -106.6 -136.6 -80.34 -90.26 (91.230) (107.700) (72.380) (84.190) ∑ Foreign Assets Share * Foreign Covid Exposure * Capital 0.783 3.36 -3.56 -1.862 {t-2 to t-1} (6.983) (8.152) (5.460) (6.252) ∑ Capital {t-2 to t-1} -2.047* -2.620* -3.375*** -3.441*** (1.224) (1.402) (0.929) (1.096) Observations 144,261 143,596 144,261 143,596 R-squared 0.528 0.557 0.474 0.514 Year-Quarter FE X X X X Bank-Firm-Maturity FE X X Bank-Firm-Credit Rating FE X X Notes: :Foreign Assets Share is total foreign lending assets divided by total assets.In Columns 1-2, the dependent variable is quarterly change in the naturallogarithmofU.S.banks'domestic lending across firms.In Columns 3-4,the dependent variable is the quarterly change in the naturallogarithmofthenumberofU.S.banks'domestic loansacross firms and credit ratings.In Columns 1 and 3, the dependent variable is pooled across loan maturities [i.e.loan with a maturity less than one yearand loan with a maturity more than one year],and in Columns 2 and 4 it is pooled across creditratings[i.e.AAA,AA,A,BBB,BB,B,CCC,CC,C,D,NotRated].Allspecificationsincludethe followingcontrols at the bankand firmlevelforquarterst-2andt-1:Ln[TotalAssets],Returnon Asset,and LeverageRatio.Allstandard errors clustered at the bank-firm level. Robust Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A5. Quarterly Change in interest rates and rate spreads across Firms, Credit Maturities, and Credit Ratings, for banks with different Capital Ratios. Dependent variable: Interest Rate Spread VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.000468 -0.000265 0.000607 0.000597 -0.00531*** 0.000849 0.00544*** 0.00191* [0.00105] [0.000742] [0.00108] [0.000786] [0.00174] [0.000974] [0.00202] [0.00103] ∑ Foreign Covid Exposure * Capital 4.97E-05 0.000195*** 1.77E-06 0.000148*** 0.000265*** 5.66E-05 -0.000467*** -0.000168** {t-2 to t-1} [5.79e-05] [4.87e-05] [5.88e-05] [4.96e-05] [0.000103] [6.45e-05] [0.000116] [6.81e-05] ∑ Capital {t-2 to t-1} 0.00101 0.00311 0.00387 0.00732 -0.0296*** -0.00599 0.0392*** 0.0252*** [0.00803] [0.00542] [0.00785] [0.00550] [0.0104] [0.00806] [0.0113] [0.00779] Observations 105,066 105,066 113,061 113,061 62,288 62,288 71,786 71,786 R-squared 0.511 0.514 0.516 0.519 0.516 0.516 0.546 0.556 Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.00170* -0.00108 -0.000872 -0.000356 -0.00637*** 0.000168 0.00520*** 0.00134 [0.00102] [0.000702] [0.00104] [0.000728] [0.00169] [0.000884] [0.00193] [0.000940] ∑ Foreign Covid Exposure * Capital 0.000109* 0.000247*** 7.36E-05 0.000210*** 0.000313*** 0.000106* -0.000493*** -0.000130** {t-2 to t-1} [5.70e-05] [4.93e-05] [5.67e-05] [4.88e-05] [0.000102] [6.18e-05] [0.000113] [6.58e-05] ∑ Capital {t-2 to t-1} -0.00277 0.0039 -0.00101 0.00767 -0.0327*** -0.00759 0.0260** 0.0235*** [0.00793] [0.00549] [0.00778] [0.00555] [0.0101] [0.00814] [0.0109] [0.00786] Observations 105,066 105,066 113,061 113,061 62,288 62,288 71,786 71,786 R-squared 0.511 0.514 0.516 0.518 0.516 0.516 0.547 0.556 Capital Ratio Tier1 CET1 Tier1 CET1 Tier1 CET1 Tier1 CET1 Year-Quarter FE X X X X X X X X Bank-Firm-Maturity FE X X X X Bank-Firm-Credit Rating FE X X X X Notes: The dependentvariable isthe quarterlychange inaverage interestrate orspread,forloans pooledacross loanmaturities [i.e.loan witha maturityless thanone yearand loanwith amaturity morethan oneyear]inColumns 1,2,5and 6,orforloans pooledacross creditratings [i.e.AAA,AA,A,BBB,BB,B,CCC,CC,C,D,Not Rated]inColumns3,4,7and8.Allspecificationsincludethefollowingcontrolsatthebankandfirmlevelforquarterst-2andt-1:Ln[TotalAssets],Returnon Asset,and Leverage Ratio.All standard errors clustered at the bank-firm level. Robust Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A6. Quarterly Change in Domestic Bank Lending across Firms, Credit Maturities and Credit Ratings - Controlling for Firms' Covid-19 Exposure. Measure of U.S.