This paper aims to investigate how firms from emerging economies choose among different international bond markets: global, US144A and Eurobond markets. The authors explore if the ranking in regulatory stringency –global bonds have the most stringent regulations and Eurobonds have the most lenient regulations – leads to a segmentation of borrowers.
The authors use a novel data set from emerging economy firms, treating them as consolidated entities. The authors also obtain descriptive evidence and perform univariate non-parametric analyses, conditional and multinomial logit analyses to study firms’ marginal debt choice decisions.
The authors show that firms with poorer credit quality, less ability to absorb flotation costs and more informational asymmetries issue debt in US144A and Eurobond markets. On the contrary, firms issuing global bonds – subject to full Securities and Exchange Commission requirements – are financially sounder and larger. This exercise also shows that following the global crisis, firms from emerging economies are more likely to tap less regulated debt markets.
This is, to the authors’ knowledge, the first study that examines if the ranking in stringency of regulation – global bonds have the most stringent regulations and Eurobonds have the most lenient regulations – is consistent with an ordinal choice by firms. The authors also explore if this ranking is monotonic in all determinants or there are firm-specific features which make firms unlikely to borrow in a given market. Finally, the authors analyze if there are any changes in the debt-choice behavior of firms after the global financial crisis.
Fuertes, A. and Serena, J. (2018), "How firms borrow in international bond markets", Journal of Financial Regulation and Compliance, Vol. 26 No. 1, pp. 135-169. https://doi.org/10.1108/JFRC-11-2016-0100Download as .RIS
Emerald Publishing Limited
Copyright © 2018, Emerald Publishing Limited
International bond markets are an increasing source for corporate debt financing in emerging economies, especially after the global financial crisis. The easy access to debt markets over the past years is attributed to global financial conditions (Lo Duca et al., 2014; Ayala et al., 2015; Feyen et al., 2015). Among the lenders, non-bank institutions have stepped into the provision of credit in a context of low returns on traditional assets. Overall, the process is deemed to pose significant risks for international financial stability: borrowers could be raising too much debt and lenders could be underestimating the risk. However, there is considerable uncertainty about the actual credit risks and the characteristics of firms from emerging economies raising funds in international markets (Avdjiev et al., 2014). These markets have different regulatory regimes, resulting in a wide variety of funding options. The most important alternatives are global bonds, bonds issued under the rule US144A and Eurobonds.
These international bond markets impose different requirements to the disclosure of financial information. Eurobonds are not subject to any local regulation, and issuers face the lightest regulatory requirements. The Eurobond market is largely a wholesale market with bonds held by large institutional investors with “buy and hold” strategies. Because Eurobonds do not comply with US securities laws, they cannot be sold in the USA. On the contrary, global bonds are strictly regulated. These types of bonds are fully fungible securities issued simultaneously in two markets. They are typically placed in the Eurobond and at least one domestic bond market, where issuers face demanding disclosure requirements such as those imposed by the Securities and Exchange Commission (SEC) in the case of the USA. Finally, firms issuing bonds in the US market under the Rule 144A face less regulatory conditions than those issuing traditional bonds that need to be registered with the SEC. Rule 144A was introduced to ease the access to US capital markets by non-US firms, as it exempts issuers from complying with the standard regulation. In counterpart, bonds issued under the Rule 144A can be placed only among institutional investors.
The costs and benefits of borrowing in each market differ across firms, for instance, an issuance of global bonds. It implies complying with a strict regulation that widens the investor base and increases the liquidity of the securities. Though fulfilling the requirements imposes costs on issuers known as flotation costs, these costs depend on aspects such as firm size or the quality of their financial statements. The benefits depend on the degree of informational asymmetries between the firm and the borrowers or the risk of financial distress. Typically, only firms with low informational asymmetries will find it advantageous for issuing global bonds. The other firms will obtain a better pricing if the security is acquired by a narrow set of institutional investors. These investors better assess the risk of firms with uncertain credit quality, and they have better technology to maximize the recovery value in case of distress. Because of this trade-off, the type of firms attracted by each market is likely to differ.
Firms’ choice of international bond markets is similar to the decision between issuing private debt securities – placed to a narrow set of investors and subject to trade restrictions – or placing publicly bonds – which reach a wide and atomized base of investors and can be sold without restrictions in secondary markets. Issuances of global bonds are akin to public placements of securities. Issuing in the Eurobond market is more similar to borrowing privately. Placing bonds under the US Rule 144A is somewhere in between.
The importance of the trade-off in the choice between publicly and privately placed securities has been previously explored by numerous studies. Most of them have focused on firms from advanced economies and the US markets. The empirical results broadly confirm most of the hypotheses coming from the debt choice theory: for example, only large firms with higher credit quality issue public debt (Blackwell and Kidwell, 1988; Denis and Mihov, 2003; Chaplinsky and Ramchand, 2004; Krishnaswami et al., 1999). A few papers have also stated that because of differences in regulation, some international bonds are akin to publicly placed securities and others to private placements. Esho et al. (2001) posit that Eurobonds are more akin to privately placed securities than foreign bonds and thus constitute an alternative to syndicated loans. Gao (2011) studied the impact of Sarbanes–Oxley Act on non-US firms’ choice between standard Yankee bonds, bonds issued in the USA under Rule 144A and Eurobonds.
We extend the existing literature of debt choice by analyzing firms’ borrowing decision among the three key options available in international bond markets: global bonds, bonds issued under the US144A rule and Eurobonds. We use univariate analyses to compare firms borrowing in each market. Besides, we investigate firms’ incremental debt choice among existing international debt securities markets by using multivariate models: multinominal logit, ordered logit and generalized-ordered logit. This is, to our knowledge, the first study that examines if the ranking in stringency of regulation – global bonds have the most stringent regulations and Eurobonds have the most lenient regulations – is consistent with an ordinal choice by firms. We also explore if this ranking is monotonic in all determinants or there are firm-specific features which make firms unlikely to borrow in a given market. Finally, we also analyze if there are any changes in the debt-choice behavior of firms after the global financial crisis.
We research this question by using a unique database comprising firm-level and deal-specific information for all emerging economy firms active in international markets during the period of 2000-2014. In contrast to previous literature, we analyze the choice of debt type treating firms as consolidated entities, in line with the newest standards in international finance (Avdjiev et al., 2015). The database is built around the firms guarantying the debt securities, which need not be the issuer companies, and contains firms’ financial accounts, corporate structure and non-financial information, as well as the type of debt securities issued. Overall, there is information for 1,584 firms headquartered in 36 jurisdictions, which issue a total of 3,944 securities, for a total amount of US$1.2tn.
