Firms’ credit risk and the onshore transmission of the global financial cycle

BIS Working Papers

No 712


Firms’ credit risk and the onshore transmission of the global financial cycle

by Ramon Moreno and Jose Maria Serena-Garralda


Monetary and Economic Department

March 2018


JEL classification: F24, F36, G15, G30

Keywords: International bonds, Credit Risk, Global risk appetite, Firm-level data


This publication is available on the BIS website (


© Bank for International Settlements 2017. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated.

ISSN 1020-0959 (print)

ISSN 1682-7678 (online)


Firms’ credit risk and the onshore transmission of the global financial cycle

Ramon Moreno and Jose Maria Serena-Garralda



We investigate the role of firms’ credit risk in the onshore transmission of international bond market conditions. We show that reductions in the global price of risk, measured by the excess bond premium, encourage more international bond borrowing by smaller and younger firms. Due to informational asymmetries, these firms pay a higher credit spread. Thus their funding costs, and consequently their international borrowing, are more tightly linked to the global price of risk. The funds borrowed in response to favourable market conditions cause their balance sheets to deteriorate; over a three-year horizon, leverage increases, in support of capital expenditure, and cash holdings increase. Our results reveal a micro-level link between rising global risk appetite and the gradual build-up of domestic vulnerabilities.

Keywords: International bonds, Credit Risk, Global risk appetite, Firm-level data

JEL Codes: F24, F36, G15, G30.

Bank for Intenational Settlements, Centralbahnplatz 2, Basel (Switzerland). Corresponding author:

We thank Mirko Abbritti, Stefan Avdjiev, Claudio Borio, Stijn Claessens, Ben Cohen, Romain Duval, Torsten Ehlers, Angel Estrada, Alberto Fuertes, Galina Hale, Branimir Gruić, Enisse Kharroubi, Gian Maria Milesi-Ferretti, Antonio Moreno, Enrique Moral-Benito, Ugo Panizza, Hyun Song Shin, Ricardo Sousa, Tommaso Trani, Philip Turner, Christian Upper, Adrian Van Rixtel, and seminar participants at the BIS, Bank of Spain, and University of Navarre for helpful comments and suggestions. All remaining errors are our responsability.


1 Introduction

Cross-border capital flows are highly responsive to global financial conditions. Recent research has established that lending by international investors strongly depends on global risk appetite. This general pattern holds in international bond markets: firms raise large amounts of financing from international investors when global financial conditions ease; their borrowing sharply declines when conditions tighten.

The capital structure literature suggests, however, that certain firm-specific features could strongly influence how global conditions are transmitted across borders. Previous research has already iden­tified some robust patterns. A first finding is that larger and better capitalised banks, with higher non-core liabilities, more actively transmit global shocks into domestic lending (Baskaya et al. [2017a], Baskaya et al. [2017b]). On the borrower side, it is well established that exchange rate appreciations, a factor transmitting global conditions across borders, ease the financial constraints of firms with currency mismatches (Shin [2015], Kalemli-Ozcan et al. [2018], Avdjiev et al. [2018a]). This happens since collateral constraints are disproportionately relaxed for these firms, due to improvements in net worth.

Poor information constitutes an additional source of credit risk, and it is often independent on firms’ capital structure. It can represent an important constraint to borrowing in international bond markets, since institutional investors have less monitoring ability than global banks (Diamond [1984]). Informational asymmetries are severe, for instance, for small and young firms, as they are often unknown to lenders. Since credit risk assessment is difficult, these firms need to pay a premium to issue debt. Consequently, for small and young firms the cost of raising external debt fluctuates with the price of risk. This cost is particularly low when global investors’ ability or willigness to absorb risk is high. In the case of large and old firms, the cost of raising debt does not fluctuate much, since they do not pay a high premium in international markets.

This suggests the following testable hypothesis: declines in global risk appetite encourage interna­tional bond borrowing by firms perceived as riskier, due to their size and age, since they generally pay higher credit spreads to access international markets. This mechanism is complementary to the impact that improvements in firms’ net worth might have on their funding costs; fluctuations in the global price of risk have a direct and stronger impact, we presume, on the funding costs of firms with low credit quality. As we discuss our measures of credit risk reflect uncertainty as to the value of firms, and are unrelated to firms’ capital structure.

The purpose of this paper is to examine this hypothesis using of a firm-level cross-country dataset covering companies headquartered in 40 small open economies, over the period 2000-2015. Our dataset matches companies’ financial statements with their bond-level issuance in international mar- kets. These bond-level data are, on an aggregate basis, similar to the corresponding series in the BIS international debt securities (BIS IDS) statistics. Thus we are investigating firms’ cross-border financing using a micro-level dataset.

To test the hypothesis we adapt an empirical model of determinants of change in leverage, in line with the literature on capital structure (Rajan and Zingales [1995], Baker and Wurgler [2002], Farre- Mensa and Ljungqvist [2016]). We verify whether changes in global financial conditions have a larger effect on international borrowing by firms perceived as riskier by international investors.

The dependent variable is the total amount of international bond issuance by a firm relative to its total assets in the previous year. Global risk appetite is measured with the excess bond premium proposed by Gilchrist and Zakrajsek [2012]. It is the price of bearing exposure to risk in the US corporate bond markets, obtained by substracting from credit spreads the portion due to expected default risk. Low levels of the excess bond premium reflect strong risk appetite, net of reductions in firms’ credit risks which might occur simultaneously.

We use the Hadlock-Pierce index (Hadlock and Pierce [2010]) as an indicator of informational asym­metries between firms and global investors. This index is constructed using simultanously firms’ size and age. The higher values correspond to young and small firms, which are often unknown and thus riskier borrowers. The Hadlock-Pierce index is independent on the amount or quality of firms’ collateral, and as such it is particularly suited to test our hypothesis. The ratio of tangible assets is sometimes used as a measure of asymmetric information; however, empirically it is not a factor sig­naling constraints (Hadlock and Pierce [2010]), and the correlation with our measure is close to zero. To isolate supply from demand shocks in the model we include several controls for firms’ demand for finance: tangible assets, size, profitability, and lagged leverage; and also country and industry-fixed effects, and a number of country-specific controls.

In our benchmark analysis we estimate a nonlinear (Tobit) model over subsamples of firms perceived as riskier (and consequently with strong financing constraints in international markets); and as safer (and thus with weaker constraints). The subsample of strongly (weakly) constrained firms comprises the top (bottom) 25% of firms, classified according to the Hadlock-Pierce index (since higher values reflect stronger constraints). To address the various problems of defining subsamples using ad-hoc thresholds, we interact the Hadlock-Pierce index with the excess bond premium. We also conduct subsample estimations with formally identified thresholds, adapting the likelihood-ratio tests applied by Hansen [1999]. The results are stronger when we exclude countries Switzerland and, in particular, the United Kingdom, whose firms are more financially integrated in global markets.