-based lending: Quarterly Change in Log of Lending Volume Quarterly Change in Log of Number of Loans Credit Credit Credit Credit Pooled across: Maturities Maturities Maturities Maturities Ratings Ratings Ratings Ratings VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.190*** -0.180*** -0.180*** -0.168*** -0.240*** -0.237*** -0.204*** -0.198*** (0.04) (0.04) (0.05) (0.05) (0.04) (0.04) (0.04) (0.04) ∑ Foreign Covid Exposure * Capital 0.0100*** 0.00942*** 0.00967*** 0.00893*** 0.0133*** 0.0131*** 0.0116*** 0.0113*** {t-2 to t-1} (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) ∑ Firm Covid Exposure {t-2 to t-1} 0.000583 0.00221 -0.00329 -0.00231 -0.00408 -0.00332 -0.00534 -0.00461 (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.00) (0.00) ∑ Firm Covid Exposure * Capital -6.83E-05 -0.000192 0.000196 0.000126 0.0003 0.000245 0.000369 0.000315 {t-2 to t-1} (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) ∑ Capital {t-2 to t-1} -0.961*** -0.985*** -1.129*** -0.942** -1.499*** -1.533*** -1.286*** -1.194*** (0.35) (0.35) (0.39) (0.38) (0.28) (0.28) (0.30) (0.30) Observations 144,018 144,018 143,351 143,351 144,018 144,018 143,351 143,351 R-squared 0.482 0.527 0.53 0.556 0.443 0.473 0.49 0.513 Panel B: Immediate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.181*** -0.174*** -0.166*** -0.155*** -0.240*** -0.237*** -0.204*** -0.199*** (0.04) (0.04) (0.05) (0.05) (0.03) (0.03) (0.04) (0.04) ∑ Foreign Covid Exposure * Capital 0.00980*** 0.00929*** 0.00915*** 0.00846*** 0.0136*** 0.0134*** 0.0118*** 0.0115*** {t-2 to t-1} (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) ∑ Firm Covid Exposure {t-2 to t-1} 0.00116 0.00279 -0.00286 -0.00191 -0.00335 -0.00261 -0.00478 -0.00407 (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.00) (0.00) ∑ Firm Covid Exposure * Capital -0.000112 -0.000235 0.000163 9.51E-05 0.000246 0.000191 0.000326 0.000274 {t-2 to t-1} (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) ∑ Capital {t-2 to t-1} -0.859** -0.900*** -1.020*** -0.839** -1.430*** -1.469*** -1.228*** -1.137*** (0.34) (0.34) (0.38) (0.37) (0.27) (0.27) (0.29) (0.29) Observations 144,018 144,018 143,351 143,351 144,018 144,018 143,351 143,351 R-squared 0.482 0.527 0.53 0.556 0.443 0.473 0.49 0.513 Year-Quarter FE X X X X X X X X Bank-Firm FE X X X X Bank-Firm-Maturity FE X X Bank-Firm-Credit Rating FE X X Notes:FirmForeign Covid Exposure is defined as the Foreign Covid Exposure indexofthe U.S. state of the borrowing firm’s headquarters. In Columns 1-4,the dependent variable is quarterly change in the natural logarithmof U.S. banks' domestic lending across firms. In Columns 5-8, the dependent variable is the quarterly change in the naturallogarithmofthe numberof U.S.banks' domestic loans across firms and credit ratings.In Columns 1, 2,5 and 6, the dependent variableispooledacrossloanmaturities[i.e.loanwithamaturitylessthanone yearand loanwith amaturity morethan oneyear],andin Columns3,4,7and8itis pooled across credit ratings [i.e.AAA,AA,A,BBB,BB,B,CCC,CC,C,D,NotRated].Allspecifications includethe followingcontrols atthe bankand firmlevel for quarters t-2 and t-1: Ln[Total Assets], Return on Asset, and Leverage Ratio. All standard errors clustered at the bank-firm level. Robust Standard errors in