We have four major findings. First, our univariate non-parametric analyses suggest that firms are segmented by regulation: firms borrowing in Eurobond markets – the least regulated – have the lowest credit quality, lowest ability to absorb high flotation costs and exhibit the highest informational asymmetries. Firms issuing global bonds – fully subject to regulation – are at the other end of the spectrum. Firms issuing bonds under Rule 144A are in between them. Second, the results of multinomial logit models confirm a causal relationship: being less able to absorb high flotation costs and suffering informational asymmetries make firms more likely to issue debt in less regulated markets. Third, by using ordinal and generalized ordered logit models, we find a ranking among markets – Global, US144A and Eurobond – which is non-monotonic in two dimensions. Filing financial statements with local accounting standards make firms extremely unlikely to borrow in global bond markets. Furthermore, Eurobond issuers are set apart from other borrowers as they are very unlikely to be cross-listed in the USA with American depositary receipts (ADRs). Fourth, we also find that after the global financial crisis, emerging market firms’ propensity to issue bonds in less regulated markets – US144A and Eurobonds – has increased.
The rest of the paper is structured as follows. Section 2 discusses theoretically how firms choose among international debt securities markets and presents the variables used to empirically investigate their choice. In Section 3, we describe our database and present the univariate analysis. In Section 4, we present the multivariate analysis of firms’ incremental debt choice. Finally, in Section 5, we summarize the main conclusions.
2.1 Related literature
Our research question is related to previous literature analyzing firms’ choice between publicly and privately placed debt. Blackwell and Kidwell (1988), using data on US public utility companies, showed that small and risky firms almost always set debt privately because flotation and agency costs make public debt issues too expensive. They found that large and less risky firms sell larger issues and have access to both public and private debt markets. Esho et al. (2001) obtained similar results by using data on bank loans (a type of private debt) and corporate bonds (foreign bonds and Eurobonds) from large Asian companies. They also pointed out the differences between Eurobond and foreign bonds, suggesting that Eurobonds have some characteristics that indicate a monitoring role of investors. Other studies have found that firms with poor credit quality rely mostly on private debt (Denis and Mihov, 2003; Chaplinsky and Ramchand, 2004). Denis and Mihov (2003) used data on US firms borrowing from bank loans, public bonds and private bonds. They mentioned that Rule 144A bonds appear to combine some features of both bank loans and public debt. Chaplinsky and Ramchand (2004) analyzed data on debt issued by non-US firms in the public market and the US144A market.
Comparing our results with previous empirical findings is tricky because of several reasons: first, our paper focuses on the choice among all available international bond markets, while most of previous literature delved into the choice between private debt (bank loans) and public debt (corporate bonds) without exploiting the differences in the latter.
Similar to us, Gao (2011) and Resnick (2012) used data sets that include several types of international bonds, although their research question is different and they do not analyze the determinants of the decision choice among all different alternatives. Gao (2011) used data on foreign firms issuing debt in the Yankee, Eurodollar and US144A bond markets, with most of the firms being domiciled in the UK, Japan and Canada. Resnick (2012) tested different hypothesis by comparing investor yields among several bond markets: US domestic, Eurodollar, US144A, Yankee and global markets. In our study, by examining emerging market firms and covering the period of 2000-2014, we are able to exploit a rich heterogeneity at the firm level.
2.2 Regulation of international bond markets and public information: global, US144A and Eurobond markets
There are different international bond markets. Bonds are considered issued in an international market when the issuer is not domiciled in the jurisdiction (BIS-ECB-IMF, 2015). Accordingly, international bonds are those in which the issuer is subject to a non-local regulation. The aim of regulation is, quoting IOSCO, to ensure “full disclosure of information material to investors”; this is the mechanism to “protect investors, maintain fair, efficient and transparent markets, and seek to address systemic risks” (IOSCO, 2010). Overall, this implies that there are noticeable differences in the volume, quality and timeliness of public information depending on the bond market the firm chooses.
Global bond markets are tightly regulated, and issuers are required to disclose publicly financial information, fulfilling detailed and often burdensome regulatory requirements. Investors cannot influence the design of the security, which is offered on a take-it-or-leave-it basis. Accordingly, debt securities are fairly standardized and more liquid than otherwise. The risks of investment in these firms are often assessed through an external rating agency: investors do not screen privately the quality of the firm and do not monitor ex post managers’ decisions. There are many bondholders who do not have expertise in liquidating firms; their losses in an event of financial distress are expectedly larger. The US Rule 144A and the Eurobond market are institutional/wholesale and private bond markets. They are subject to lighter regulation. Borrowers disclose private information to a narrower set of informed potential lenders, which include institutional investors such as insurance companies or pension funds. These lenders have expertise in dealing with informational asymmetries surrounding the firm value and agency problems. They might require firms to include covenants or options to mitigate adverse selection or moral hazard problems. Debt securities are less standardized and henceforth less liquid. Also, the role of external credit agencies is far less important. In an event of distress, liquidation is more efficient as lenders have expertise and have designed provisions to cope with it.
The main features of these markets are described in Table I. Regulation requires firms issuing debt in public markets to provide substantial public information. The disclosure of information reduces the asymmetry of information and gives access to a broad pool of non-sophisticated investors. US debt securities regulation constitutes a relevant example. The requirements are ruled by the Securities Act of 1933, which requires SEC registration. The process is lengthy and particularly burdensome for foreign firms, as these firms have to file their financial accounts by using the US generally accepted accounting principles (GAAP).
Global bonds are debt securities publicly placed by non-resident firms in at least one local market. A Mexican firm issuing a debt security simultaneously in the US and Eurobond debt markets is issuing a global bond. Global bonds are often very large and usually place at least one tranche in the USA; therefore, they are fully subject to the SEC requirements. Similar to Yankee bonds or US domestic public debt placements, there is public information available.
Overall, investors in global bonds dispose public and high-quality information; they do not need any specific technology to monitor borrowers’ credit quality, and retail investors are able to buy and trade these securities. Thus, the investors’ base is much atomized, and securities are relatively standardized. Global bonds are liquid, and there is a secondary market for them.
Debt securities issued under the US Rule 144A constitute another market. It is a specific and very relevant one. The 144A private placement market developed after the SEC was introduced the Rule 144A in 1990. Rule 144A created a second-tier market for both US and non-US firms. Firms issuing under the US Rule 144A are subject to much softer requirements on the quality, volume and frequency of financial accounts disclosure. Rule 144A allows trading these debt securities among “qualified institutional buyers”. Non-US firms became very active in this market, as the SEC requirements to issue in the US public markets were particularly burdensome for them. Rule 144A debt securities have low liquidity, and a substantial fraction of international investors is banned from investing in them. Thus, it has features of non-bank private debt (Chaplinsky and Ramchand, 2004; Fenn, 2000; Esho et al., 2001; Arena, 2011; Gomes and Phillips, 2012).