The results are consistent with our expectation. International borrowing by firms with a very con­strained access to global markets strongly respond to shifts in global lending standards; the size of the response decreases as firms become less constrained; and it is negligible for firms with weak con­straints. Similarly in the threshold estimation we find that the impact of global financial conditions is much stronger for firms identified as constrained.

We conduct a number of robustness checks. We check the sensitivity of our results to: (1) the exclusion of energy and utility firms, whose funding patterns in international markets can be related to their foreign currency asset-side exposures; (2) alternative measures of global financial conditions -VIX, MOVE, and US dollar term premium; and (3) alternative measures of constraints in global markets: firm-size, the Whited-Wu index, and the Kaplan-Zingales index. The results are broadly consistent with our hypothesis. As a robustness check, we use the number of analysts covering each company to assess if firms’ access to international markets is constrained; firms with less coverage are less well known, and thus more constrained. We reach similar conclusions. Next we demonstrate that risk appetite does not translate into stronger issuances in domestic corporate bond markets. This underscores that the impact of risk appetite occurs through additional offshore borrowing.

Our measure of firms’ credit risks is independent on their capital structure. Accordingly risk-taking by global investors are not apparent on ex-ante balance-sheet metrics. But balance-sheet vulnera­bilities can build up later on. Thus we examine how international bond issuance impacts on firms’ balance-sheets using a standard model (Kim and Weisbach [2008], Bruno and Shin [2015]). Our objective is to gauge if global risk appetite has a stronger impact on the balance-sheets of firms with a more constrained access to global markets due to their uncertain credit quality; these are the firms borrowing offshore when global financing conditions are favourable. The evidence is consistent with the hypothesis: international borrowing raised in response to strong risk appetite has a higher impact on smallar and younger firms: leverage increases, resulting in higher capital expenditures and particularly cash holdings. These results suggest there is a link between greater global risk appetite in international bond markets, and the gradual build-up of domestic vulnerabilities.

Our paper is primarly a contribution to the literature on cross-border capital flows. Recent research has shown that extreme swings in cross-border gross capital inflows are caused by global financial factors. There is a subsequent literature using micro-level data to investigate how global finan­cial conditions are transmitted into domestic economies. This literature has underscored that the transmission through the domestic banking sector strongly depends on some cross-sectional aspects such as bank-specific attributes (Baskaya et al. [2017a]); or certain features of domestic banks’ loan portfolio (De Jonghe et al. [2016]). Research on the transmission through global corporate bond markets has highlighted the importance of carry-trade incentives; certain features make firms more likely to engage in such activity: ex ante high cash holdings (Bruno and Shin [2017]); debt capacity (Acharya and Vij [2017]); or ability to arbitrage capital controls (Caballero et al. [2016]). Currency mismatches also enhance the positive impact of exchange rates appreciations on leverage due to net worth effects (Kalemli-Ozcan et al. [2018]). Our contribution to this literature consists in showing that risk-taking by international investors is also transmitted into local economies through additional borrowing by firms of higher credit risk. The findings are consistent with the patterns identified in the US economy. Greenwood and Hanson [2013] show that issuers’ credit quality deteriorates in response to decreases in the pricing of credit risk in US corporate bond markets. A difference is that they measure issuers’ credit quality with ratings and EDF; instead we exploit measures unrelated to capital structure (size, age, and analysts coverage). The variables they use are encompassing mea­sures of credit risk, which account for balance-sheet vulnerabilities; data limitations prevent us from conducting that exercise. Our findings are thus more specific: we show a deterioration in issuers’ quality which would go unnoticed in ratio analyses.

The rest of the paper is structured as follows. Section 2 reviews related literature and discusses the theoretical rationale. Section 3 describes the data. Section 4 discusses the empirical setup. Section 5 examines the influence on constraints on international borrowing. In section 6 we examine the impact on capital structure. Finally, section 7 summarizes the main conclusions.

2 Constrained access to international markets

Literature review

This paper is related to two strands of the literature: drivers of capital flows, and capital structure.

Most recent research on capital flows has underscored the importance of global financial factors as driver of gross capital flows (Forbes and Warnock [2012],Broner et al. [2013], Miranda-Agrippino and Rey [2013], Bruno and Shin [2015] , Milesi-Ferretti and Tille [2011], McCauley et al. [2015] , Lo-Duca et al. [2016], Feyen et al. [2015] or Rodriguez-Bastos et al. [2015]). However Cerruti et al. [2017] find that the quantitative importance of common shocks is not that large. Besides some papers have found that this importance has varied over time, and highlighted it depends on policy initiatives on creditor countries (Avdjiev et al. [2017]); or on country-level fundamentals during stress (Fraztchter [2012]). These findings underscore the importance of understanding the micro-level link between global conditions and capital flows.

In response the literature is growingly using firm-specific data to investigating the issue, which is also the approached follow in this paper. Several studies use country-level data to better investigate aspects of the transmission through the banking sector. Baskaya et al. [2017a] use Turkish data to show that the transmission through the domestic banking sector depends on bank-specific attributes: positive shocks have a positive impact on lending, which is stronger for larger, more capitalised banks, with higher non-core liabilities. Baskaya et al. [2017b] find that this impact is reflected on lower rates, since banks with higher non-core funding price loans more aggressively. De Jonghe et al. [2016] use data for Belgium and show that banks credit reallocation after an external funding shock is tilted towards low-risk firms and core sectors. Ongena et al. [2015] construct a matched firm-bank dataset for 14 emerging economies and find that international shocks spread to domestic credit; though, credit-dependent firms suffer more from global financial shocks. There are several papers that match firm and bond level data to investigate companies’ bond issuance decisions. On a cross-country setup, Bruno and Shin [2017] highlight that interest rate differentials and exchange rate strength increase firms’ propensity to hoard cash; they find that firms with high cash holdings are more likely to borrow in response to dollar carry-trade opportunities. Caballero et al. [2016] underscore the relevance of capital controls as determinant of such carry trade activity. Acharya and Vij [2017] use detailed data on Indian companies and find that low-leverage firms are more likely to respond to easy global financial conditions by issuing foreign currency liabilities. On the other hand, Kalemli-Ozcan et al. [2018] use a large sample of Asian firms, and highlight the importance of the net-worth effects of exchange rate appreciations: firms with higher foreign currency debt increase their leverage more in response to appreciations. Finally, Calomiris et al. [2018] investigate the impact of foreign equity inflows on firms’equity issuances; they find that inflows have a stronger impact on issuances by large and liquid firms.