Table A7. Quarterly Change in Domestic Bank Lending across Firms, Credit Maturities and Credit Ratings, for banks with different Tier1 Capital Ratios - For borrowing firms in Covid-19-sensitive and insensitive industries. Measure of U.S.-based lending: Quarterly Change in Log of Lending Volume Quarterly Change in Log of Number of Loans Insensitiv Insensitiv Insensitiv Insensitiv Industry Covid Sensitivity: Sensitive Sensitive Sensitive Sensitive e e e e VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.169*** -0.290*** -0.149** -0.324*** -0.249*** -0.288*** -0.218*** -0.292*** (0.06) (0.09) (0.06) (0.11) (0.05) (0.08) (0.05) (0.10) ∑ Foreign Covid Exposure * Capital 0.00825*** 0.0172*** 0.00748** 0.0186*** 0.0137*** 0.0161*** 0.0119*** 0.0168*** {t-2 to t-1} (0.00) (0.01) (0.00) (0.01) (0.00) (0.00) (0.00) (0.01) ∑ Capital {t-2 to t-1} -0.853** -2.107*** -0.694 -2.112*** -1.593*** -1.880*** -1.344*** -1.660** (0.42) (0.72) (0.49) (0.79) (0.34) (0.69) (0.40) (0.75) Observations 88,990 45,032 85,504 48,598 88,990 45,032 85,504 48,598 R-squared 0.53 0.524 0.554 0.56 0.47 0.482 0.508 0.527 Panel B: Immediate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.157*** -0.305*** -0.127** -0.325*** -0.244*** -0.294*** -0.206*** -0.295*** (0.05) (0.09) (0.06) (0.10) (0.04) (0.08) (0.05) (0.09) ∑ Foreign Covid Exposure * Capital 0.00784*** 0.0181*** 0.00656* 0.0189*** 0.0136*** 0.0167*** 0.0115*** 0.0172*** {t-2 to t-1} (0.00) (0.01) (0.00) (0.01) (0.00) (0.00) (0.00) (0.01) ∑ Capital {t-2 to t-1} -0.758* -2.111*** -0.558 -1.997** -1.501*** -1.851*** -1.238*** -1.564** (0.40) (0.70) (0.47) (0.78) (0.32) (0.68) (0.38) (0.72) Observations 88,990 45,032 85,504 48,598 88,990 45,032 85,504 48,598 R-squared 0.53 0.524 0.554 0.56 0.47 0.482 0.508 0.528 Year-Quarter FE X X X X X X X X Bank-Firm-Maturity FE X X X X Bank-Firm-Credit Rating FE X X X X Notes: Odd columns are restricted to firms belonging to Covid-sensitive industries and even columns are restricted tofirms belongingto Covid-insensitive industries.COVID-sensitive industries are definedbased onKaplan,Molland Violante(2020),"Thegreat lockdownand the big stimulus:Tracing the pandemic possibility frontierfor the U.S.", NBERWorking PaperNo. 27794.In Columns 1-4, the dependent variableis quarterlychangeinthenaturallogarithmofU.S.banks'domestic lendingacross firms.In Columns 5-8,thedependent variableis thequarterlychangeinthenaturallogarithmofthe numberofU.S.banks'domesticloans across firms andcredit ratings.In Columns 1,2,5 and 6,the dependent variable is pooled across loan maturities [i.e.loan with a maturity less than one yearand loan with a maturity more thanoneyear],andinColumns 3,4,7and8itis pooledacross creditratings [i.e.AAA,AA,A,BBB,BB,B,CCC, CC,C,D,Not Rated].All specifications include the following controls at the bank and firm levelfor quarters t-2 and t-1: Ln[TotalAssets], Return on Asset,and Leverage Ratio. All standard errors clustered at the bank-firm level. Robust Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A8. Quarterly Change in Domestic Bank Lending across Firms and Credit Maturities for banks with different Capital Ratios -- including industry*quarter fixed effects. Measure of U.S.-based lending: Quarterly Change in the Log of Lending Quarterly Change in the Log of the Number of Loans VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.00342** -0.00316* 0.00714 0.00796 -0.00379*** -0.00510*** 0.00378 0.00378 (0.002) (0.002) (0.011) (0.012) (0.001) (0.001) (0.008) (0.008) ∑ Foreign Covid Exposure * Capital -0.000503 -0.000784 -0.000574 -0.000574 {t-2 to t-1} (0.001) (0.001) (0.001) (0.001) ∑ Capital {t-2 to t-1} -0.1 -0.0802 -0.00753 -0.00753 (0.066) (0.075) (0.048) (0.048) Panel B: Immediate Counterparty Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.00396** -0.00359* 0.00516 0.00796 -0.00380*** -0.00566*** 0.00478 0.00378 (0.002) (0.002) (0.013) (0.012) (0.001) (0.001) (0.009) (0.008) ∑ Foreign Covid Exposure * Capital -0.000371 -0.000784 -0.000484 -0.000574 {t-2 to t-1} (0.001) (0.001) (0.001) (0.001) ∑ Capital {t-2 to t-1} -0.107 -0.0802 -0.00706 -0.00753 (0.079) (0.075) (0.050) (0.048) Observations 45,000 41,944 41,944 41,944 45,000 41,944 41,944 41,944 R-squared Capital Ratio None None Tier1 CET1 None None Tier1 CET1 Industry-Quarter FE X X X X X X X X Bank Controls X X X X X X Notes: In Columns 1-4,the dependent variable is quarterly change in the naturallogarithmofU.S.banks'domestic lendingacross firms and inColumns 5-8,the dependentvariableis thequarterlychangeinthenaturallogarithmofthenumberofU.S.banks'domestic loans across firms.Bankcontrols includes thefollowing variables for quarters t-2 and t-1: Ln[Total Assets], Return on Asset, and Leverage Ratio. Robust Standard errors (clustered at the bank-firm level) are in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Table A9. Quarterly Change in Domestic Bank Lending across Firms, Credit Maturities and Credit Ratings, for banks with different Tier1 Capital Ratios - For Small and Large borrowing firms. Measure of U.S.-based lending: Quarterly Change in Log of Lending Volume Quarterly Change in Log of Number of Loans Firm Size: Small Large Small Large Small Large Small Large VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] Panel A: Ultimate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.431*** -0.122** -0.303** -0.125** -0.437*** -0.203*** -0.433*** -0.160*** (0.123) (0.051) (0.131) (0.060) (0.114) (0.041) (0.119) (0.046) ∑ Foreign Covid Exposure * Capital 0.0222*** 0.00633** 0.0152** 0.00653** 0.0229*** 0.0114*** 0.0226*** 0.00935*** {t-2 to t-1} (0.007) (0.003) (0.007) (0.003) (0.006) (0.002) (0.006) (0.003) ∑ Capital {t-2 to t-1} -2.934*** -0.585 -1.896** -0.597 -3.249*** -1.245*** -3.051*** -0.884** (0.914) (0.386) (0.938) (0.451) (0.798) (0.318) (0.805) (0.360) Observations 47,973 96,288 50,039 93,557 47,973 96,288 50,039 93,557 R-squared 0.472 0.544 0.497 0.572 0.464 0.484 0.492 0.527 Panel B: Immediate Risk Weighted ∑ Foreign Covid Exposure {t-2 to t-1} -0.445*** -0.123*** -0.317** -0.116** -0.439*** -0.209*** -0.430*** -0.167*** (0.124) (0.047) (0.127) (0.056) (0.112) (0.038) (0.114) (0.043) ∑ Foreign Covid Exposure * Capital 0.0243*** 0.00655** 0.0168** 0.00625** 0.0246*** 0.0119*** 0.0235*** 0.00983*** {t-2 to t-1} (0.007) (0.003) (0.007) (0.003) (0.006) (0.002) (0.006) (0.002) ∑ Capital {t-2 to t-1} -3.005*** -0.565 -1.979** -0.523 -3.244*** -1.220*** -2.987*** -0.867** (0.930) (0.375) (0.911) (0.439) (0.804) (0.305) (0.791) (0.347) Observations 47,973 96,288 50,039 93,557 47,973 96,288 50,039 93,557 R-squared 0.472 0.544 0.497 0.572 0.464 0.484 0.492 0.527 Year-Quarter FE X X X X X X X X Bank-Firm-Maturity FE X X X X Bank-Firm-Credit Rating FE X X X X Notes: : Large firms are firms with total assets above the sample median firm asset size.Small firms are firms with totalassets belowthe median.In Columns 1-4,the dependent variable is quarterly change in the naturallogarithmofU.S.banks'domestic lendingacross firms.In Columns 5-8,thedependentvariableis the quarterlychange inthe naturallogarithmofthe numberofU.S.banks'domesticloans across firms andcreditratings.InColumns 1,2,5and6,thedependentvariableis pooledacross loanmaturities [i.e.loanwithamaturity less than oneyear and loan with a maturity more than one year],and in Columns 3,4,7and 8it is pooled across credit ratings [i.e.AAA,AA,A,BBB, BB,B, CCC, CC, C, D, Not Rated]. All specifications include the following controls at the bank and firm level for quarters t-2 and t-1: Ln[Total Assets],ReturnonAsset,andLeverageRatio.Allstandarderrors clusteredatthe bank-firmlevel.RobustStandard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.