Finally, the last international debt market available is the Eurobond market. Eurobonds are debt securities in which investors need their own technology to monitor borrowers’ risks. In terms of existing regulation, Eurobonds are totally different from each of the other markets described: debt securities are not subject to any local jurisdiction (Esho et al., 2001; Miller and Puthenpurackal, 2002). Indeed, the Eurobond market was developed to bypass the existing local jurisdictions. SEC rules imply that Eurobonds cannot be sold in US capital markets: bonds are bearer and are not registered in the SEC, as global bonds are; they are not subject to the Rule 144A regulatory requirements either. Eurobonds markets are necessarily less liquid, and the pool of potential investors is limited to sophisticated institutional investors.
2.3 Firms’ choice between issuing public and institutional/wholesale offerings
There is a theory of placement structure of non-bank debt (Blackwell and Kidwell, 1988; Krishnaswami et al., 1999; Kwan and Carleton, 2010). It stresses that public and non-bank private debt exhibit significant differences in three dimensions: flotation costs, effectiveness in addressing informational asymmetries and the efficiency of liquidation and renegotiation in an event of distress. Firms will issue debt securities in the market in which their funding costs are smaller. We will study how these factors affect the choice of debt among the different markets: global bonds, Rule 144A bonds and Eurobonds.
First, flotation costs are larger in the public debt securities markets (as discussed, for instance, Blackwell and Kidwell, 1988). These costs include all the expenses that borrowers need to pay to issue the debt security and include underwriter compensation, legal fees, accountants’ fees and the costs of fulfilling the regulatory requirements. Compliance is costly, in particular, for emerging economies firms, which need to provide a substantial amount of information following non-local accounting standards [international financial reporting standards (IFRS) or US GAAP]. Though the bulk of flotation costs are fixed, they do not increase the size of the amount issued. Hence, public debt issuances become less and less expensive as a firm becomes larger, the amount it wants to raise becomes larger or if the issuer is already complying with the required disclosure of information.
Second, private debt securities markets are more effective to address informational asymmetries between firms and investors (Krishnaswami et al., 1999 and references therein). Problems of adverse selection plague debt-financing (Myers and Majluf, 1984). They occur when investors have incomplete information on firms’ actual value or disagree with the managers about its value. Agency problems are also frequent: managers’ commitment to maximize the firms’ value can change after lending occurs (Green, 1984; Green and Talmor, 1986; Myers, 1977). In the process of issuing security privately, lenders can require value-decreasing concessions to mitigate agency costs: they include negative covenants to align the incentives of equity and bond holders or additional guarantees to mitigate some specific risks. Through the process, the risks of adverse selection can also be mitigated: informed lenders can require firms to include dissipative signals, such as short maturities and collateral posting, establishing a sinking fund or embedding convertible rights.
Third, firms’ distress is less costly for investors in the private debt securities markets. Investors in privately originated debt are fewer; they are more sophisticated and probably have included covenants to ease the management of the process. In contrast, investors in public debt markets face important coordination problems and difficulties to maximize the liquidation value. Overall, the liquidation value of the firm is higher for investors in private debt markets. These arguments were first raised for bank private debt (Berlin and Loyes, 1988; Chemmanur and Fulghieri, 1994); they apply to other non-bank private lenders as well (Denis and Mihov, 2003). Overall, firms with higher credit risk pay a large spread to issue debt in public markets; similarly, a larger spread will be charged for firms with less fungible assets, such as goodwill or patents. Thus, these firms are more likely to borrow privately.
2.4 Determinants of debt choice
To analyze how the factors explained in the previous section affect the choice among the different international bond markets, we use two types of firm-level variables: firm-specific information and financial contract characteristics. Table II summarizes the variables we use.
2.4.1 Flotation costs.
To measure flotation costs, we include the amount raised by the firm: the larger the amount, the lower the fixed flotation cost relative to the total proceeds. As an alternative, we include the firm size, measured by its total assets (Denis and Mihov, 2003; Kwan and Carleton, 2010; Blackwell and Kidwell, 1988; Krishnaswami et al., 1999). Incidentally, flotation costs can be lower for firms which are already complying with the disclosure requirements imposed by securities regulation. For example, firms whose equity is cross-listed in US exchange markets are already complying with SEC requirements. Hence, we use as a proxy for the ability to absorb flotation costs the existence of exchange-listed ADRs. Finally, we use firms’ reporting GAAP. Firms reporting the financial statements with their local GAAP cannot issue debt in regulated international capital markets unless they provide supplementary information using IFRS or US GAAP. This creates an additional cost of issuing a debt security in the global bond market.
2.4.2 Financial contracting costs due to informational asymmetries.
To measure informational asymmetries, we include the following variables. First, we gauge the actual value of firms’ assets including the ratio of fixed assets to total assets (Denis and Mihov, 2003). Holding everything else equal, lenders face less uncertainty investing in firms with a higher proportion of fixed assets. Hence, firms with a higher proportion of fixed assets to total assets are less likely to issue debt in the institutional/wholesale markets (Eurobond and US144A).
Second, to measure firms’ willingness to improve the quality of public information, we use the existence of a credit rating on a debt-security.We expect that firms issuing unrated debt are more likely to issue in the institutional/wholesale markets. We also exploit information on bonds’ convertible rights. These are embedded options that grant holders the right to convert the debt security into equity. Firms attach convertible rights to debt securities to alleviate contracting costs (Krishnaswami and Yaman, 2008; Brennan and Swartz, 1988; Lewis et al., 1998). Firms suffering more from informational asymmetries are more likely to issue bonds with convertible rights, and we use it as a proxy of informational asymmetries. Finally, firms can choose issuing debt securities at short-term maturities in case of uncertainty about their future investment decisions. Thus, we include the term (i.e. bond maturity) as a proxy of informational asymmetries – in this case, moral hazard.