Our research is also related to the literature investigating the impact of credit supply shocks on firms’ borrowing decisions. Some papers have explored the impact of shocks to specific sources of capital. For instance Lemmon and Roberts [2010] investigate US firms’ response to exogenous contractions in the supply of below-investment-grade credit; below-investment grade firms responded by decreasing net issuances, since were not able to substitute towards alternative sources of funding. Leary [2009] shows that bank-dependent firms decreased their net funding in response to bank loans contraction, whereas firms with access to public bond markets are resilient. Our paper is particularly close to Greenwood and Hanson [2013], that explore shifts in the pricing of credit risk; they demonstrate that in the US economy this type of supply shock has a stronger impact on borrowing by low-quality firms -measured with credit ratings and EDFs.

Hypothesis development

Our hypothesis is that global risk appetite particularly fosters the international borrowing of firms which, as a result of their credit risk, have a more constrained access to international markets.

We emphasize the role of informational asymmetries as a source of constraints. International investors find difficult to assess the credit risk of some firms -young and small. As a result, they require a higher premium for bearing additional credit risk. This premium is reflected in an upward-sloping supply of external funds -ie it reveals a constrained access to markets. When the price that investors charge for bearing risk decreases (risk aversion is low), the slope of the supply of funds flattens. The funding cost of risky firms falls, and they react by increasing their borrowing. In contrast low-risk firms do not borrow additional funds in response to reductions in the price of credit risk. Since they can always raise additional debt without paying an extra premium, the price of risk does not affect their funding costs.

Our expectation is that shifts in global risk appetite impact on international borrowing, and not on domestic or aggregate borrowing. It derives from the following assumption: investors in international and domestic corporate bond markets are segmented.

The assumption means that international investors do not buy corporate bonds issued in domestic markets. Thus constrained firms need to issue in international markets to benefit from low global risk aversion.

Our assumption is not implausible. A similar segmentation of investors with different preferences in credit markets has been assumed by some recent literature (Liao [2016], Ma [2016]). The evidence suggests that international investors exibit home bias, and are unlikely to buy corporate bonds issued in the local bond markets of small open economies (Maggiori et al. [2018]). Moreover international and domestic securities differ in important dimensions: currency, maturity, or average size (Gozzi et al. [2015]); this suggests that international and local investors have different preferences, and are effectively segmented.

Moreover, the bias introduced if the assumption does not hold goes against finding results. If international investors buy bonds issued in domestic corporate markets, stronger risk appetite should foster domestic bond issuance. The channel we suggest would be weaker.

3 Data

3.1 Firm-level data

We use a firm-level panel for the period 2000-2015. The original sample comprises firms listed in the main equity indexes worldwide. The total number of firms is 10,917 companies. After removing financial companies, the total number of firms in our sample is 8,708. We further restrict the sample to focus on firms headquartered in small open economies. The final number of firms is 2,781 firms, based in 40 countries, emerging and advanced.

For each company we record the items included in the consolidated financial statements, as well as key non-financial information, such as country of incorporation of the issuer, nationality of the firm, sector, year of incorporation, among others. Besides we retrieve several common identifiers: CUSIP, ISIN, SEDOL, LEI, equity ticker and the organization ID. All of them are unique identifiers: each firm has its own identifer, and viceversa. However they are not always available.

We determine the nationality of the firm using the country of headquarters of the company. The jurisdictions covered include Latin America, Emerging Europe, Africa and Middle East, Emerging asia, and the universe of small advanced economies. We classify firms’ industry by the sector of the guarantor, using the MSCI-GICS classification. Firms are classified into 9 groups: Materials, Communications, Consumer Discretionary, Consumer Staples, Energy, Health Care, Industrials, Ma­terials, Technology, Utilities. The cross-tabulation of firms by region and industry is shown in Table 1.

In the main analyses we use all firms. In robustness checks we remove firms with missing data in more than half of the years, which leaves us with 1,928 firms. The qualitative results shown below do not change when we use this alternative, balanced sample.

3.2 Bond-level data

We collect data on all bonds issued by non-financial firms based in the 40 jurisdictions covered, in the period 2000-2015. We measure firms’ international bond borrowing on a consolidated basis. For each firm we gather all the bonds guaranteed, even when they have been issued by its affiliates.

More specifically we assume a bond is ultimately a liability of the company with an explicit legal exposure to it (ie legally liable in case of default); if this information is not available we assume it is the ultimate parent company of the issuer; if the latter is not available, we assume it is the direct issuer. This rule differs from consolidation at the ultimate parent level when firms have affiliates that guarantee liabilities, acting as standalone entities. Appendix A provides an example. Bond- level data are obtained from Reuters. We retrieve the aforementioned common identifiers for each company: CUSIP, ISIN, SEDOL, LEI, equity ticker, and the organization ID.

Besides we obtain certain non-financial information (industry, country of incorporation, etc). We use it to classify bonds according to the nationality and industry of the company. A debt issuance is  classified as domestic when the company issues in its local bond market, and as international when it does not. We classify bonds according to the primary market of issuance, following the methodology of the BIS international debt securities (BIS IDS).

In our sample selection we leave aside firms based in the major economies. Consequently there is a clear distinction between domestic and international bond issuances. We implicitely assume that fluctuations in global risk appetite do not directly impact on funding costs in domestic corporate bond issuances, since we assume segmentation between investors in domestic and international markets. Overall we identify 49,094 bonds with a total value of 5.5 USD trillion.

While we exploit this data on a firm-level basis, the aggregate volume is roughly similar in size to the corresponding series in the BIS IDS (Figure 1). The differences between both series can be explained by differences in data coverage, and by some methodological differences.

Table 2 summarizes the patterns of bond issuance by region. Our focus is on international bonds, marketed outside the local jurisdiction. Panel A shows that, throughout the period, firms have issued a total of 3.2 USD trillion in international markets. International bond issuance outweights domestic placements in terms of volume in most countries, with the exception of Emerging Asia.

The bulk of international bond issuance is denominated in foreign currency. But the breakdown by region shows that some of the firms based in advanced economies issue overseas in domestic currency.

In terms of number of bonds, the pattern is slightly different: there are more deals in local makets, underscoring a smaller average size (Table 2, Panel B). This pattern is similar in all regions, although sharper in Asia -where local markets are more developed.

