Cross-Border Debt Flows and Credit Allocation: Firm-Level Evidence from the Euro Area

University of Groningen

Cross-Border Debt Flows and Credit Allocation

te Kaat, Daniel

Published in:

Journal of Money, Credit, and Banking DOI:

10.1111/jmcb.12776

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2021

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

te Kaat, D. (2021). Cross-Border Debt Flows and Credit Allocation: Firm-Level Evidence from the Euro Area. Journal of Money, Credit, and Banking. https://doi.org/10.1111/jmcb.12776

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

DANIEL MARCEL TE KAAT

Cross-Border Debt Flows and Credit Allocation:

Firm-Level Evidence from the Euro Area

This paper employs euro area firm-level data covering the years 2002–18 to examine the impact of cross-border debt flows on the domestic allocation of credit across firms conditional on their profitability. As only debt flows driven by global push factors are exogenous with respect to domestic credit allocation, I overcome the endogeneity of debt flows by instrumenting them with a measure of global uncertainty (VIX). My results show that debt flows raise the credit growth rates of low performing firms significantly more than those of high performing firms. This result is driven by domestic banking sectors with lower capitalization.

JEL codes: D22, F32, F41, G15 Keywords: credit allocation, debt flows, agency problems The international integration of financial markets and banking systems has increased significantly over the recent decades. Financial in-stitutions and investors have expanded their foreign investments, raising the amounts of cross-border capital flows. As a result, local firms and banks located in recipi-ent countries have access to additional funding at more favorable terms. This phe-nomenon is also marked in the 11 founding members of the euro area and Greece, where cross-border debt inflows, as one major component of total capital flows, took an average value of 1.8% of GDP during 2002–07 and of 0.6% during 2002–18, with I thank the editor, two anonymous referees, Luís A.V. Catão, Nathan Converse, Lisa Cycon, Valeriya Dinger, Halit Gonenc, Olivier Jeanne, Katja Mann, Alexander Mayer, Steven Ongena, Alessandro Rebucci, Frank Westermann, Joachim Wilde, and conference participants at the 2018 EEA Annual Congress, at the 2017 AEA Annual Meeting (ASSA), at the 2016 German Economic Association Annual Meeting, at the 2016 German Finance Association Annual Meeting, at the 8th International Conference of the Financial Engineering and Banking Society (F.E.B.S.), at the 10th Conference for Macroeconomics (ifo Dresden and Helmut Schmidt University), at Johns Hopkins Carey Business School, at the University of Osnabrück, and at the Deutsche Bundesbank for valuable comments. A previous version circulated under the title “International Capital Flows and the Allocation of Credit Across Firms.”

Daniel Marcel te Kaat is at University of Groningen (E-mail: d.m.te.kaat@rug.nl).

Received April 16, 2019; and accepted in revised form September 22, 2020.

Journal of Money, Credit and Banking, Vol. 00, No. 00 (xxxx 2020)

© 2021 The Authors. Journal of Money, Credit and Banking published by Wiley Periodicals LLC on behalf of Ohio State University

This is an open access article under the terms of the Creative Commons Attribution-NonCom-mercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.

peaks of 2.9% in 2002 and 2017, and of 4%–6.2% in 2007 and 2008. However, al-though these debt inflows into the euro area have increased the quantity and reduced the price of funding, they have not been associated with higher output growth (Lane 2013). In fact, the correlation between debt inflows and real GDP growth is equal to −17% during 2002–07, and −16% when the crisis and postcrisis episodes of 2008– 18 are included. One reason put forward is that cross-border capital inflows can lead to a misallocation of credit (Bank for International Settlements 2015).

Indeed, as shown by Obstfeld (2012), European countries subject to large capital inflows experienced booms in less productive sectors, such as real estate and (gov-ernment) consumption, rather than in nonfinancial business investment. Samarina and Bezemer (2016) find empirically that credit in countries with larger capital inflows has been reallocated from nonfinancial businesses toward households. In addition to this shift in credit from firms toward the household and public sector, several papers establish that capital flows also lead to an expansion of the nontradable sector relative to the tradable sector (Lane 2013, Reis 2013, Benigno and Fornaro 2014). Benigno, Converse, and Fornaro (2015) provide empirical evidence that large capital inflows are associated with growth of the nontradable sectors at the expense of the tradable sectors, such as agriculture and manufacturing.

While these papers identify the relation between foreign capital flows and cross-sectoral shifts in credit, in this paper, I focus on nonfinancial business lending and examine shifts in credit within industries. Specifically, I analyze to what extent and through which channels cross-border debt inflows lead to higher credit growth of ex ante low performing firms, relative to their more profitable industry peers. Employing a comprehensive data set of euro area firms covering the period 2002–18, I show that debt inflows lead to a disproportionate increase in credit growth of low performing relative to high performing firms.

The main empirical challenge when examining the effects of cross-border debt flows is that they might be driven by global push factors external to a country (e.g., global uncertainty) or local pull factors (e.g., domestic macro-economic fundamen-tals). As only debt flows that are driven by global push factors are exogenous with respect to domestic credit allocation, I overcome the endogeneity problem of debt flows by instrumenting them with a measure of global uncertainty (the VIX), which has been shown to be an important exogenous push determinant of cross-border capi-tal flows (e.g., Forbes and Warnock 2012, Rey 2013, Fratzscher, Lo Duca, and Straub 2018, Cerutti, Claessens, and Puy 2019, di Giovanni et al. 2019, Miranda-Agrippino and Rey 2020). In particular, during high (low) levels of global uncertainty, debt in-flows to (outin-flows from) advanced economies typically increase, as do debt outin-flows from (inflows to) emerging economies. As I detail below, this pattern also holds in the euro area, with a sensitivity of debt flows to the VIX that is stronger in the core of the euro area as opposed to the periphery.

The main result of this paper is that higher debt inflows lead to a disproportionate increase in credit growth of low performing firms relative to high performing ones. The effect is statistically significant and economically important: a 1 percentage point (henceforth pp) increase in cross-border debt inflows raises the annual credit growth

rates of firms at the 25th percentile of the distribution of profitability by 0.6 pp more than of firms at the 75th percentile. This result is robust to several robustness checks in which I employ different measures of firm profitability, vary the time and country coverage of the sample, and differentiate between small and large firms, as well as those in the tradable and nontradable sectors.

In the second part of the paper, I show that risk taking and search for yield of the financial sector are important transmission channels of cross-border debt flows to shifts in credit allocation. In particular, financial institutions and investors seem to raise their credit supply to low performing firms significantly more than to high per-forming ones, as the former have an increased probability of default and pay a higher interest rate on their debt. Consistent with the theories of Acharya and Naqvi (2012) and Martinez-Miera and Repullo (2017), which link push-driven cross-border capi-tal flows to risk-taking and search for yield and which depart from agency problems in the financial sector, I also explore the variation in the magnitude of the effects of debt flows on credit allocation across financial sectors with different capitalization as a proxy for agency problems (Holmstrom and Tirole 1997). I find that the observed relation between debt flows and the disproportionate increase in lending to low per-forming, relative to high perper-forming, firms is reversed in better capitalized financial systems less subject to agency problems.