2.4.3 Liquidation and renegotiation costs.
We use the Altman score as a measure of a firm probability of distress. The Altman score is a synthetic measure of a firm ability to repay debt obligations. Following Denis and Mihov (2003), we define a binary variable taking the value 1 if the Altman score is lower than 1.21; this defines firms with a high probability of financial distress. Second, we also use the existence of a sinking fund attached to the debt security as a proxy for the likelihood of financial distress. A sinking fund is a fund set up to pay back the bond. Firms set up sinking funds to minimize the cost of funding when the risk of financial distress is perceived to be large. Low-quality issues involve sinking funds; high-quality issues rarely do so (Allen et al., 2013). We define a binary variable taking the value 1 if the bond issued has a sinking fund. Besides, we use a number of financial ratios: interest coverage ratio, return on assets (ROA), leverage and current ratio. Finally, we investigate a firm liquidation value by using the fixed assets to total assets ratio. In case of distress, fixed assets lose less value during the liquidation process as they are more fungible than intangible assets (Esho et al., 2001). Hence, we expect firms with a lower proportion of fixed assets to issue debt in private debt markets.
2.4.4 Global financial conditions.
We measure global financial conditions using the average of the MOVE index for 20 days before the issuance. The impact of this variable is uncertain. On the one hand, a tightening in global financial conditions can alter firms’ choices among the different markets: underwriters’ risks are deemed to increase disproportionally in public offerings, thus making more convenient issuing in institutional/wholesale markets (Blackwell and Kidwell, 1988), though existing evidence is mixed so-far (Kwan and Carleton, 2010). On the other hand, issuances in institutional/wholesale markets take less time, and firms can choose better the market timing; this argument suggests that it is less likely to see firms issuing in US144A or Eurobond during periods of high volatility, as they can more easily avoid placing bonds in such circumstances.
3. Data set
3.1 A micro-level database for macrofinancial analysis
We gathered all the debt securities issued in international markets during the period of 2000-2014 and guaranteed by emerging economy firms. We cover 36 countries of four emerging economies regions: emerging Asia, Latin America, emerging Europe and Africa and Middle East. We have obtained the database by using Bloomberg. To carry out the analysis, we construct a firm-year database by using the deal-level information.
The database has three defining features. First, it is built around the firms guaranteeing the debt securities instead of the issuer entities; this allows mimicking properly the risk analysis carried out by international investors when deciding to invest in a given debt security. Second, the debt securities information contains bonds from the global market, US144A market and Eurobond market. Finally, the firm-level information is comprehensive and contains the debt securities guaranteed by unlisted firms, firms listed in local exchanges and firms cross-listed also in US exchanges.
Overall, these three features make our database comprehensive and entirely consistent: it contains 3,944 debt securities issued by 1,584 firms in the period of 2000-2014, which make up a total amount of US$1.2tn.
3.2 A criterion: analyzing firms that guarantee debt securities
The organizational structures of emerging economies firms have become very complex. Firms have affiliates incorporated all over the world. These affiliates have different degrees of financial autonomy relative to their parent institutions: some are fully supported, and others are fully independent; some others receive explicit guarantees for specific financial operations.
Our purpose is to mimic the risk analysis carried out by international investors; this is key to understand firms’ choices of market of issuance as they depend on the yield that international investors require. We assume investors price the risk of investing in a debt security analyzing the firm guaranteeing it: legally, it is the entity liable in case of distress. This criterion is superior to the other two alternatives: the analysis of the issuer firm or the parent company.
Accordingly, we obtain all debt securities guaranteed by emerging economy firms by following a previous contribution (Fuertes and Serena, 2014). We are the first to use this criterion, although the importance of assigning deals to the firm guaranteeing the debt security had already been suggested (Esho et al., 2001). Previous papers have used a more conservative approach and have focused only on observations in which debt issuers and debt guarantors coincide.
Instead, we use the following rule: if a debt security is issued by an entity and guaranteed by another, we match the deal with the corresponding information of the firm guaranteeing it. We interpret that the issuer entity is transferring upstream its risk to the guarantor. This criterion applies to all issuer entities: non-financial affiliates and offshore/onshore financial vehicles. If, alternatively, an entity issues debt without explicit guarantee of another company, we use the issuer information. Most probably, the issuer will be a non-financial affiliate, fully independent from its parent company.
Figure 1 provides an example in which a company guarantees the debt securities issued by an affiliate, which can be incorporated domestically or overseas. This company has a second affiliate, which is financially independent: the debt securities it issues do not receive any explicit guarantee.
This criterion prevents two problems. First, it ensures gathering systematically all the debt guaranteed by emerging economies firms: we obtain information on all debt whose financial risks lay in an emerging economy. Second, the criterion mimics the investment decision process of international investors, as the focus lies on the entity backing the debt securities. For the purpose of understanding firms’ choices, it is important to mirror their approach.
The organizational structures of emerging economies firms have two features which reinforce the importance of using this criterion. First, firms use financial vehicles to issue securities in international markets. In these cases, investors price the risk analyzing the guarantors of the debt. Second, emerging economy firms’ have large non-financial affiliates incorporated all over the world with different degrees of autonomy from their controlling interest.
Figure 2 decomposes the total volume issued by emerging economies firms according to the country of incorporation of the debt issuer. Debt issued onshore refers to debt in which the country of the debt issuer and the debt guarantor coincide. In the other two groups, they differ. Debt issued from offshore (financial or non-financial) centers refers to that issued by entities incorporated in a country different from that of the debt guarantor. The gap between onshore and offshore financing is a measure of the debt which would be improperly classified – or remain hidden – had not we introduced our criterion. Offshore financing accounts for 70 per cent of total debt guaranteed by emerging economy firms in emerging Europe and close to 50 per cent in Africa and Middle East; in Latin-America and emerging Asia, it represents 28 and 25 per cent, respectively. Debt issued from non-offshore centers is noticeably high in Africa and Middle East and emerging Asia, underscoring the importance of guarantees extended by their multinationals on debt issued by foreign non-financial affiliates.
Identifying the ultimate guarantor of a debt security is challenging. Hence, as a short cut, it has been proposed to obtain debt securities issued by firms headquartered in emerging economies and all their affiliates, irrespective of the support they receive from their parent companies.
On aggregate basis, this measure is referred to as the nationality measure. This measure is also flawed; it has two biases: there are foreign affiliates of emerging economy firm which are standalone entities and should not be included (e.g. Jaguar-Land Rover from the UK is a standalone affiliate of Tata Motors from India); and there are standalone affiliates of advanced economy firm in emerging economies, which should be included (e.g. Kansas City Southern de Mexico from Mexico is a standalone affiliate of Kansas City Southern Lines from the USA).
As these biases have opposite effects, they could offset each other and go unnoticed in aggregate analyses. Finally, the criterion proposed has another advantage; it provides a globally consistent breakdown of international debt among firms. Affiliates whose debt is not guaranteed by their parent companies are treated as independent entities, preventing double counting.