Table 3 summarizes the patterns of funding by industry. There are remarkable differences accross industries, which seem related to their (technological) demand for external finance. Industries with more needs of external funds -such as energy, industry, and utilities companies- raise more funds than the others. Energy firms raise a substantial fraction of total international bonds, mostly in foreign currency. The bulk of international bond issuances in local currency is carried out by utilities and industrials.

3.3 Merging the datasets

In both datasets we have retrieved firms’ information on a consolidated basis. Thus we can merge them using the common identifiers of the company, which are available in both datasets. We have different options, since we have retrieved several identifiers.

We illustrate the procedure with Korea Electric Power Corp. This company is included, for instance, in the Bloomberg World Index -its weight as of October 2016 was 0.063994%. As a result the company is part of our sample, and we have detailed information on financial statements, cash-flow statements, and non-financial items. The company has the following common identifiers: CUSIP, Y48406105; ISIN, KR7015760002; SEDOL, 6495730; LEI, 988400QG33WS7JEIV602; ticker, 015760; and orga­nization ID, 4295881588. In this particular case we can use any of them to merge both datasets. In the period covered, Korea Electric Power Corp has backed 51 bonds issued in international markets, which amount for 16 US billion.

However, some common identifiers are not always available. For instance CUSIP, SEDOL and ISIN are missing in around 10% of the firms. This percentage increases to 53% in the case of the LEI. In contrast the ticker and the organization ID are available for all firms. Thus we opt to use the latter.

The 2,781 firms in our sample have issued 58% of the 5.5 US trillion guaranteed by firms based in the countries covered in the analysis. The total number of bonds they have guaranteed is 17,005, and the total amount guaranted is 3.2 US trillion. 2 US trillion were issued in international markets, and the remaining 1.2 US trillion in domestic markets. The proportion of the total amount issued in international markets by firms in our sample rises to 62% of the total, and the total number of international bonds is 4,507 (45% of the total).

To make sure we are not losing meaningful information, we examine the type of firms guaranteeing international bonds issued in the period, but not included in our sample of companies. Since we are covering listed firms, we expect the majority of these firms to be unlisted companies.We sort the 2,809 companies which have guaranteed international bonds and are not in our sample according to the total amount they have backed. We focus on the 1,091 companies which have issued bonds for an amount higher than 100 US million. These companies account for 98% of the total amount raised in international markets by companies not in our sample. Consistent with our expectations, 811 of them are unlisted companies, which account for 89% of the amount issued in international markets not in our sample; listed firms not in our sample account for the remaining 11%.

3.4 Informational asymmetries

International investors require a premium on firms perceived as risky, due to informational asymetries. This spread results in a constrained access to international debt markets, in the sense that firms cannot borrow as much as desired unless they pay a premium to compensate investors.

We cannot observe if a firm suffers such constrained access to international debt markets, and there is no agreed procedure to assess this. Some papers use certain firm-level characteristic which expectedly signal that companies have problems to raise external finance: firm size (Almeida et al. [2004], Baum et al. [2011], Devereaux and Schiantarelli [1990], Hennessy and Whited [2007]), age (Lyandres [2007] and Devereaux and Schiantarelli [1990]), payout policy (Almeida et al. [2004], Fazzari et al. [1988],Baum et al. [2011], Farre-Mensa and Ljungqvist [2013] and Hennessy and Whited [2007]), credit rating (Lemmon and Roberts [2010]), or existence of a rating on a bond (Almeida et al. [2004], Farre-Mensa and Ljungqvist [2013], and Whited [1991]). This literature assumes that firms are more likely to have a constrained access if they are small, young, not paying dividends, have a poor credit rating, or do not have an outstanding bond with public rating.

Instead of using a single variable, we decide to use the Hadlock and Pierce [2010] index. Hadlock and Pierce [2010] related qualitative information on firms’ access to finance with observable charac­teristics. Firms’ status were categorized reading statements made by managers in SEC filings, on a random sample of US firms. Managers discussing liquidity problems or losses of sources of credit were revealing constraints in access to finance. These assessments were related to observable charac­teristics. Constraints depended negatively on size and age, and positively on the quadratic term of size: older and larger firms are bettter known, so have a less constrained access to debt markets (ie, a lower cost of external finance). Based on their results, Lamont et al. [2001] defined the following index:


Higher values reveal more costly access to debt markets. Following their approach, we measure size with total assets, deflated with US CPI. They cap size and age at percentile 95, which in their sample are respectively 4.5 US billion and thirty-seven years. In our sample these values are 4.9 US billion and sixty-eight years. Our firms are older, but in our main analyses we follow their approach and cap values at percentile 95. We cross-check that the results are not sensitive to alternative definitions.

 By construction firms with weak constraints are larger and older. They have higher leverage ratios, and lower cash to capital ratios (as in Hadlock and Pierce [2010]). Thus the index is not clearly related to standard measures of balance-sheet vulnerabilities.

The Hadlock-Pierce index is unevenly distributed across industries. Energy and utility companies have low values, since are often large (see Table A8); and technology companies have high values, reflecting they tend to be small. In robustness checks we make sure this does not distort our results.

This index has advantages over the two other major alternatives: the Kaplan-Zingales (Kaplan and Zingales [1997]) and the Whited-Wu index (Whited and Wu [2006]).

The main drawback of the Kaplan-Zingales index is that Kaplan and Zingales [1997] used balance­sheet information to categorize firms’ financial constraints, as a way of complementing qualitative information on firms’ statements. By construction, the subset of financially constrained firms defined by Kaplan and Zingales [1997] has low cash ratios, hight debt, low dividends. The Kaplan-Zingales index seems to measure of the need for external finance (Hennessy and Whited [2007]).

The drawback of the Whited and Wu [2006] is their assumption that that higher investment-cash sensitivities reveal higher financial constraints -since they investigated how investment-cash sensitiv­ities relate to observable characteristics. Research suggests that the assumption is strong (Kaplan and Zingales [2000],Almeida and Campello [2007],Cleary et al. [2007],Lyandres [2007]).

3.5 Global risk appetite

We use the excess bond premium (Gilchrist and Zakrajsek [2012]) to measure global risk appetite. We interpret it reflects the price of credit charged by investors in global corporate capital markets; it reveals the level of lending standards, independently of shifts in companies’ credit risks.