These results contribute to the literature in several dimensions. First, my results can help to explain the difficulties of the empirical literature to identify a uniform positive relationship between cross-border capital inflows and aggregate economic growth (e.g., Grilli and Milesi-Ferretti 1995, Edison et al. 2002, Bonfiglioli 2008, Bussière and Fratzscher 2008, Kose, Prasad, and Terrones 2009, Eichengreen, Gul-lapalli, and Panizza 2011, Aizenman, Jinjarak, and Park 2013). Notably, I find that cross-border debt flows change the composition of credit in the economy, benefiting disproportionately more ex ante low performing than high performing firms. Rela-tive to the extant literature on the nexus between foreign capital, credit allocation, and real performance that focuses on a shift in credit across industries (e.g., Reis 2013, Benigno and Fornaro 2014, Benigno, Converse, and Fornaro 2015, Samarina and Bezemer 2016), I show (i) that international debt flows also affect the allocation of credit within industries and (ii) that risk taking and search for yield of the financial sector are important mediating channels of cross-border debt flows to within-industry shifts in credit allocation. Thereby, my results also broadly speak to the literature on the relationship between foreign capital flows and the incidence of financial crises (e.g., Rancière, Tornell, and Westermann 2008, Reinhart and Rogoff 2008, Obstfeld 2012, Gourinchas and Obstfeld 2012, Lane and McQuade 2014).

This paper is structured as follows. In Section 1, I describe the data set. Section 2 introduces the empirical strategy. The main results are presented in Section 3. In Section 4, I explore possible transmission mechanisms of cross-border debt flows to credit allocation. Section 5 concludes. An Online Appendix provides the results of several additional regressions and robustness checks.

TABLE 1

The Cross-Country Distribution of Sample Firms

Country Number of Firms

Austria 58 Belgium 78 Finland 109 France 497 Germany 528 Greece 197 Ireland 21 Italy 203 Luxembourg 25 Netherlands 77 Portugal 38 Spain 111  1,942 1. DATA 1.1 Firm-Level Data

To conduct the empirical analysis, I assembled a comprehensive euro area firm-level data set. It comprises nonfinancial firms in the 11 founding members of the euro area (Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, Portugal, and Spain) and Greece at annual frequency from 2002 to 2018. The distribution of firms across countries is depicted in Table 1. The firm-level data stem from Thomson Reuter’s Worldscope database, which is a high-quality database for firm balance sheet and income data and has been used widely in the empirical finance literature (e.g., La Porta et al. 2000, Leuz, Nanda, and Wysocki 2003, Wei and Zhang 2008, Houston, Itzkowitz, and Naranjo 2017).

Thomson Reuter’ s Worldscope database covers publicly quoted companies, which altogether represent about 95% of global market value, as well as a smaller number of private companies. In sum, slightly more than 20% of the firms in my sample are small and medium sized, i.e., they have less than 250 employees. The remaining firms are classified as large firms. While large firms in the euro area represent only 0.2% of all firms, they are a critical element for economic dynamics. Specifically, in the euro area, they account for 29%–48% of total value added and are the main source of pro-ductivity growth (ECB 2013). Indirectly, the importance of large firms for economic dynamics has recently been confirmed by Greenstone, Mas, and Nguyen (2020), who show that small business lending is not an important determinant of overall economic activity in the United States.

In order to relate firms’ credit growth to cross-border debt flows, the overpro-portional representation of large and listed firms in my sample is an advantage be-cause these firms are likely to benefit disproportionately more from debt flows due to their access to the international bond markets, which can play a critical role in pass-ing through global credit to domestic firms (Borio, McCauley, and McGuire 2011).

TABLE 2

Summary Statistics

Obs. 25th Median 75th Definition

Dependent variable

DEBT 26,668 −16.62 −0.01 20.27 Growth rate of debt

INTEREST RATE 25,992 −1.25 −0.08 1.08 Change in interest expenses/total debt Regressors

DEBT INFLOWS 36,363 −2.83 0.29 4.82 Net portfolio debt and other inflows/GDP VIX 36,363 14.23 16.67 22.55 CBOE Volatility Index

PROFITABILITY 29,369 0.59 4.18 7.24 return on assets SIZE 30,249 10.48 11.98 13.73 ln(total assets)

LIQUIDITY 29,449 0.68 0.99 1.51 liquid assets/current liabilities

Notes: This table depicts the summary statistics of the baseline variables.DEBT is the firm-level credit growth rate and  INTEREST RATE is the change in the firm-level interest rate defined as interest expense relative to total debt. I employ net debt inflows to GDP as the country-level measure of cross- border capital flows and VIX is the CBOE volatility index. The analysis further includes several firm controls: the returns on assets (PROFITABILITY), the logarithm of total assets (SIZE) and liquid assets relative to current liabilities (LIQUIDITY) .

Further, larger firms often maintain credit relationships with large banks,1_{which the}

empirical literature has shown to be most significantly affected by international cap-ital flows via an increase in the supply of wholesale funding (e.g., di Giovanni et al. 2019, Dinger and te Kaat 2020). In fact, as I show in the Online Appendix, my results are statistically and economically stronger if I restrict the sample to large firms with more than 250 employees. An additional advantage of large/listed firms is the higher quality of their accounting data.

I define the dependent variable (firm credit growth) as the log difference in firm-level debt volumes, which include all interest bearing financial liabilities, in particular bank loans and bonds.2 In the empirical analysis, I also include different firm-level regressors. Most importantly, I include firms’ returns on assets (PROFITABILITY ), calculated as earnings before interest after tax divided by total assets, which allows me to focus on the profitability of operations regardless of the way the assets are fi-nanced. The results are materially unchanged when using earnings before interest and taxes in the numerator of the return on assets (not reported) or employing the return on equity as the profitability measure (see the evidence in the Online Appendix). I further control for the logarithms of total assets (SIZE) and the shares of liquid assets relative to short-term liabilities (LIQU IDITY ).3 _{All variables are winsorized at the}

3% and 97% levels to reduce the potential impact of outliers. My results, however, are generally robust to alternative ways of handling those outliers.

Table 2 contains the summary statistics for these variables. The median (mean) rate of firm debt growth is−0.01% (5.1%). The variation in credit growth, however, is distinct, indicated by a 25th percentile of−16.6% and a 75th percentile of 20.3%. 1. As small firms are more subject to information asymmetries, they often maintain credit relationships with smaller banks that are headquartered closer to relationship customers, reducing problems related to asymmetric information (Berger and Udell 2002).

2. Values of firm debt growth smaller than−100% are set equal to −100%.

3. In Section 4.1, I control for additional firm-level covariates that are linked to firm risk, such as leverage, interest coverage and asset tangibility. The results are unaffected.

At the same time, the median firm in my sample also experiences a reduction in its interest expenses by 8 basis points per year. The average return on assets is equal to 4.2%. Again, the variation is pronounced and ranges from 0.6% (25th percentile) to 7.2% (75th percentile). Finally, the median natural logarithm of total assets is equal to 11.98 thousand€ and the average liquidity ratio is equal to 0.99%.

1.2 Capital Flow Data

I match the firm-level data with data on cross-border capital flows. Since I presume that firm-level debt dynamics are affected more by cross-border debt flows than by foreign direct investments or portfolio equity flows, my measure of cross-border cap-ital flows in this paper are international debt flows. In fact, in previous versions of this paper, I used the current account deficit as an additional proxy for capital flows, which in addition to debt flows includes foreign direct investments and equity flows, and the economic and statistical significance of my estimates was lower.