3.3 Univariate analysis: deal characteristics and firms’ features in global and institutional/wholesale markets
Table III shows the total amount raised per year by market of issuance. This is a very large number, as we are focusing on non-financial firms and excluding firms from large debt issuers such as China. The bulk of debt securities guaranteed by emerging economy firms are issued in the Eurobond market, though its relative importance has decreased over time. Debt securities issuances in the US144A bond market are second in importance; in the last year, they experienced a noticeable increase. The number of debt securities issued in the global bond markets is comparatively much smaller.
Figure 3 displays the total amount raised by region and breaks it down by year. Latin-American firms have guaranteed debt securities for an amount of more than US$600bn; emerging Asian firms are second in importance and have guaranteed for US$250bn and emerging Europe and Africa and Middle East firms stand as third and fourth, with a total of US$250 and 171bn, respectively (see Table AIII for a country breakdown).
For each debt security, we obtain details of the structure of the operation: the amount issued, the maturity, yield, currency of denomination, issuer name and its country, guarantor name and its country, ultimate parent company name and its country and market of issuance. Besides, we obtain information on the type of security issued: if it is a straight bond, whether it has embedded call or put options, has convertible rights, has a sinking fund, has any combination of these features (i.e. sinking fund plus call option), whether it is registered in the SEC (or it is a bearer bond), etc.
Next, we match each debt security with the financial and non-financial information of the firm guaranteeing it. We use firms’ financial information to compute relevant financial ratios: interest coverage ratio, ROA, current ratio, debt to total assets, fixed assets to total assets, the Altman-score, etc. The non-financial information includes the firm industry, total assets, number of employees, reporting GAAP, listing status in the local market, cross-listing in the US exchange market.
To examine how firms choose among the different international debt markets, we create a firm-year database. This adjustment is important as some firms issue debt securities on a delayed basis (particularly in Eurobond market), dividing their annual funding needs into several tranches. Treating them as different deals could introduce biases: deals will appear as smaller, and markets in which delayed issuance is more frequent would become overrepresented. Similar methodological decisions have been taken in previous research (Esho et al., 2001). Then, we classify firms according to the market in which they issue debt in three different groups: firms able to issue in the global bond market, firms issuing debt only in the US144A private debt market and firms issuing only in the Eurobond market.
Next, we compare the characteristics of the deals and the type of borrowers among the global (public) and institutional/wholesale markets. We use descriptive statistics, kernel estimation and tests of stochastic dominance. Table IV shows descriptive statistics. All variables shown are firm-specific. They can be classified in two groups: firm-level information and variables related to the type of financial contracts that firms subscribe. We refer to them as financial contract characteristics that include measures of the type of debt securities and are a valuable source of information of firms’ credit quality, which reflects how lenders assess their risk.
3.3.1 Panel A reports firm information.
Firms guaranteeing debt issued in institutional/wholesale debt securities markets are smaller and are less frequently cross-listed in the USA through ADR. This is consistent with our expectation: firms with more capacity to absorb flotation costs will issue debt in public debt markets. Firms without access to the global market show more severe informational asymmetries: have a lower proportion of fixed assets to total assets and are more likely to report their financial statements using a local GAAP.
Also, firms issuing debt only in institutional/wholesale markets seem to have weaker financial conditions. They show lower profitability (ROA) and less capacity to pay interest expenses interest coverage ratio (ICR). Overall, firms without access to the global market show worst financial conditions.
3.3.2 Panel B shows the financial contract characteristics.
Under this heading we include the features of the debt securities issued by firms in a given year: total amount issued, maturity of security, existence of rating, etc. As expected, the amount issued by firms in the Eurobond and US144A markets is smaller.
The maturity of debt is shorter, consistent with lower informational asymmetries. Firms are more likely to issue bonds with sinking funds, reflecting a perceived lower quality, and to include convertible rights as decreasing-value concessions to reduce their funding costs. Their debt securities are less frequently rated, especially in the Eurobond market, consistent with the high asymmetry of information in this market. Overall, the proportion of firms granting any sort of bond holder rights (sinking funds, convertible rights or put options) in the global market is substantially lower. Table V shows the results of the tests of identity of distributions (continuous variables) and proportions (categorical variables) for the variables in Table IV among the three groups of firms. Most of the tests reject the null of equal distributions/proportions, implying that the three groups of firms and the type of bonds issued by each group have different characteristics. The tests also reject the null that firms issuing in the Eurobond and US144A markets are similar. This is important because even though both markets share common features (less regulation and wholesale investors), the issuers and the characteristics of the deals are not the same. This will be crucial when conducting multivariate analysis.
Figure 4 plots the density function of a number of variables for different types of firms: firms with access to global bond market; firms issuing debt only in the Eurobond market; and firms issuing debt only in the US144A market. Visual inspection suggests that firms with access to global bond markets are larger, have a better ratio of fixed assets to total assess and issue at longer maturities. These firms also have better interest coverage ratios and larger ROA.
Firms issuing only in the Eurobond market are smaller and have lower ratio of fixed assets than those issuing in the US144A market. Finally, we further analyze these apparent relations by conducting empirical tests of stochastic dominance for the distributions of different variables among the three groups of firms. When conducting the tests, we follow the procedures given by Barrett and Donald (2003). Unlike test of equality of distributions, these tests rank unequivocally independent distributions; for instance, they determine if the distribution of firms with access to the global bond market have larger assets than the firms in any of the other two groups at any level of total assets. The results of the analysis are shown in the third column of Table VI. They confirm the different features of each group of issuers. Firms with access to global markets are larger, have a higher fixed assets ratio and issue larger amounts at longer maturities. This is also true for firms only issuing in the US144A market compared to firms issuing only in the Eurobond market (Table VI, Panel C). Regarding financial conditions, firms with access to global markets show better balance sheet ratios overall. Issuers in the Eurobond market seem to have more deteriorated ratios and less capacity to absorb flotation costs.
The analysis of the distributions of firm-level information confirms that differences in the type of firms acceding to each market are not random. They suggest that poorer credit quality, more severe informational asymmetries and less ability to cope with high flotation costs lead firms to issue in the institutional/wholesale markets, though we cannot claim causality. Differences in the distributions can stem from an unobserved variable. For instance, larger firms have easier access to the global bond market; if the firm size is positively correlated with cross-listing in the USA through ADR, the unconditional distribution can lead to erroneous inference about the impact of cross-listing on firms’ choices.