Most previous research uses the VIX (implied volatility of S&P 500 equity index options) to measure global risk appetite, due to its strong impact on cross-border capital flows (Forbes and Warnock [2012]). Other potential measures are the Merrill-Lynch Option Volatility Estimate (MOVE) Index, and the US dollar term premium. The MOVE index measures the implied volatility of the 1-month Treasury options, and can be considered as the counterpart of the VIX in bond markets, capturing the risk of holding Treasury securities. The US dollar term premium is interpreted as a measure of investors’ appetite in bond markets, since it reflects the compensation which investors receive for holding long-term US treasuries.

These three variables are imperfect measures of the price of credit, since also reflect how default risks fluctuate along the financial cycle. Consequently they are not useful to identify supply shocks; for instance, a negative relationship between the VIX and firms’ borrowing could reflect how higher default risks translate into lower demand for funds.

To address this identification problem Gilchrist and Zakrajsek [2012] decomposed firms’ credit spreads into two time-varying measures: a measure of firms’ credit risk; and the effective price of bearing exposure to risk required by US financial intermediaries. The latter measure is the excess bond premium, reflecting investor sentiment or risk appetite. It is positively correlated with the three mesaures described above, but the coefficient is below one.

The excess bond premium is computed using US corporate bond markets data. Given the impor­tance of US financial intermediaries we assume it measures risk appetite in global corporate bond markets. When the excess bond premium is low, credit spreads compress, the capital supply curve of constrained companies flattens.

4 Econometric framework

4.1 Model

Our empirical model to investigate the hypothesis is based on the model of determinants of change in leverage developed by Rajan and Zingales [1995], and subsequently adapted by Baker and Wurgler [2002], Farre-Mensa and Ljungqvist [2016] or Lemmon and Roberts [2010]. In this model we define the dependent variable yn as the total amount of international bond issuance of firm i, relative to total assets in the previous year.


In the vector of controls xa we include firm-level determinants of borrowing: tangible assets, size, profitability, and leverage, as in Baker and Wurgler [2002]. Tangible assets is defined as the sum of property, plant and equipment to lagged total assets. Size is the logarithm of total assets, CPI adjusted and measured at 2000 USD prices. Firm profitability is measured with the lagged return-on-assets. Book leverage is defined as book debt (sum of short and long-term debt) to lagged total assets, and it is introduced lagged. In the equation we include as well industry and country-level controls. At the country-level we include a country’s GDP per capita, financial depth and sovereign rating, as factors which can affect a company’s demand for finance. We treat unobserved heterogeneity at the industry level ai and the country level ac as fixed effects. Finally, we include global GDP growth.

We expect international borrowing by constrained firms to exhibit a stronger response to shifts in the excess bond premium. Thus we expect 71 to be negative, and higher in absolute terms for the subsample of constrained firms. In all models standard errors are clustered at the year level.

4.2 Identification issues

There are four potential identification problems. First that supply and demand shock may be con­founded. This happens for instance when shifts in global financial conditions also reflect changes in firms’ default risks. Since we are covering firms based in small open economies, we consider it is extremely unlikely that unexpected shifts in the firm-level demand for finance affect the price of risk in global markets. Besides we have addressed this potential problem using the excess bond premia (Gilchrist and Zakrajsek [2012]). Finally, to account for firms’ demand for finance we have included several observable firm-level determinants of firms’ demand for finance: tangible assets ratio, size, profitability, and lagged leverage (inversely related to net worth). Overall we consider this problem is not important.

The second issue is the measurement of constraints in access to international markets. The Hadlock- Pierce index appears to be the best way of gauging if firms face a high cost of external finance, but this measure has been critized on empirical grounds. Using a sample of US firms Farre-Mensa and Ljungqvist [2013] find that firms categorized as constrained using this variable behave as if they were unconstrained; they find similar drawbacks in the other measures examined in this paper (size, Kaplan-Zingales, and Whited-Wu index). Instead we find that firms classified as constrained behave as if they were constrained. Following Farre-Mensa and Ljungqvist [2016] heuristical assessment we consider that high values of the Hadlock-Pierce index do signal a constrained access to international markets. As a robustness check, we use as an alternative the number of analysts covering a company; we will assume that better analyst coverage reduces informational asymmetries, and thus signals that companies have an easier access to international markets.

Third, we have defined the dependent variable as the total amount of international bond issuance relative to total assets in the previous year, and not as the change in net leverage (as done by Baker and Wurgler [2002], Farre-Mensa and Ljungqvist [2013] or Lemmon and Roberts [2010]). Since international and domestic markets are partially segmented, as shown by Gozzi et al. [2010], these sources of finance are not perfect substitutes. It follows that shocks to the price of risk should affect firms’ international bond borrowing -and only through this on their aggregate borrowing.

Finally, in our first exercise we have defined subsamples of strongly/weakly constrained companies using arbitrary thresholds. This is standard practice in the capital structure literature. However this has drawbacks, since results can be sensitive to alternative thresholds. Moreover splitting the sample diminishes the power of the model. We tackle the issue in two different ways. One is to interact the financial constraint measure and the excess bond premium. Another is to split the sample between strongly and weakly constrained firms estimating the thresholds as suggested by Hansen [2000]. Once the threshold is identified we augment a linear model with an interaction between the excess bond premium and an indicator variable which takes value 1 if firms’ financial constraints are above a certain threshold.

5 Impact on international borrowing

5.1 Main results

Table 6 reports the results obtained using the Hadlock-Pierce index to gauge firms’ financial con­straints in international bond markets. The first column reports the results of estimating the model defined by 2 and 3 for the full sample. As expected the excess bond premium has a negative impact on firms’ borrowing (the coefficient 71 on BPt is negative), and the impact of the excess bond pre­mium is negative also in the three subsamples (columns 2, 3 and 4). Consistent with our expected the size of the impact depends on cross-sectional differences in firms’ degree of financial constraints: it is smallest for firms with weak constraints (column 2), and it is largest for companies with strong financial constraints in international markets (column 4). It is worth noting, however, that few firms with strongly constrained access to international bond markets are able to tap international markets. Thus the results of this subsample can be unstable. More generally, the results can be sensitive to alternative definitions of the subsamples.

Since the results of the subsample estimations are not reliable, we estimate a model with an in­teraction between the excess bond premium and the Hadlock-Pierce index, modifying Equation 3

The necessary condition for our hypothesis to hold is that 73 is negative: this indicates that the impact of the excess bond premium on firms’ international borrowing increases as financial con­straints become more binding. Since HadlockPierceit is continuous, the statistical significance of the impact is different from a Wald test on 73. The confidence interval depends on cov (71,73), so the statistical significance can change substantially with the value of financial constraints. The excess bond premium has a negative impact, and the coefficient of the interaction is also negative: firms’ international borrowing becomes more sensitive to the excess bond premium the more constrained companies are. As a way of summarizing this result, we plot the partial effect of the excess bond premium. The range leaves aside the top and bottom 1% of the distribution-. The solid blue line represents the partial effect, and the dashed lines 95% confidence intervals. We overlay a histogram on the distribution of the Hadlock-Pierce index. For companies with a very easy access to mar­kets (Hadlock-Pierce index is 2 standard deviations below the mean) the impact of the excess bond premium is not different from zero.