Cross-border debt flows are defined as the sum of net portfolio debt inflows and other inflows, which contain interbank credit flows, relative to GDP. Specifically, I calculate net debt inflows as the difference between a country’ s change in the stock of portfolio debt liabilities and other liabilities over nominal GDP and a country’ s change in the stock of portfolio debt and other assets over nominal GDP. While the data from 2002–15 are available in Lane and Milesi-Ferretti (2017), the missing data from 2016–18 are hand-matched using the International Financial Statistics. Similar to the firm-level variables, cross-border debt flows can take extreme values, espe-cially in the international financial centers Ireland and Luxembourg, and thus distort the estimation results. I hence winsorize them at the 3% and 97% levels. After this winsorization, as can also be seen from Table 2, cross-border debt flows have a me-dian value of 0.29% relative to GDP. Therefore, most countries in my sample exhibit significant inflows of cross-border debt, as can also be seen from the pronounced 75th percentile of 4.82%.

2. EMPIRICAL STRATEGY 2.1 Econometric Specification

The main hypothesis that I test in this paper is whether cross-border debt inflows lead to a disproportionate increase in credit growth of ex ante low performing firms relative to high performing ones. Similar to a specification in Jiménez et al. (2014), which explores whether aggregate monetary policy changes (debt flows in my case) affect the credit growth of ex ante risky firms (low performing firms in my case), I estimate the following regression equation:

DEBTf i jt = αf i j+ αjt+ αit+ β · (DEBTINFLOWSj_,t−1

where f indexes firms, i industries,4 j countries, and t year. The dependent variable in equation (1) is the growth rate of firm debt. On the right-hand side of equa-tion (1), DEBT INFLOW Sj,t−1 are lagged country-level cross-border debt flows,

PROFITABILITYf,i, j,t−1is lagged firm-level return on assets, and Xf,i, j,t−1includes

the firm-level controls introduced in Section 1.1, all of which are lagged by 1 year to reduce endogeneity concerns. The coefficient of interest in the following analysis is β, determining whether higher debt inflows lead to increased credit growth of ex ante low performing, relative to high performing, firms.5

αf i j,αjt, andαitare firm, country-year, and industry-year fixed effects. Firm fixed

effects are important to control for individual firm-specific effects that do not vary over time, such as the location of the firm. I also include industry-year fixed effects to control for industry-specific, time-varying trends, such as different sensitivities of industries to the business cycle. Industry-year fixed effects also allow me to identify within-industry shifts in credit across firms with different returns on assets. Country-year fixed effects are included to control for domestic macro-economic variables, such as GDP growth and inflation. Controlling for inflation is particularly important because nominal debt growth (the dependent variable) can be expected to be higher in countries with higher inflation levels. Note that country-year fixed effects also absorb the direct effect of cross-border debt flows, which therefore cannot enter the regres-sions individually. As PROFITABILITY is not absorbed by the set of fixed effects, it enters the regressions individually and is subsumed in the vector Xf,i, j,t−1. The

stan-dard errors are clustered at the industry-year level to account for the within-industry correlation across firms.

2.2 Identification via Instrumental Variables

Cross-border debt flows can be driven by global push factors that are external to a country and affect the supply of funding in the recipient economy, such as global uncertainty, or local pull factors, such as domestic macro-economic fundamentals, which rather affect the domestic demand for funding (e.g., Calvo, Leiderman, and Reinhart 1996, Fratzscher 2012, Rey 2013, Ghosh et al. 2014, Bruno and Shin 2015). In the latter case, therefore, debt flows are correlated with (potentially unobservable) domestic fundamentals and hence subject to omitted variable concerns. For instance, to the extent that local pull factors are associated with improved domestic funda-mentals and higher expectations about the returns of domestic projects, they shift upward the demand for domestic credit and result in debt inflows. In this framework, as changes in domestic fundamentals raise domestic credit demand and, thereby, in-crease both international debt inflows and firms’ observable levels of credit, pull-driven debt flows are clearly endogenous to the allocation of credit across firms.

4. The number of industries according to the Worldscope industry identifier is slightly larger than the three-digit NAICS and two-digit SIC codes. A similar disaggregation of industries is, for instance, also used in the empirical literature on the effects of finance on growth (e.g., Rajan and Zingales 1998). My results, however, are also robust to a broader definition of industries, e.g., using the one-digit SIC codes (not reported).

5. In unreported regressions, I also included the lagged level of firm-level debt. The results were largely unaffected.

TABLE 3

The Effect of The VIX on Debt Flows

(1) (2)

DEBT INFLOWS DEBT INFLOWS

VIX 0.169∗∗∗ 0.210∗∗∗

(0.006) (0.006)

VIX× DUMMY_GIIPS −0.131∗∗∗

(0.014)

Country Fixed Effects Yes Yes

Observations 24,028 24,028

R-squared 0.186 0.189

Notes: The table depicts the results of regressing country-level cross-border debt flows over GDP on the interaction between the VIX and its interaction with a dummy equal to one for Greece, Ireland, Italy, Portugal, and Spain. Both regressions include country fixed effects. The robust standard errors are shown in parentheses. *p< 0.10, **p < 0.05, ***p < 0.01.

Following this line of arguments, only the component of debt flows that is driven by global push factors is exogenous with respect to domestic credit allocation. I therefore overcome the endogeneity of debt flows by instrumenting them in a classical 2SLS regression with a measure of global uncertainty—the VIX—that has been shown to be one of the most important exogenous (push) determinants of cross-border capital flows (e.g., Forbes and Warnock 2012, Rey 2013, Fratzscher, Lo Duca, and Straub 2018, Cerutti, Claessens, and Puy 2019, di Giovanni et al. 2019, Miranda-Agrippino and Rey 2020).6 _{Specifically, during high (low) levels of global uncertainty, debt}

inflows to (outflows from) advanced economies typically increase, as do debt outflows from (inflows to) emerging economies.

This pattern is also apparent in the euro area (a region comprising only advanced economies), as I show in Table 3. It indicates that, on average, higher global un-certainty is associated with cross-border debt flows into the countries of the euro area (column 1). However, as can be seen from the positive linear effect of the VIX and the negative coefficient on the interaction term between the VIX and a dummy equal to 1 for Greece, Ireland, Italy, Portugal, and Spain (henceforth GIIPS dummy), the strength of this effect is stronger in the core countries of the euro area than in the periphery, where concerns regarding financial stability and eco-nomic perspectives, especially post-2007, are stronger than in the core (column 2). In economic terms, whereas a 1 pp increase in the VIX raises debt flows to the core countries of the euro area by 0.21 pp, the effect is only equal to 0.08 pp for the GIIPS countries.7

6. As I show in Figure A1 of the Online Appendix, the VIX is in most circumstances exogenous to domestic fundamentals in the euro area and, instead, closely associated with key events in the United States and several emerging market economies. One exception is the European sovereign debt crisis of 2010–14, which did affect the VIX. However, in the Online Appendix, I also present specifications that exclude this episode and my results are unaffected.

7. This is the sum of the coefficients on debt flows and the interaction between debt flows and the GIIPS dummy.

Although the VIX linearly would already be a relevant instrument for debt flows in the euro area, following the evidence of Table 3, I use both the VIX and the inter-action between the VIX and the GIIPS dummy as instruments for debt flows, thereby capturing the different sensitivities of debt flows in core and periphery countries of the euro area with respect to global uncertainty. I therefore, essentially, run the first-stage regression of debt flows on the VIX separately for countries in the core and in the periphery of the euro area.8

3. MAIN RESULTS 3.1 Fixed Effects Results

As a benchmark for the following second-stage instrumental variable results, I esti-mate equation (1) without instrumentation. As argued in Section 2, this is problematic because cross-border debt flows are not only driven by global push factors that are external to a country and increase the supply of funding, but also by endogenous pull factors, such as macro-economic fundamentals, that affect domestic credit demand. Thus, my ordinary least squares (OLS) estimates are likely to be biased if I include both periods of push- and periods of pull-driven debt inflows. Indeed, as can be seen from column (1) of Table 4, debt flows are not associated with a differential impact on the credit growth of low versus high performing firms, which should not be over-interpreted given the aforementioned potential endogeneity of debt flows.