4. Multivariate analysis of firms’ choice of market of issuance
In this section, we investigate firms’ incremental debt choice among their existing options in international debt securities markets. Conditional on their decision to issue a debt security in an international market, they can choose between global, US144A and Eurobond markets. Thus, this exercise focuses on the marginal decision of firms, reporting the results of multinomial logit regressions and ordered logit regression. This type of estimation has already being used in previous literature analyzing the determinants of the debt choice (Denis and Mihov, 2003; Altunbas et al., 2010; Esho et al., 2001). For example, Altunbas et al (2010) analyzed the marginal choice of issuing debt between the corporate bond market and the syndicated loans market for European firms. All independent variables are firm-specific. Common factors are captured by the intercept and the time dummies.
4.1 Multinomial logit
Given the results obtained from the univariate analysis, we need to treat the three groups of firms separately and study how firms choose among the three alternative options: global bond market (baseline option), US144A and Eurobond markets. We fit a multinomial logit to do so. This specification allows for potential differences in the slope of the coefficient and relaxes the proportional odd ratios assumption that holds in an ordered logit specification. In the first model, we include only the financial contracting variables, and in the second model, we include the firm variables as well. Table VII shows the results.
First, the decision to issue in the global bond market depends positively on the total amount issued; but it is not a determinant on firms’ choice for the US144A bond market. Not being cross-listed in the US exchanges through ADRs implies that firms are more likely to issue securities in both the US144A and Eurobond markets. Firm size has a negative impact on the decision to issue in the Eurobond and US144A bond markets. Overall, flotation costs impact firms’ choices: the amount issued and firm size introduce a wedge between firms; those large enough will issue in the global bond market.
Second, informational asymmetries are a key factor behind firms’ choice of US144A and Eurobond markets, but they are evident in different financial contract characteristics. Firms issuing unrated debt are more likely to enter the Eurobond market; this is not a significant factor behind firms issuing in the US144A bond market. On the other hand, firms granting bondholders rights (sinking fund, convertible rights, any combination of them and other embedded options) are more likely to enter institutional/wholesale markets. Bond maturities impact negatively on firms’ decision to issue debt in any of the two institutional/wholesale markets, and the ratio of fixed assets to total assets is not significant. Third, firms with higher risk of financial distress are less likely to issue debt in the Eurobond market.
This may indicate that having healthy financial ratios is more relevant for the small firms issuing in the Eurobond market than for the large and well-known firms with access to the global market. The opposite happens with firms issuing in the USD144A market; firms with higher risk of financial distress tend to issue in that market, although the coefficient is not significant. Market volatility makes firms less likely to issue in the Eurobond market. This is consistent with firms being better able to choose the market timing in institutional/wholesale markets. Finally, firms are more likely to issue in both types of institutional/wholesale debt securities markets after the global financial crisis – the market-based dummy has a negative and similar impact in both columns.
Table VIII shows the marginal effects for the multinomial logit. They show the change in the probability of issuing in a given market after a standard deviation change in continuous variables and a unit change in categorical variables. By construction, the sum of the three changes is zero. The last row shows the baseline values or frequency of firms issuing in each of these markets.
4.2 Ordered logit: Eurobond, US144A and global bond markets
The multinomial logit does not establish any order between the different alternatives. We want to investigate if the ranking in regulatory stringency leads to a segmentation of borrowers across markets. We fit an ordered logit defining the global bond market as the base outcome, the one where firms provide more public information and comply with more regulatory requirements. The other two options are the US144A bond market and the Eurobond market, information is less public in the US144A bond market. The ordered logit has a critical assumption: the coefficients are equivalent in the three potential bivariate estimations; thus, it imposes such restriction on the coefficients, minimizing the number of parameters estimated. This is so-called parallel regression assumption. It is potentially restrictive as it implies that odd ratios do not change and slope coefficients are identical in all the three options.
Table IX shows the results for the specification including only the financial contracting variables; and Table X expands the analysis and includes firm information. We find that the less a firm is able to cope with high flotation costs, the more probable it opts to access institutional/wholesale markets (negative sign of amount issued); and firms with higher asymmetries of information will head towards less public markets (local GAAP, term and rated security).
Next, we run tests of model specification. First, we run a test of identity of the cut points (also known as thresholds). The test does not reject the null hypothesis; two cut points identified are different, although the Brant tests reject the parallel regression assumption. Overall, the results suggest that the three choices are different: US144A and Eurobond markets are two different options firms have. Though the differences between the US144A and Eurobond markets are not constant across variables, the results of the Brant tests suggest that the slope coefficients are different in the bivariate logits; the restrictions on coefficients imposed by the ordered logit are too restrictive.
Table XI reports the marginal effects. By construction, the change in the probability of choosing the outer options (global bond market as the more public debt market and Eurobond market as the less public debt market) have opposite signs, and the sign of the inner option – the US144A bond market – is uncertain; and the sum of the three changes is zero. When the ability to absorb floatation cost improves (larger amount and the existence of an ADR), the probability that a firm chooses the US144A bond market increases; similarly, if informational asymmetries decrease (rated securities, maturities, bond holder rights and local GAAP), firms switch from the Eurobond to the US144A market. An increase in firms’ risk of financial distress (Altman score below 1.21) has a similar impact: the probability that a firm chooses the Eurobond market decreases at the expenses of the US144A market. Overall, this suggests a migration of firms from the Eurobond market to the inner option, confirming that US144A and Eurobond markets are more akin than US144A and global bond markets are.
4.3 Generalized ordered logit
The Brant tests reject the parallel regression assumption of the ordered logit. This assumption states that the odd ratios are constant across all alternatives, i.e. there is a single set of coefficients for all covariates. The rejection of this hypothesis suggests that the restrictions of the ordinal logit model are too stringent, though this does not rule out the existence of an ordinal raking between the three primary bond markets. To investigate this possibility, we fit a generalized ordered logit. This estimation method does not impose ex ante any restriction on the coefficients between the different choices; it tests their existence, estimating restricted and unrestricted models sequentially for each of the variables. Overall, the estimator is flexible enough to nest the multinomial and the ordinal models as particular cases. The results are shown in Table XII. Marginal effects are shown in Table XIII. In the first model, we do not include firms’ financial variables (Columns 1 and 2).
The results suggest that the differences between the odd ratios of global bond issuers and US144A issuers with respect to Eurobond issuers are related to the cost of fulfilling with the reporting standards of US public offerings. More formally, the iterative estimation identifies that the parallel regression assumption does not hold for maturity, local GAAP and ADR. When including balance sheet variables, the restrictions are not imposed on local GAAP and total assets. Overall, the results show that the ranking between the three primary markets is not linear in several dimensions related with informational asymmetries and ability to cope with regulatory requirements of public offerings. Indeed, global issuers are identified as more different than the joint group of US144A and Eurobond issuers. More specifically, having a local GAAP makes unlikely than a firm issues in the global bond market; it does not make a difference when it comes to explain the choice between Eurobonds and the other markets. Similarly, the effect of ADR decreases: it is quite important in explaining the choice between global and the other markets; less so for the choice between Eurobonds and other markets. When we include financial ratios, we find a similar effect for total assets.