A negative sign would indicate that shifts of the excess bond premium particularly impact on borrowing by financially-constrained firms. The results are consistent with the hypothesis: 71 is negative, underscoring that firms’international borrowing is negatively impacted by increases in global risk aversion; and 74 is also negative, indicating that this impact is stronger for financially constrained firms. The results are economically important: for firms weak constraints one standard deviation of the excess bond premium translates into a 6.4% decrease in the ratio of international borrowing to total assets; this impact is multiplied by 13 for firms with constraints above the identified threshold.


These results indicate that firms with higher credit risk borrow more in international markets in response to strong risk appetite. Our measure of credit risk is not related to firms’ capital structure, so this pattern can be somehow unnoticed in financial ratio analyses.

5.2 Robustness checks

5.2.1 Influential sectors

We test whether our results are driven by influential sectors. Energy companies in particular need to borrow frequently in international markets to match the currency composition of assets and liabilities. This issue might also affect utility companies. For these firms, international funding is likely to be less responsive to global risk appetite. Since energy and utility firms are large, typically they are categorized as low risk borrowers (ie weakly constrained) using the Hadlock-Pierce index (see Table A8). This could result in a spurious relation beween low risk and lack of sensitivity to global risk appetite. However our results hold when we remove energy or utility firms; and also when we simultaneously remove both (in particular in this case, they are slightly stronger).

5.2.2 Financial constraints measures​​​​​​​

We estimate the model using other variables which can signal constrained access to global markets: firm-size, the Whited-Wu, and Kaplan-Zingales index. The results are similar using firm-size, and are weaker using the Whited-Wu and particularly the Kaplan-Zingales index (Tables C1 C2 and C3 in Appendix C).

We find that higher values of the Kaplan-Zingales index are associated to higher international bond borrowing, reflecting that these firms borrow more than their counterparts in international markets. Firms classified as constrained using the Kaplan-Zingales index seem to have high funding needs, and not necessarily a high cost of external finance, which is our interest (Hennessy and Whited [2007]). For this reason we conclude that the results are not inconsistent with the hypothesis.

5.2.3 Global risk appetite measures​​​​​​​

We use the VIX, the MOVE, and the US dollar term premium as our measure of global financial conditions. Although as discussed earlier these measures may confound demand and supply effects, it is worth checking the results obtained using them. In particular the VIX is still the most popular measure of risk appetite. We find that in all cases the results hold qualitatively, although the  quantitative importance is smaller in some instances. They are particularly weaker when we US dollar term premium. We think it is not a major problem, given the low unconditional impact of this variable on firms’ international bond borrowing (see Table B1).

5.3 Analysts coverage

We use the number of analysts covering each firm to assess informational asymmetries in interna­tional markets. Existing empirical evidence suggests that extensive coverage by analysts signals low asymmetries: good coverage reduces informational asymmetries; and analysts tend to extend cover­age to more transparent firms (Chang et al. [2016]). We assume that less analyst coverage reflect more uncertainty surrounding the firm. Thus less analyst coverage should be associated to a more costly access to international markets.

To examine the importance of analysts coverage, we split firms into two groups: firms with good (bad) coverage are those with a number of analysts above (below) the average.

Stronger risk appetite leads to higher international borrowing by firms with bad coverage, but does not alter borrowing by firms with good coverage. These results further validate our conclusion: firms perceived as riskier disproportionally borrow when the price of risk is low in international markets.

5.4 Domestic markets​​​​​​​

Our hypothesis is based on the premise that domestic and international corporate bond markets are segmented, in the sense that international investors do not acquire securities into domestic corporate bond markets.

We explore if this assumption is sensible re-estimating the main results using as dependent variable domestic (instead of international) bond issuance.


The results show that global risk appetite does not have a negative impact on international borrowing by the average firm (column 1). This finding validates our assumption.

However we do find a different reaction by firms with normal and strong constraints. The former react by increasing bond issuance in local markets (column 3), while the latter decrese them (column 4). We do not further explore the finding. However it suggests that global risk appetite could indirectly enhance domestic borrowing of firms with high risk. This would ocurr if firms with medium constraints decreased their domestic borrowing, in response to easier access to overseas funding.

5.5 Discussion​​​​​​​

Our findings point out that global risk appetite is transmitted into local economies through the offshore borrowing of smaller, younger companies. These firms are typically more unknown, and are charged a premium by international investors. This measure of credit risk is not related to collateral availability, since the tangible assets ratio has a very low correlation with the Hadlock-Pierce index. Thus risk-taking can ocurr without an apparent increase in issuers’ balance-sheet vulnerabilities.

We acknowledge that exchange rates appreciations vis-a-vis the currency of denomination of liabilities can ease funding costs in the presence of currency mismatches between assets and liabilities (Avdjiev et al. [2018a]). We cannot check the issue since lack of data on the currency composition of assets (see Kalemli-Ozcan et al. [2018] for a paper exploring this channel). However we think that our results are not biased, since the strength of the effective exchange rate has a very low correlation with the excess bond premium.

Finally it might seem strange that we have not measured firms’ effective credit risk with credit ratings -as in Greenwood and Hanson [2013]. However measuring firms’ effective credit risk is challenging in our sample. Credit ratings are rarely available (only in 14% of the firm-year observations). If global risk appetite leads to more borrowing by firms with high expected default risks merits further investigation, but we leave it for further research.

5.6 Limitations​​​​​​​

Time-invariant unobserved heterogeneity across firms could potentially bias our results, because we

cannot include firm fixed-effects in the Tobit model due to the incidental parameter problem. We have tried, as a robustness checks, to estimate the model with the correlated random effect estimator, which includes the average of firm-specific variables to account for time-invariant unobserved heterogeneity. However, the maximum-likelihood function did not converge. This is a limitation of our analysis.

6 Onshore transmission: Impact on capital structure

Recent research (Baker [2009]) underscores that credit market developments strongly impact on firms’ capital structure. Accordingly we examine how firms’ balance-sheets change when risk appetite fosters borrowing in international markets. These changes would reflect the transmission of global risk appetite into domestic activity.