Therefore, I next try to disentangle debt flows that are driven by exogenous push factors from those driven by pull factors. This is difficult because, in equilibrium, debt flows reflect the confluence of supply (push) and demand (pull) factors and it is impossible to attribute the observed flows alone to one of these factors. In order to identify episodes that are at least likely to be mainly driven by supply (push) factors, I define push-driven debt inflow episodes as periods when the average country-level interest rate across firms relative to the euro area average of all firms in my sam-ple decreases significantly. Consistent with Ghosh et al. (2014), the idea is that the domestic interest rate decreases when the supply of funding due to push-driven debt inflows goes up, while the demand stays relatively constant. In contrary, should higher debt flows be driven by shifts in domestic credit demand, this should result in higher domestic interest rates. Following this argument, I define country-specific episodes when cross-border debt flows are driven by supply (push) factors as those episodes in which an increase (decrease) in debt inflows is associated with a change in average, country-level firm interest rate spreads in the lowest (highest) 33% of the distribu-tion and reestimate the models for only those episodes.9_{For the sake of comparison,}

8. Strictly speaking, I instrument the interaction between cross-border debt flows and firm-level returns on assets with the double interaction between the VIX and returns on assets, as well as the triple interaction between the VIX, the GIIPS dummy and returns on assets.

9. I deliberately choose a relatively high threshold of the highest 33% of interest rate changes to ensure that I indeed identify episodes that are mainly supply (push) driven. For lower thresholds, endogenous pull (demand) factors are still likely to play an important role, biasing the results.

TA B L E 4 B a seline Resul t s OLS, full sample OLS, pull factor s dominate O LS, push factor s dominate IV IV (1) (2) (3) (4) (5)  DEBT  DEBT  DEBT  DEBT  INTEREST R A T E DEBT INFLO W S × PR OFIT A BILITY − 0.007 − 0.003 − 0.038 ∗∗ − 0.090 ∗∗ 0.016 ∗∗∗ (0.007) (0.009) (0.015) (0.043) (0.006) SIZE − 12.094 ∗∗∗ − 13.812 ∗∗∗ − 11.890 ∗∗∗ − 12.304 ∗∗∗ 0.808 ∗∗∗ (1.122) (1.542) (2.272) (1.159) (0.149) LIQ U IDITY 6 .727 ∗∗∗ 7.036 ∗∗∗ 6.143 ∗∗∗ 6.824 ∗∗∗ − 0.495 ∗∗∗ (0.739) (0.992) (1.448) (0.739) (0.110) PR OFIT A BILITY 0.422 ∗∗∗ 0.409 ∗∗∗ 0.441 ∗∗∗ 0.521 ∗∗∗ − 0.048 ∗∗∗ (0.063) (0.074) (0.133) (0.083) (0.012) Firm Fix ed E ff ects Y es Y es Y es Y es Y es Industry-Y ear Fix ed Ef fects Y es Y es Y es Y es Y es Country-Y ear Fix ed Ef fects Y es Y es Y es Y es Y es Observ ations 24028 16261 7023 24028 23843 R -squared 0 .197 0.258 0.390 0.189 0.130 First-Stage F -Statistic -3 5.5 37.0 p (o v eridentification) -0 .41 0 .68 Notes: In this table, I sho w the b aseline re g ression results. T he dependent v ariable in columns (1)–(4) is the fi rm-le v el credit gro w th rate. In column (5), the d ependent v ariable is the change in firm-le v el interest rates, defined as the ratios o f interest expenses o v er total debt. T he main re gressor are lagged country-le v el net d ebt inflo w s o v er GDP interacted with lagged firm-le v el return o n assets. I control for the one-year lag o f the follo wing firm co v ariates: size (log arithm o f total assets), liquidity (liquid assets o v er current liabilities), and profitability (the return on as sets). In addition, I include firm, country-year and industry-year fix ed ef fects. Columns (1)–(3) are estimated with OLS, where column (1) uses the full fi rm sample, and columns (2)–(3) restrict the sample to episodes during w hich an increase (decrease) in debt inflo ws is associated with a change in country-le v el firm interest rate spreads in the lo west (highest) 33% of the d istrib ution (push-dri v en debt flo ws, column 3 ), and all other episodes (pu ll-dri v en debt flo ws, column 2). Columns (4)–(5) use an IV re gression, emplo y ing the VIX and its interaction w ith an indicator that is one for G reece, Italy , Ireland, Portugal, and Spain as instruments for d ebt fl o w s. The st andard errors are clustered at the industry-year le v el and are sho wn in parentheses. * p < 0. 10, ** p < 0. 05, *** p < 0. 01.

I also re-estimate equation (1) for all other episodes that, therefore, can be seen as being rather demand driven.

As is apparent from column (3), cross-border debt flows are associated with a dif-ferential impact on credit growth of low and high performing firms during episodes where push factors are predominant. This effect is statistically significant at the 5% level and economically important: a 1 pp increase in cross-border debt inflows raises the annual credit growth rates of low performing firms at the 25th percentile of the profitability distribution by 0.25 pp more than of firms at the 75th percentile.10_In

con-trast, during episodes where pull factors are likely to affect debt flows, I find no link between debt flows and a disproportionate increase in lending toward firms with low, relative to high, returns on assets, as can be seen from the low interaction coefficient in column (2). This is evidence that only debt flows driven by supply (push) factors, i.e., where higher inflows are accompanied by significant reductions in the domestic interest rate, correspond with increased credit growth of ex ante low performing, rel-ative to high performing, firms, corroborating the importance of instrumenting debt flows with a supply (push) factor.

As I argue in Section 4, one reason for that only push (supply)-driven debt flows are associated with a differential impact on credit growth of low performing, relative to high performing, firms is that, in this case, the domestic interest rate decreases, in-ducing financial intermediaries to increase their risk taking and to search for yield. As firms with lower returns on assets have a higher probability of default and, as a conse-quence, pay a higher interest rate on their debt, they seem to benefit disproportionately more from (push-driven) debt inflows than firms with higher returns on assets.

Following the evidence of Section 3.1, in the remainder of this paper, I instrument cross-border debt flows by the VIX (and its interaction with the GIIPS dummy) as one of the most important supply (push) factors driving their dynamics. In unreported specifications, I establish that, while debt flows in general have a positive relation with domestic interest rates at the country level, debt flows instrumented by the VIX (and its interaction with the GIIPS dummy) do have a negative one. This indicates that, following the above argument, the VIX captures exogenous supply effects that affect debt flows in the euro area.

3.2 IV Results

In this section, I present the 2SLS second-stage instrumental variable results em-ploying the VIX and the interaction between the VIX and a GIIPS dummy as in-struments for debt flows. Column (4) of Table 4 shows that cross-border debt flows still induce an overproportional credit allocation toward firms with lower, relative to higher, returns on assets, as can be gauged from the negative interaction term that is statistically significant at the 5% level. The estimate implies that a 1 pp increase in cross-border debt inflows raises the annual credit growth rates of firms at the 25th 10. The 25th percentile of the distribution of profitability is equal to 0.59 and the 75th percentile is equal to 7.24. Using these values, I obtain the economic effect as follows: (0.59− 7.24) × (−0.038) = 0.253.

percentile of profitability by almost 0.6 pp more than of firms at the 75th percentile. For comparison, the annual credit growth rate in my sample is, on average, 5.1% per year. So a differential of 0.6 pp is a relatively large number. Taking into account the high standard deviation of firm-level credit growth in my sample and considering the 75th percentile of its distribution (20.27%, see Table 2), the economic importance of this differential seems much smaller, but it is still nonnegligible.