Marginal effects – shown in Table XIII – synthesize these differences. Filing with a local GAAP, or having ADR, decreases the chances of issuing in the global bond market. However, the odds of issuing in US144A and Eurobond markets are not evenly split: having a local GAAP increases the chances of issuing through Rule 144A and less through the Eurobond market.
4.4 Robustness checks
As we mentioned when analyzing the decision of issuing debt, firms may choose to tap less regulated markets because of the speed of issuance or the need to diversify their funding sources, even though they are able to cope with the regulatory requirements imposed by a public offering. We take into account these circumstances by excluding firms which have ever issued a global bond in our sample from the US144A or Eurobond issuers. Indeed, these firms might be still subject to ongoing disclosure requirements due to previous global bond issues. Many of the features which make them able to comply with regulatory requirements are likely to hold over time – reporting GAAP and existence of a US ADR. The results are robust to the exclusion of these firms, which consistent with our prior studies (Table AVI).
We also examine the differences between the two types of firms which have access to the global bond market: the switchers and non-switchers. We have argued that switchers should be classified as firms with access to the global bond market. We conjecture that firms issuing also in another market seek other advantages –speed of issuance, bespoke financing conditions, funding diversification – and not to a sudden or temporary inability to issue in the global bond market. Consistent with our expectations, the non-parametric analysis suggests that switchers are larger and sounder than non-switchers (Table AVII). They are statistically different in some dimensions, larger size, higher ratio of fixed assets to total assets, higher profitability and better Altman score, though they seem to invest less, signaling that they are more mature companies.
So far, we are not controlling for local shocks to the credit supply, which could impact on firms’ market choice, thus distorting our results. To cross-check that there is no omitted variable bias, we estimate the model by including two alternative measures of local credit supply as additional covariates: the average local interest rates in the last month before the issuance and its rate of growth in the last month. We summarize the main results (not reported for the sake of brevity), focusing on the ordered logit model. The results confirm our previous conclusions. Moreover, we find that local interest rates have a negative impact (statistically significant at 10 per cent) on the likelihood of issuing in less regulated markets. In contrast, the rate of growth has no effect. Overall, firms based in countries with poorer access to credit (i.e. higher funding costs) seem less likely to borrow in less regulated markets.
Finally, omitted country and industry time invariant factors could be biasing our results. To cross-check it is not the case, we re-estimate the ordered logit with firm information including country and industry dummies. Some variables become statistically insignificant, which in part is due to the loss of degrees of freedom, but our conclusions hold: the probability of issuing in less regulated markets is negatively related to firms’ size, MOVE index, fixed assets ratio and existence of rating; it is positively related to reporting with a local GAAP and increases in the post-crisis period. The results underscore that firms with less ability to absorb flotation costs and more informational asymmetries issue in less regulated markets; thus, they confirm that our main conclusions are not driven by omitted variable bias.
In this paper, we investigated how non-financial borrowers of emerging economies choose among the existing international bond markets: global, US144A and Eurobond markets. First, we showed that monitoring the entities guaranteeing the debt securities is key to prevent biases. Thus, to carry out the analysis, we matched data on debt securities with the corresponding firm-level information on the firm guaranteeing it. This way, we accomplished a comprehensive database of 3,944 debt securities and over US$1.2tn.
We have discussed how the regulation of international debt markets is uneven: global and foreign bond markets are strictly regulated; the regulation of the US144A and, in particular the Eurobond market, is much lighter. The results of univariate and multivariate analyses confirm corporate finance theory predictions: firms of poorer credit quality, less ability to absorb high flotation costs and exhibiting more informational asymmetries tend to issue in the less regulated debt markets. This implies that credit risks are more severe in the US144A and the Eurobond markets. Moreover, the propensity to issue in these markets has increased after the global financial crisis.
This can reflect the reduction in liquidity in public debt markets after the global financial crisis.
These results have far-reaching policy implications. They suggest that there is a two-tier system of regulation in international debt markets. The sizable regulatory gap implies that only very large and sound firms, with low informational asymmetries, find it convenient to issue in public bond markets. Reforms aimed at making easier the access to public debt markets could be helpful. Local regulators could enhance firms’ access to global bond markets by converging toward international reporting standards. Regulators in key jurisdictions could collaborate to reduce the costs that issuing in the public bond market has for foreign firms.
This paper leaves some unanswered questions for further research. The first one has to do with the role of local debt markets. The development of local bond markets could either foster or restrain the access to international debt markets. Well-developed local bond markets reduce the need to obtain external financing. But they imply more familiarity with bond financing and thus could ease access to external markets. Moreover, if local markets are developed, regulators could have very demanding country-specific regulation such as a local reporting GAAP; thus, local firms might find costly to raise funds in the fully regulated international debt markets and head toward the Eurobond market. The second question has to do with advanced economy firms and their access to international debt markets. Do they follow the same pattern? These economies differ in the stringency of local surveillance, legal systems and institutions. More stringent local surveillance could imply less reliance in regulated debt markets. However, firms in advanced economies might have easier access to these markets because of lower informational asymmetries and better legal systems and institutions. At the end of the day, these are empirical questions which deserve further analyses. Answering them is the key to understand the risks that market-based financing poses for global financial stability.