Presumably the effects will be stronger for firms with a more constrained access to international debt markets. To shed light on this issue, we investigate companies’ use of proceeds over a three-year horizon, adapting a standard model (see Kim and Weisbach [2008] and Bruno and Shin [2017]) :

The dependent variable yn is, alternatively, the change in net book leverage, the change in cash holdings, and (cumulated) capital expenditures between t and t-3. Assetsu is the logarithm of assets measured at t-3. Proceedsц measures international bond proceeds over the last three years (between t and t-3). Revenuesa is defined as the cash obtained by the firm in regular operations (ie cash from operations) in the same period. HadlockPiercen is the average Hadlock-Pierce index over the last three years. BPt is our measure of global risk appetite: the excess bond premium over the last three years (it is standardized, as in the previous section). We restrict the estimation to observations with positive values of international bond proceeds at t-3.

To examine the impact of global risk appetite on firms’ use of proceeds, we interact Proceedsц with BPt. A negative coefficient in the interaction term въ would indicate that the impact of international borrowing on firms’ capital structure is stronger when risk appetite is high. It is worth mentioning that in this model we are not allowing the impact of global risk appetite to depend on firms’ credit risks. In all models we include country and industry fixed effects.

Columns 4, 5 and 6 in Table 10 reports the results (columns 1 to 3 show results of a model without the interaction terms). The impact of international bond proceeds on net book leverage (ві) is, as expected, positive. We find that the impact is higher if bonds are raised when the market-wide price of risk is cheap (ie the coefficient въ of the interaction term has a negative and statistically signif­icant sign). Expansions in risk appetite positively impact on firms’ leverage through international borrowing. This impact is reflected in higher firms’ capital expenditures and cash holdings.

Next we examine if this impact is stronger on firms with higher credit risk (ie more costly access to international markets); we include two additional terms on the equation:

We find the expected result, since P7 is negative and highly significant: firms with higher credit risk issuing bonds during periods of easy global financial conditions experience a sharper increase in leverage than the rest of firms; this translates into higher cash holdings, and (to a lower extent) capital expenditures. The marginal effects are shown in Table 11.

We conclude that balance-sheet vulnerabilities build up when funding constraints in global bond markets are relaxed due to stronger risk appetite.

7 Conclusions

We use cross-country, micro-level data to better understand the transmission mechanism of global risk appetite across borders.

We document that firms with higher credit risks help to transmit onshore the financial conditions prevailing in international bond markets. Our findings show that small and young firms tend to tap global bond markets when the price of risk is low; in contrast borrowing by larger and older firms is less sensitive to the price of risk. This distinct sensitivity to the price of risk is a consequence of differences in credit spreads. Due to severe informational asymmetries, spreads are generally higher for small and young firms.

These results are consistent with previous research that finds that cycles of international borrowing are strongly related to global risk appetite. Our contribution consists in showing that this response varies across firms; in particular we stress the importance of informational asymmetries between firms and international investors.

Our findings show that global risk appetite can trigger an influx of risky firms in international bond markets. This pattern can remain unnoticed in financial ratio analyses, since the measures of firms’ credit quality which we exploit are independent of firms’capital structure. However, we find that international bond issuances lead, over a three-year horizon, to some deterioration in the financial ratios of risky borrowers.

These results unveil a micro-level link between risk appetite in international corporate bond markets, and the build-up of domestic risks. Through their fund raising in international bond markets, firms with higher credit risk link their local economy with the global financial cycle.