Note that the coefficient estimate of column (4) is more than twice as large as the estimate without instrumentation presented in column (3) of Table 4. Given the evidence of the previous section, this amplified impact is not surprising. It shows that especially the exogenous (supply-driven) component of cross-border debt flows leads to a shift in within-industry credit allocation toward ex ante low performing, relative to high performing, firms. While the potentially endogenous, demand-driven component of debt flows was, at least partially, still present in the OLS estimate of column (3), the IV estimate is better able to extract the exogenous, supply-driven component of debt flows, leading to a higher coefficient estimate.

Across the specifications of columns (1)–(4), an increase in firm size reduces firms’ credit growth. Thus, larger firms tend to grow less, which is broadly consistent with the literature on aggregate growth convergence (e.g., Baumol 1986). I also identify a positive relation between the ratio of liquid assets, profitability, and credit growth. The linear impact of profitability on credit growth might seem surprising, but merely indi-cates that, in the absence of cross-border debt inflows, high-profitability firms obtain more credit. This positive relation, however, is reverted in the case of significant debt inflows (indicated by the negative interaction coefficient). Importantly, these results suggest that debt inflows induce an increased credit allocation toward low performing firms, relative to high performing ones, even after controlling for several firm-level covariates. As I show in the Online Appendix, this main result is further robust to several robustness checks in which I employ different measures of firm profitability, vary the time and country coverage of the sample, and differentiate between small and large firms, as well as those in the tradable and nontradable sectors.

For the VIX to be a valid instrument, it does not only need to be a strong predic-tor of debt flows, as shown by the high first-stage F-statistics. In addition, the VIX should only affect firms’ credit growth via international debt flows. This assumption might be violated if the VIX is correlated with firm-level demand for credit, in which case the VIX does not only affect the observable levels of firm credit via higher push-driven debt inflows that increase the domestic supply of funding, but also via changes in domestic credit demand. For instance, to the extent that movements in the VIX are related to the business cycle, the VIX may also affect firms’ credit demand. This does not seem to be the case because, as I argue at the end of Section 3.1, VIX-induced debt inflows lead to a reduction in the domestic interest rate at the country level (re-sult not reported), which is evidence that instrumented debt flows in the euro area are rather driven by supply, as opposed to demand, factors. Yet, in order to strengthen this evidence and to validate the exclusion restriction, I next examine the impact of debt flows on the change in average borrowing costs of firms conditional on firms’ returns on assets. I expect the borrowing costs of low performing, relative to high

performing, firms to decrease since, in this case, consistent with the standard the-ory of loanable funds markets, credit supply effects should dominate credit demand effects. Indeed, column (5) documents that debt inflows reduce the average interest rates of ex ante low performing firms disproportionately more than those of high per-forming ones. Thus, low perper-forming firms, relative to their high perper-forming peers, do not only experience an increase in the volume of credit, but also a reduction in the price of borrowing. This is evidence of a credit supply story and validates the exclu-sion restriction of the instrument, i.e., shows that the VIX is unlikely to affect credit allocation via changes in loan demand.

To sum up, Section 3 shows that cross-border debt flows induce an increased credit supply toward ex ante low performing, relative to high performing, firms. The next section will identify possible channels through which this relation materializes.

4. TRANSMISSION MECHANISMS

Martinez-Miera and Repullo (2017) theoretically show that cross-border capital inflows, driven by an exogenous increase in the supply of savings, suppress domestic interest rate margins and, therefore, lead to a decline in the monitoring intensity of financial market participants and a decrease in the quality of loan portfolios. Alterna-tive theoretical channels generating similar predictions with regard to the relationship between push-driven capital inflows and risk taking, depart from the assumption that capital inflows generate excess liquidity, which aggravates agency problems of finan-cial intermediaries and, in search for yield, leads to softer lending conditions (Dell’ Ariccia and Marquez 2006, Acharya and Naqvi 2012). Consistent with this theoret-ical literature, Section 4 identifies risk-taking and search for yield of the financial sector as important transmission mechanisms of international debt flows to changes in credit allocation. Specifically, cross-border debt inflows seem to induce financial institutions and investors to expand their credit supply to low performing, relative to high performing, firms because these firms are riskier and pay a higher interest rate on their debt.11

Introducing additional firm risk variables in their interactions with cross-border debt flows, the first set of tests, presented in Section 4.1, shows that surges in debt flows are also associated with a disproportionate increase in credit growth of risky firms according to these firm risk measures. I also show that the main interaction of 11. For instance, estimating a duration model to explain firms’ survival time to default, Carling et al. (2007) relate low firm profitability to a higher default probability. In a recent empirical study on banks’ maturity composition, Paligorova and Santos (2017) argue that profitability is one important dimension of firm risk. In my sample, consistent with this argument, the correlation between firm-level profitability and interest rates is equal to−5% (p-value = 0). Note that this relatively low number has to be interpreted in light of the zero interest rate environment and the substantial cross-firm heterogeneity in my sample. Indeed, when I “smooth” this heterogeneity by comparing the median interest rate of the 25% of the least profitable firms to all other sample firms, I find that this spread is economically significant and equal to 60 basis points.

debt flows and firms’ profitability gets economically and statistically less significant once I control for the interaction between debt flows and firm-level interest rates, suggesting that my baseline effects are, to large extents, driven by the higher interest rate of low performing, relative to high performing, firms.

The results presented so far are consistent with a situation where, under the pre-sumption that the financial sector always funds low-risk (high-profitability) firms and lends to riskier (low-profitability) borrowers only when the amount of fund-ing rises, cross-border debt flows increase credit allocation in favor of low perform-ing firms, relative to high performperform-ing ones, by shiftperform-ing up the amount of fundperform-ing available to financial intermediaries and investors. This line of argument reflects a rather mechanical link between the increase in funds’ availability and credit al-location and is not necessarily related to any shifts in risk-taking incentives. In Section 4.2, I show that higher capitalization as a proxy for lower agency prob-lems in the financial sector (Holmstrom and Tirole 1997) significantly reduces the observed relation between debt flows and increased lending to firms with lower, relative to higher, returns on assets, indicating that this relation is not of a purely mechanical nature, but also reflects shifts in the risk-taking behavior of financial sys-tems that are subject to greater agency problems, as suggested by the aforementioned theories.

4.1 Introducing Additional Firm Risk Proxies

To establish the role of risk taking as a mediating channel of cross-border debt flows to shifts in credit allocation, I introduce other firm risk proxies interacted with border debt flows over GDP, in addition to my key interaction between cross-border debt flows and firm-level returns on assets. These firm risk proxies include the ratio of debt divided by the EBIT (in %), which is a key ratio used in evaluating the creditworthiness of a firm and measures the ability of a firm to service its debt (henceforth leverage for brevity). In addition, I employ the inverse of the interest cov-erage ratio (100× interest expenses on debt/EBIT, henceforth just interest coverage for brevity) and the intangible asset ratio (100× intangible assets/total assets), which the literature has shown to be significantly related to the probability of firm default (Ben-Zion and Shalit 1975, Carling et al. 2007, Duchin and Sosyura 2014, Palig-orova and Santos 2017). I hypothesize that foreign debt flows also disproportionately increase credit growth of firms that are risky according to these additional dimensions of firm risk, relative to their safer peers.

Columns (1)–(3) of Table 5 show that the interactions between debt flows and firm leverage and interest coverage, respectively, enter the model with positive and sta-tistically significant coefficients, suggesting that the credit growth rates of levered firms with high interest expenses, relative to firms with lower leverage and interest expenses, are affected disproportionately more by international debt flows. In eco-nomic terms, a 1 pp increase in cross-border debt inflows leads to a 0.19–0.55 pp annual credit growth differential between firms at the 75th percentile and firms at

TABLE 5

The Interaction with Other Firm Risk Variables

(1) (2) (3) (4)

DEBT DEBT DEBT DEBT

DEBT INFLOWS× PROFITABILITY −0.095∗∗ −0.096∗∗ −0.094∗∗ −0.067∗ (0.043) (0.044) (0.043) (0.040) DEBT INFLOWS× LEVERAGE 0.001∗∗∗

(0.000)

DEBT INFLOWS× INTANGIBLE ASSETS −0.020 (0.019)

DEBT INFLOWS× INTEREST COVERAGE 0.007∗

(0.004)

DEBT INFLOWS× INTEREST RATE 0.079∗∗

(0.035) PROFITABILITY 0.557∗∗∗ 0.536∗∗∗ 0.546∗∗∗ 0.430∗∗∗ (0.083) (0.083) (0.084) (0.080) LEVERAGE −0.003∗∗∗ (0.001) INTANGIBLE ASSETS 0.028 (0.061) INTEREST COVERAGE −0.031∗∗∗ (0.010) INTEREST RATE 1.133∗∗∗ (0.074)

Firm-Level Controls Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes

Country-Year Fixed Effects Yes Yes Yes Yes

Industry-Year Fixed Effects Yes Yes Yes Yes

Observations 24023 24001 24023 24028

R-squared 0.186 0.188 0.188 0.223

Notes: In this table, I introduce additional firm risk proxies apart from the returns on assets. The dependent variable is firm-level credit growth. The regressors are country-level lagged debt flows, interacted with firms’ lagged returns on assets, leverage (debt to EBIT), interest coverage ratio (interest expense to EBIT), intangible in total assets and interest rates (interest expenses to total debt). Debt inflows are instrumented with the VIX and its interaction with a an indicator, which is equal to one for Greece, Italy, Ireland, Portugal, and Spain. I control for the lag of various firm variables (logarithm of total assets, liquid assets over current liabilities, the returns on assets), as well as firm, industry-year, and country-year dummies. Standard errors are clustered at the industry-year level and are shown in parentheses. *p< 0.10, **p < 0.05, ***p< 0.01.

the 25th percentile of the distribution of the respective risk variable.12 _{In contrast,}

the interaction between debt flows and firm tangibility is not statistically significant at conventional significance levels. The coefficient estimate on the main interaction between debt flows and firm-level profitability remains negative and statistically sig-nificant in all specifications, confirming the robustness of the estimate to controlling for other firm risk interactions.

The linear effects of leverage and interest expenses over EBIT are negative and sta-tistically significant at the 1% level. Thus, in the absence of debt flows, riskier firms obtain less credit. However, as can be seen from the positive interactions between in-ternational debt flows and both firm risk proxies, this negative impact is mitigated by higher debt inflows and, once they reach a value of 3.0% and 4.4% of GDP, respec-tively, the linear effect of the firm risk proxies on firm debt growth turns positive, 12. The 75th percentile of leverage (interest coverage ratio) is equal to 549.93 (26.96), the 25th per-centile is equal to 0 (0.04). Using these perper-centiles and the respective interaction coefficient, I obtain the economic effects as follows: (549.93− 0) × 0.001 = 0.55 and (26.96 − 0.04) × 0.007 = 0.19.

indicating that riskier firms, following cross-border debt inflows, obtain dispropor-tionately more credit than safer ones.13

These results suggest that cross-border debt flows affect the within-industry al-location of credit through higher risk taking. This firm-level evidence is consistent with the macro-economic literature exploring the relationship between foreign cap-ital flows, financial sector risk and the incidence of crises (e.g., Rancière, Tornell, and Westermann 2008, Reinhart and Rogoff 2008, Obstfeld 2012, Gourinchas and Obstfeld 2012, Lane and McQuade 2014).

The result that surges in cross-border debt inflows are associated with higher credit growth rates of risky, relative to safer, firms raises the question of why financial insti-tutions and investors allocate disproportionately more credit to these firms. One rea-son brought up in the theoretical literature is that higher risk-taking allows financial markets to reach for yield (Martinez-Miera and Repullo 2017). Thus, the overpro-portional rise in credit growth of low performing firms, relative to high performing ones, might be driven by the higher interest rate that they pay relative to high per-forming firms. In fact, the correlation between firm profitability and interest rates in my sample is negative and statistically significant at the 1% level. If the overpropor-tional credit allocation toward low performing firms is also driven by interest rate differences between low and high performing firms, I should find that controlling for the interaction between debt flows and firm-level interest rates, measured by the ra-tio of interest expenses on debt over total debt, reduces the statistical and economic significance of the interaction between debt flows and profitability. Indeed, as can be seen from column (4) of Table 5, saturating the baseline regression of equation (1) with the firm-level interest rate in its interaction with debt flows reduces both the economic and statistical significance of the interaction between debt flows and prof-itability. This is evidence that my baseline effects are, at least to large extents, driven by interest rate differentials between low and high performing firms. In turn, the in-teraction coefficient between debt flows and interest rates is positive and statistically significant at the 5% level, suggesting that the credit growth rates of high-interest rate firms are disproportionately higher than those of low-interest firms.

4.2 The Role of Agency Problems

The theoretical studies of Dell’ Ariccia and Marquez (2006), Acharya and Naqvi (2012), and Martinez-Miera and Repullo (2017) that relate push (supply)-driven in-ternational capital inflows to risk taking and search for yield are based on agency problems in the financial sector. Therefore, I further strengthen the evidence on risk taking and search for yield as transmission channels of cross-border debt flows to cross-firm credit allocation by examining whether the differential impact on the credit growth rates of low performing and high performing firms vanishes when the financial system is less subject to agency problems. Particularly, I focus on agency problems 13. I obtain these thresholds by dividing the linear coefficient on LEVERAGE (INTEREST COVER-AGE) by the interaction between DEBT INFLOWS and LEVERAGE (INTEREST COVERCOVER-AGE).