V. Acharya and S. Vij. Foreign currency borrowing of corporations as carry trades: Evidence from india. mimeo, 2017.
H. Almeida and M. Campello. Financial constraints, asset tangibility and corporate investment. The Review of Financial Studies, 20(5):1429-1460, 2007.
H. Almeida, M. Campello, and M. Weisbach. The cash flow sensitivity of cash. The Journal of Finance, . LIX(4), August 2004 2004.
S. Avdjiev, L. Gambacorta, L. S. Goldberg, and S. Schiaffi. The shifting drivers of global liquidity. BIS Working Papers No. 644, 2017.
S. Avdjiev, V. Bruno, C. Koch, and H. Shin. The dollar exchange rate as a global risk factor: evidence from investment. BIS Working Papers No 695, 2018a.
S. Avdjiev, M. Everett, P. R. Lane, and H. S. Shin. Tracking the international footprint of global firms. BIS Quarterly Review, March 2018, 2018b.
M. Baker. Capital market-driven corporate finance. Annual Review of Financial Economics, 1: 181-205, 2009.
M. Baker and J. Wurgler. Market timing and capital structure. The Journal of Finance, LVII(1), Feb 2002 2002.
Y. Baskaya, J. di Giovanni, S. Kalemli-Ozcan, J.L. Peydro, and M. Ulu. Capital flows and the international credit channel. Journal of International Economics, 108 S1 (May 2017):S15-S22, 2017a.
Y. Baskaya, J. di Giovanni, S. S. Kalemli-Ozcan, and M.Ulu. International spillovers and local credit cycles. Barcelona GSE Working Paper Series No. 953, 2017b.
C. Baum, D. Schafer, and O. Talavera. The impact of the financial system’s structure on firms’ financial constraints. Journal of International Money and Finance, 30:678-691, 2011.
T. Beck, A. Demirguc-Kunt, L. Laeven, and V. Maksimovic. The determinants of financing obstacles. Journal of International Money and Finance, 25:932-952, 2006.
F.    Broner, T. Didier, A. Erce, and S. Schmukler. Gross capital flows: Dynamics and crises. Journal of Monetary Economics, 60(1):113-133, 2013.
V. Bruno and H. Shin. Cross-border banking and global liquidity. The Review of Economic Studies, 82:535-564, 2015.
V. Bruno and H. Shin. Global dollar credit and carry trades: a firm-level analysis. The Review of Financial Studies, 3(3):703-749, 2017.
J. Caballero, U. Panizza, and A. Powell. The second wave of global liquidity: why are firms acting like financial intermediaries? CEPR Working Paper, (10926), 2016.
C.    W. Calomiris, M. Larrain, and S. L. Schmukler. Capital inflows, equity issuance activity, and corporate investment. NBER Working Paper Series No. 24433, 2018.
M. Carabin, A. de la Garza, and O. Moreno. Global liquidity and corporate financing in mexico. mimeo, 2015.
E. Cerruti, S. Claessens, and A. K. Rose. How important is the global financial cycle? evidence from capital flows. BIS Working Papers No. 661, 2017.
X. Chang, S. Dasgupta, and G. Hilary. Analyst coverage and financing decisions. The Journal of Finance, LXI(NO. 6), December 2016 2016.
S. Cleary, P. Povel, and M. Raith. The u-shaped investment curve: theory and evidence. Journal of Financial and Quantitative Analysis, 42:1-39, 2007.
O. De Jonghe, H. Dewachter, K. Mulier, S. Ongena, and G. Schepens. Some borrowers are more equal than others: Bank funding shocks and credit reallocation. mimeo, 2016.
M. Devereaux and F. Schiantarelli. Asymmetric Information, Corporate Finance, and Investment, chapter Investment, Financial Factors, and Cash Flow: Evidence from U.K. Panel Data. University of Chicago Press, Chicago IL, 1990.
D.     Diamond. Financial intermediation and delegated monitoring. Review of Economic Studies, LI: 393-414, 1984.
Bat el Berger. Debt weighted exchange rate indices. Box, BIS Quarterly Review, December 2016, 2016.
J. Farre-Mensa and A. Ljungqvist. Do mesures of financial constraints measure financial constraints. NBER Working Papers, (19551), 2013.
J. Farre-Mensa and A. Ljungqvist. Do measures of financial constraints measure financial constraints? The Review of Financial Studies, 29(2), 2016.
S. Fazzari, R. Hubbard, and B. Petersen. Financing constraints and corporate investment. Brooking Papers of Economic Activity, 19:141-206, 1988.
E.     Feyen, K. Gosh, K. Kibuuka, and S.Farazi. Global liquidity and external bond issuance in emerging economies and developing economies. World Bank Research Working Paper 7363, 2015.
K. Forbes and F. Warnock. Capital flow waves: Surges, stops, flights and retrenchment. Journal of International Economics, 88(2):235-251, 2012.
M. Fraztchter. Capital flows, push versus pull factors and the global financial crisis. Journal of International Money and Finance, 88(2):314-356, November 2012 2012.
S. Gilchrist and E. Zakrajsek. Credit spreads and business cycle fluctuations. The American Economic Review, 102(4):1692-1720, 2012.
J. P. Gozzi, R. Levine, and S. Schmukler. Patters of international capital raisings. Journal of International Economics, 80:45-57, 2010.
J.C. Gozzi, R. Levine, M. Martinez Peria, and S. Schmukler. How firms use corporate bond markets under financial globalization. Journal of Banking & Finance, 58:532-551, 2015.
R. Greenwood and S. Hanson. Issuer quality and corporate bond returns. The Review of Financial Studies, 26(6), 2013.
B.     Gruic and P. Wooldridge. Enhancements to the bis debt securities statistics. BIS Quarterly Review, December 2012 2012.
C.     Hadlock and J. Pierce. New evidence on measuring financial constraints: Moving beyond the kz index. The Review of Financial Studies, 23:1909-1940, 2010.
B. Hansen. Threshold effects in non-dynamic panels: Estimation, testing and inference. Journal of Econometrics, 93:345-368, 1999.
B.     Hansen. Sample splitting and threshold estimation. Econometrica, 68(3):575-603, May 2000 2000.
C.     Hennessy and T. Whited. How costly is external financing? evidence from a structural estimation. The Journal of Finance, LXII(4), August 2007 2007.
S. Kalemli-Ozcan, X. Liu, and I. Shim. Exchange rate appreciations and corporate risk taking. BIS Working Papers No 710, 2018.
S. Kaplan and L. Zingales. Do investment-cash flow sensitivities provide useful measures of financing constraints. The Quarterly Journal of Economics, 112(1):16-215, 1997.
S. Kaplan and L. Zingales. Investment-cash flow sensitivities are not valid measures of financing constraints. The Quarterly Journal of Economics, 115(2):707-712, 2000.
W. Kim and M. Weisbach. Motivations for public equity offerigns: An international perspective. Journal of Financial Economic, 87:281-307, 2008.
O. Lamont, C. Polk, and J. Saa-Requejo. Financial constraints and stock returns. The Review of Financial Studies, 14(2):529-554, 2001.
M. Leary. Bank loan supply, lender choice, and corporate capital structure. The Journal of Finance Vol. LXIV, NO. 3 June 2009, 2009.
M. Lemmon and M. Roberts. The response of corporate financing and investment to changes in the supply of credit. Journal of Financial and Quantitative Analysis, 45(3):555-587, 2010.
G.    Liao. Credit migration and covered interest rate parity. Project on Behavioral Finance and Financial Stability Working Paper Series 2016-07, 2016.
M. Lo-Duca, G. Nicoletti, and A. Vidal-Martinez. Global corporate bond issuance: What role for us quantitative easing? Journal of International Money and Finance, 60:114-150, 2016.
E. Lyandres. Costly external financing, investment timing, and investment-cash flow sensitivities. Journal of Corporate Finance, 13:959-980, 2007.
Y. Ma. Non-financial firms as cross-market arbitrageurs. mimeo, 2016.
M. Maggiori, B. Neiman, and J. Schreger. International currencies and capital allocation. Columbia Business School Research Paper No. 17-96, 2018.
R. McCauley, P. McGuire, and V. Suskho. Global dollar credit: links to us monetary policy and leverage. Economic Policy, 30(82):187-229, April 2015 2015.
G.    M. Milesi-Ferretti and C. Tille. The great retrenchment: international capital flows during the global financial crisis. Economic Policy, pages 289-346, 2011.
S. Miranda-Agrippino and Helen Rey. World asset markets and global liquidity. mimeo, 2013.
S. Ongena, Jose-Luis Peydro, and Neeltje van Horen. Shocks abroad, pain at home? bank-firm-level evidence on the international transmission of financial shocks. IMF Economic Review November 2015, Volume 63, Issue 4, pp 698-750, 2015.
R. Rajan and Luigi Zingales. What do we know about capital structure some evidence from inter¬national data. The Journal of Finance, 5(5):1421-1460, Dec 1995 1995.
E.    Rodriguez-Bastos, H. Kamil, and B. Sutton. Corporate financing trends and balance-sheet risks in latin america: Taking stock of the bond(d)anza. IMF Working Paper 15/10, 2015.
H.     Shin. Exchange rates and the transmission of global liquidity. Speech at the Bank of Korea-IMF Conference “Leverage in Asia: Lessons from the Past, What’s New Now?, and Where to Watch Out For?'', Seoul, 11 December 2015., 2015.
T. Whited. Investment and financial asset accumulation. Journal of Financial Intermediation, 1: 307-334, 1991.
T. Whited and G. Wu. The financial constraints and risks. Review of Financial Studies, 19:531-559, 2006.