TABLE 6

The Role of Agency Problems

Normal Capitalization High Capitalization

(1) (2)

DEBT DEBT

DEBT INFLOWS× PROFITABILITY −0.055∗ 0.212∗

(0.033) (0.113)

PROFITABILITY 0.471∗∗∗ 0.313

(0.078) (0.213)

Firm-Level Controls Yes Yes

Firm Fixed Effects Yes Yes

Country-Year Fixed Effects Yes Yes

Industry-Year Fixed Effects Yes Yes

Observations 20097 2897

R-squared 0.214 0.331

Notes: This table depicts the regression results corresponding to equation (1), conditioning the effects on bank’s capital-to-asset ratio. In column (1), I show the effect for banking sectors with normal capitalization and column (2) restricts the sample to banking sectors in the top 25% of the distribution. The dependent variable is firm credit growth. The key regressor is lagged country-level debt flows interacted with lagged firm profitability. Debt flows are instrumented with the VIX and its interaction with a dummy that is equal to one for Greece, Italy, Ireland, Portugal, and Spain. I control for the lag of firm covariates (log of total assets, liquid assets over current liabilities, the returns on assets), and firm, industry-year, and country-year fixed effects. The standard errors are clustered at the industry-year level and are shown in parentheses. *p< 0.10, **p < 0.05, ***p < 0.01.

in the banking sector, which is not only the main provider of loans, but also a large debt security holder in the euro area.14

As the measure of bank agency problems, I employ the average country-level cap-italization of the domestic banking sector, retrieved from the World Bank’ s World Development Indicators. As shown by Holmstrom and Tirole (1997), poorly capital-ized banks do not sufficiently internalize their risk of default, and are thus more prone to excessive risk taking. This theoretical hypothesis has been confirmed empirically by Jiménez et al. (2014) and Ioannidou, Ongena, and Peydró (2015), among oth-ers. Equipped with this measure of agency problems, I estimate equation (1) for two distinct subsamples. First, banking sectors with normal capitalization that are likely to be subject to agency problems, which I define as having a capital-to-asset ratio in the lowest 75% of the distribution. Second, well-capitalized banking sectors that are unlikely to be subject to agency problems that I define as having capitalization in the top 25% of the distribution.15 Consistent with the aforementioned literature, I hypothesize that the disproportionate increase in credit growth of low performing firms, relative to high performing ones, indicated by a negative interaction term be-tween debt flows and firms’ returns on assets, is weakened if banking sectors have high capitalization.

As can be gauged from columns (1) and (2) of Table 6, cross-border debt flows only raise the credit growth rates of low performing firms disproportionately more than those of high performing ones if banking sectors have a capital ratio in the

14. For instance, see Timmer (2018) for evidence on Germany.

15. The World Bank reports data on banking sector capital-to-asset ratios at the country level starting in 2005. Therefore, if those data are missing for a particular year and country, I treat all firm observations of that country and year as being subject to banking sectors with normal capitalization.

lowest 75% of the distribution, i.e., when agency problems in the banking sector are likely to be present. In contrast and as opposed to all previous results, international debt flows are even associated with an overproportional increase in lending to high performing, relative to low performing, firms if the degree of agency problems in the banking sector is low, despite a lower relevance of my instruments in this subsample. The difference in the estimates between both subsamples is statistically significant at the 5% level, suggesting that the relation between debt flows and credit allocation, documented in this paper, indeed seems to reflect a shift in the risk-taking behavior of financial systems that are subject to greater agency problems.

5. CONCLUDING REMARKS

Using a euro area firm-level data set, this paper explores the impact of cross-border debt flows on credit allocation. Instrumenting debt flows by global uncertainty, I find that surges in international debt inflows have a differential impact on the credit growth rates of low performing and high performing firms. Particularly, a low performing firm at the 25th percentile of the profitability distribution has a 0.6 pp higher annual credit growth rate relative to a firm at the 75th percentile for each 1 pp increase in debt inflows. I further document that risk taking and search for yield of the financial sector are important transmission channels of cross-border debt flows to credit allocation: low performing firms have a higher probability of default and they pay higher interest rates on their debt, which seems to allow financial intermediaries to reach for yield. These effects are driven by domestic banking sectors that have lower capitalization.

These results suggest that one reason for the low (and even negative) correlation between debt inflows and economic growth in the euro area is that credit within the nonfinancial business sector is reallocated from high performing toward low perform-ing firms. This evidence is broadly consistent with the misallocation literature that identifies increases in the dispersion of the return to capital within industries and productivity losses from misallocation associated with capital flows into countries in South Europe (Gopinath et al. 2017).

Finally, I would like to note that my sample only contains countries within a mone-tary union and, hence, subject to fixed exchange rates. However, as shown by Magud, Reinhart, and Vesperoni (2014) or Obstfeld, Ostry, and Qureshi (2019), the link be-tween capital inflows and domestic credit growth is significantly stronger in countries subject to fixed compared to flexible exchange rate regimes. Given this evidence, it would be interesting to study the relation between cross-border debt flows and credit allocation for a sample that covers firms in countries with fixed and flexible exchange rate regimes, and to see whether the impact of debt flows on cross-firm credit dy-namics depends on the exchange rate regime. I leave this interesting question for future research.

LITERATURE CITED

Acharya, Viral, and Hassan Naqvi. (2012) “The Seeds of a Crisis: A Theory of Bank Liquidity and Risk-Taking over the Business Cycle.” Journal of Financial Economics, 106, 349–66. Aizenman, Joshua, Yothin Jinjarak, and Donghyun Park. (2013) “Capital Flows and Economic

Growth in the Era of Financial Integration and Crisis, 1990-2010.” Open Economics Review, 24, 371–96.

Bank for International Settlements. (2015) “85th Annual Report.” Basel.

J Baumol, William. (1986) “Productivity Growth, Convergence, and Welfare: What the Long-Run Data Show.” American Economic Review, 76, 1072–85.

Ben-Zion, Uri, and Sol S. Shalit. (1975) “Size, Leverage, and Dividend Record as Determinants of Equity Risk.” Journal of Finance, 30, 1015–26.

Benigno, Gianluca, and Luca Fornaro. (2014) “The Financial Resource Curse.” Scandinavian

Journal of Economics, 116, 58–86.

Benigno, Gianluca, Nathan Converse, and Luca Fornaro. (2015) “Large Capital Inflows, Sec-toral Allocation, and Economic Performance.” Journal of International Money and Finance, 55, 60–87.

Berger, Allen N., and Gregory F. Udell. (2002) “Small Business Credit Availability and Re-lationship Lending: The Importance of Bank Organisational Structure.” Economic Journal, 112, 32–53.

Bonfiglioli, Alessandra. (2008) “Financial Integration, Productivity and Capital Accumulation.” Journal of International Economics, 76, 337–55.

Borio, Claudio, Robert McCauley, and Patrick McGuire. (2011) “Global Credit and Domestic Credit Booms.” BIS Quarterly Review, September. Available at https://www.bis.org/publ/ qtrpdf/r_qt1109f.htm

Bruno, Valentina, and Hyun Song Shin. (2015) “Cross-Border Banking and Global Liquidity.”

Review of Economic Studies, 82, 535–64.

Bussière, Matthieu, and Marcel Fratzscher. (2008) “Financial Openness and Growth: Short-Run Gain, Long-Short-Run Pain?” Review of International Economics, 16, 69–95.

Calvo, Guillermo A., Leonardo Leiderman, and Carmen M. Reinhart. (1996) “Inflows of Cap-ital to Developing Countries in the 1990s: Causes and Effects.” Journal of Economic

Per-spectives, 10, 123–39.

Carling, Kenneth, Tor Jacobson, Jesper Lindé, and Kasper Roszbach. (2007) “Corporate Credit Risk Modeling and the Macroeconomy.” Journal of Banking and Finance, 31, 845–68. Cerutti, Eugenio, Stijn Claessens, and Damien Puy. (2019) “Push Factors and Capital Flows to

Emerging Markets: Why Knowing Your Lender Matters More than Fundamentals.” Journal

of International Economics, 119, 133–49.

Giovanni Dell’, Ariccia,, and Robert Marquez. (2006) “Lending Booms and Lending Stan-dards.” Journal of Finance, 61, 2511–46.

di Giovanni, Julian, Sebnem Kalemli-Özcan, Mehmet F. Ulu, and Yusuf S. Baskaya. (2019) “International Spillovers and Local Credit Cycles.” Federal Reserve Bank of New York. Dinger, Valeriya, and Daniel M. te Kaat. (2020) “Cross-Border Capital Flows and Bank

Risk-Taking.” Journal of Banking & Finance, 117, 105842.

Duchin, Ran, and Denis Sosyura. (2014) “Safer Ratios, Riskier Portfolios: Banks’ Response to Government Aid.” Journal of Financial Economics, 113, 1–28.

Edison, Hali J., Ross Levine, Luca Ricci, and Torsten Slok. (2002) “International Financial Integration and Growth.” Journal of International Money and Finance, 21, 749–776. Eichengreen, Barry, Rachita Gullapalli, and Ugo Panizza. (2011) “Capital Account

Liberal-ization, Financial Development and Industry Growth: A Synthetic View.” Journal of

Inter-national Money and Finance, 30, 1090–1106.

European Central Bank. (2013) “ECB Monthly Bulletin July 2013.” Frankfurt a. M. Forbes, Kristin J., and Francis E. Warnock. (2012) “Capital Flow Waves: Surges, Stops, Flight,

and Retrenchment.” Journal of International Economics, 88, 235–51.

Fratzscher, Marcel. (2012) “Capital Flows, Push versus Pull Factors and the Global Financial Crisis.” Journal of International Economics, 88, 341–56.

Fratzscher, Marcel, Marco Lo Duca, and Roland Straub. (2018) “On the International Spillovers of US Quantitative Easing.” Economic Journal, 128, 330–77.

Ghosh, Atish R., Jun Kim, Mahvash S. Qureshi, and Juan Zalduendo. (2014) “Surges.” Journal

of International Economics, 92, 266–85.

Gopinath, Gita, Sebnem Kalemli-Özcan, Loukas Karabarbounis, and Carolina Villegas-Sanchez. (2017) “Capital Allocation and Productivity in South Europe.” Quarterly Journal

of Economics, 132, 1915–67.

Gourinchas, Pierre-Olivier, and Maurice Obstfeld. (2012) “Stories of the Twentieth Century for the Twenty-First.” American Economic Journal: Macroeconomics, 4, 226–65. Greenstone, Michael, Alexandre Mas, and Hoai-Luu Nguyen. (2020) “Do Credit Market

Shocks Affect the Real Economy? Quasi-experimental Evidence from the Great Recession and ‘Normal’ Economic Times.” American Economic Journal: Economic Policy, 12, 200– 25.

Grilli, Vittorio, and Gian Maria Milesi-Ferretti. (1995) “Economic Effects and Structural De-terminants of Capital Controls.” IMF Staff Papers, 42, 517–51.

Holmstrom, Bengt, and Jean Tirole. (1997) “Financial Intermediation, Loanable Funds, and the Real Sector.” Quarterly Journal of Economics, 112, 663–91.

Houston, Joel F., Jennifer Itzkowitz, and Andy Naranjo. (2017) “Borrowing Beyond Borders: Foreign Assets, Lender Choice, and Loan Pricing in the Syndicated Bank Loan Market.”

Journal of Corporate Finance, 42, 315–34.

Ioannidou, Vasso, Steven Ongena, and José-Luis Peydró. (2015) “Monetary Policy, Risk-Taking, and Pricing: Evidence from a Quasi-Natural Experiment.” Review of Finance, 19, 95–144.

Jiménez, Gabriel, Steven Ongena, José-Luis Peydró, and Jesús Saurina. (2014) “Hazardous Times for Monetary Policy: What Do Twenty-Three Million Bank Loans Say about the Ef-fects of Monetary Policy on Credit Risk-Taking?” Econometrica, 82, 463–505.

Kose, M. Ayhan, Eswar S. Prasad, and Marco E. Terrones. (2009) “Does Openness to Inter-national Financial Flows Raise Productivity Growth?” Journal of InterInter-national Money and

Finance, 28, 554–80.

La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. (2000) “Agency Problems and Dividend Policies around the World.” Journal of Finance, 55, 1–33. Lane, Philip R. (2013) “Capital Flows in the Euro Area.” Economic Papers 497. Available at https://ec.europa.eu/economy_finance/publications/economic_paper/2013/pdf/ ecp497_en.pdf

Lane, Philip R., and Gian Maria Milesi-Ferretti. (2017) “International Financial Integration in the Aftermath of the Global Financial Crisis.” IMF Working Paper, 17/115.

Lane, Philip R., and Peter McQuade. (2014) “Domestic Credit Growth and International Cap-ital Flows.” Scandinavian Journal of Economics, 116, 218–52.

Leuz, Christian, Dhananjay Nanda, and Peter D. Wysocki. (2003) “Investor Protection and Earnings Management: An International Comparison.” Journal of Financial Economics, 69, 505–27.

Magud, Nicolas E., Carmen M. Reinhart, and Esteban R. Vesperoni. (2014) “Capital Inflows, Exchange Rate Flexibility and Credit Booms.” Review of Development Economics, 18, 415– 30.

Martinez-Miera, David, and Rafael Repullo. (2017) “Search for Yield.” Econometrica, 85, 351–78.

Miranda-Agrippino, Silvia, and Hélène Rey. (2020) “US Monetary Policy and the Global Fi-nancial Cycle.”Review of Economic Studies. Available at https://academic.oup.com/restud/ advance-article/doi/10.1093/restud/rdaa019/5834728

Obstfeld, Maurice. (2012) “Does the Current Account Still Matter?”American Economic

Re-view, 102, 1–23.

Obstfeld, Maurice, and Jonathan D. Ostry, and Mahvash S. Qureshi. (2019) “A Tie that Binds: Revisiting the Trilemma in Emerging Market Economies.” Review of Economics and

Statis-tics, 101, 279–93.

Paligorova, Teodora, and Joao A. C. Santos. (2017) “Banks’ Exposure to Rollover Risk and the Maturity of Corporate Loans.” Review of Finance, 21, 1739–65.

Rajan, Raghuram G., and Luigi Zingales. (1998) “Financial Dependence and Growth.”

Amer-ican Economic Review, 88, 559–86.

Rancière, Romain, Aaron Tornell, and Frank Westermann. (2008) “Systemic Crises and Growth.” Quarterly Journal of Economics, 123, 359–406.

Reinhart, Carmen M., and Kenneth S. Rogoff. (2008) “Is the 2007 US Sub- prime Financial Crisis So Different? An International Historical Comparison.” American Economic Review, 98, 339–44.

Reis, Ricardo. (2013) “The Portuguese Slump and Crash and the Euro Crisis.” Brookings

Pa-pers on Economic Activity, 46, 143–93.

Rey, Hélène. (2013) “Dilemma not Trilemma: The Global Financial Cycle and Monetary Pol-icy Independence.” Federal Reserve Bank of Kansas City Economic PolPol-icy Symposium. Samarina, Anna, and Dirk Bezemer. (2016) “Do Capital Flows Change Domestic Credit

Allo-cation?” Journal of International Money and Finance, 62, 98–121.

Timmer, Yannick. (2018) “Cyclical Investment Behavior across Financial Institutions.”

Jour-nal of Financial Economics, 129, 268–86.

John Wei, K. C., and Yi Zhang. (2008) “Ownership Structure, Cash Flow, and Capital In-vestment: Evidence from East Asian Economies before the Financial Crisis.” Journal of

Corporate Finance, 14, 118–32.

SUPPORTING INFORMATION

Additional supporting information may be found online in the Supporting Infor-mation section at the end of the article.

Table A1: The Linear Effect of Debt Flows Table A2: Robustness Tests (1)

Table A3: Robustness Tests (2) Table A4: Robustness Tests (3)

Figure A1: This figure plots the monthly VIX over time. Supplementary Material