Markets of international debt securities issuances
|Characteristics||Global bond market||US Rule 144A bond market||Eurodollar bond market|
|Type of offering||Public offering||Institutional/wholesale offering||Institutional/wholesale offering and private|
|Description||Public offering in at least one foreign local market; we assume it involves placement in USA||US institutional offering, following Rule 144A||EU institutional offering and private offshore offering|
|Regulation||US regulation, SEC full regulation||US regulation, SEC Rule 144A exemptions||MIFID and exchange-regulated|
|Disclosure requirements||High. Firms need to file form 20-K; financial accounts using US GAAP||Medium||Medium/low|
|Type of investors||No restrictions; includes retail investors||Qualified institutional buyers||Qualified institutional buyers|
Sources: Own elaboration; see also Gao (2011)
|Firm-specific features||Variables (expected sign)||Description|
|Flotation costs||Firm total assets (−)|
|Amount issued (−)|
|US ADR (−)||1 if firm is cross-listed in the USA through ADR|
|Local GAAP (+)||1 if firm files using a local GAAP|
|Informational asymmetries||Fixed assets to total assets (−)|
|Firm total assets (−)|
|Local GAAP (+)||1 if firm files using a local GAAP|
|Credit rating on the debt security (−)||1 if firms issue bonds with credit rating (any)|
|Bond with convertible rights (+)||1 if firm issues bonds with convertible rights|
|Five-year growth in firm total assets (+)|
|Liquidation and renegotiation||Altman score (−)||1 if Altman score is below 1.21 (distress zone)|
|Bond with s1nking fund (+)||1 if firm issues bonds with sinking fund|
|Fixed assets to total assets (−)|
|Primary market efficiency and global conditions||Merrill Lynch option volatility estimate (+/−)||Average MOVE in 20 working days before issuance|
|General||Bond holders’ rights (+)||1 if firm issues bonds with sinking fund put option, or convertible rights|
|Bond with call option (−)||1 if firm issues bonds with call option|
|Bond with put option (+)||1 if firm issues bonds with put option|
Main variables and expected impact on likelihood of firms to issue through institutional/wholesale offering
Total amount issued per year (billion dollars)
|Year||Total||Global bond markets||US144A bond markets||Eurodollar bond market|
Breakdown by market of issuance
|Characteristics||Firms with access to global market||Firms issuing only on Eurobond market||Firms issuing only on US144A market|
|A. Firm information|
|Total assets (million dollars)||11,620||936||4,295|
|Fixed assets to total assets||0.58||0.38||0.46|
|Interest coverage ratio||3.47||2.55||2.88|
|Local GAAP (%)||6||27||20|
|US ADR (%)||48||13||25|
|Ratio LT/ST debt||5.25||1.77||3.56|
|B. Financial contract characteristics|
|Amount issued million dollars)||650||55||300|
|Sinking fund (%)||4||6||18|
|Convertible rights (%)||6||29||6|
|Call option (%)||32||15||38|
|Put option (%)||3||35||4|
|Bond holder rights (%)||9||33||23|
The table shows median values for firm-year observations; descriptive statistics (median values)
Tests of identity of distributions/proportions
|Characteristics||Global issuers vs Eurobond issuers||Global issuers vs UA144A issuers||Eurobond issuers vs US144A issuers|
|A. Firm information|
|Total assets (million dollars)||−10.41***||−4.41***||9.20***|
|Fixed assets to total assets||−6.62***||−2.84***||4.18***|
|Interest coverage ratio||−2.76***||−1.21||1.93**|
|Local GAAP (%)||5.62***||3.84***||−2.73***|
|US ADR (%)||−11.19***||−4.74***||5.76***|
|Ratio LT/ST debt||−7.41***||−2.45***||5.62***|
|B. Financial contract characteristics|
|Amount issued (million dollars)||−12.06***||−4.82***||14.34***|
|Sinking fund (%)||0.73||4.02***||8.03***|
|Convertible rights (%)||5.85***||−0.23||−8.71***|
|Call option (%)||−5.04***||1.35||9.70***|
|Put option (%)||7.84***||0.71||−10.89***|
|Bond holder rights (%)||5.98***||3.60***||−3.47***|
Global issuers refer to those firms with access to the global market and Eurobond (US144A) issuers refer to firms issuing only in the Eurobond (US144A) market. The table shows median values for firm-year observations. Test of identity of distributions: Wilkoxon rank-sum test for continuous variables; tests of equality of proportions for categorical variables. Table shows the z-values; *, ** and *** denote statistical significance at 10%, 5% and 1%, significance, respectively; and null is identity (z-values)
Tests of stochastic dominance in EMEs NFCs (2000-2014; p-values)
|Test GLOBALSD1 Eurobond||Test EUROBONDSD1 Global||SD1: distribution preferred|
|A: global vs Eurobond|
|Fixed assets to total assets||1.000||0.000***||GLOBAL|
|Interest coverage ratio||0.771||0.004***||GLOBAL|
|B: global vs US144A|
|Fixed assets to total assets||1.000||0.004***||GLOBAL|
|Interest coverage ratio||0.727||0.110||–|
|C: Eurobond vs US144A|
|Test EUROBOND SD1 US144A||Test US144A SD1 EUROBOND||SD1: distribution preferred|
|Fixed assets to total assets||0.000***||0.978||US144a|
|Interest coverage ratio||0.007||0.931||US144a|
Note: *, ** and ***denote rejecting the null of SD at the significance level of 10%, 5% and 1% significance, respectively
Source: Own elaboration
Determinants of market of issuance of debt securities
|Variables||Financial contracting features||Including firm information|
|Amount issued||0.664*** [0.062]||0.432*** [0.041]||0.651*** [0.030]|
|Term||0.969** [0.013]||0.932*** [0.014]||0.962** [0.014]||0.960** [0.017]||0.906*** [0.022]||0.943*** [0.019]|
|Local GAAP||2.858*** [1.141]||2.683*** [1.002]||0.939 [0.171]||3.125** [1.519]||2.668** [1.216]||0.854 [0.195]|
|ADR||0.505*** [0.124]||0.480*** [0.110]||0.951 [0.163]||0.621* [0.174]||0.462*** [0.113]||0.745 [0.145]|
|MOVE index||0.997 [0.004]||0.991** [0.004]||0.994** [0.003]||0.999 [0.006]||0.994 [0.005]||0.996 [0.004]|
|Rated security||0.699 [0.172]||1.515* [0.350]||2.169*** [0.327]||0.896 [0.275]||1.796** [0.498]||2.005*** [0.406]|
|Bond holder rights||1.847* [0.618]||1.621 [0.510]||0.878 [0.150]||4.503*** [2.615]||4.964*** [2.751]||1.103 [0.270]|
|No financial information||0.964 [0.246]||0.920 [0.221]||0.954 [0.143]|
|Market-based||2.538*** [0.620]||1.803** [0.427]||0.711** [0.111]||2.333*** [0.747]||1.733* [0.524]||0.743 [0.149]|
|Firm assets||0.845* [0.076]||0.720*** [0.063]||0.852*** [0.043]|
|Fixed assets to total asset||0.999 [0.007]||0.995 [0.006]||0.997 [0.004]|
|Altman score < 1.21||0.749 [0.209]||0.590**[0.147]||0.787 [0.143]|
|Constant||2.486 [1.399]||11.014*** [5.972]||4.431*** [1.569]||10.844*** [9.648]||429.940*** [360.850]||39.646*** [20.307]|
Robust standard errors in brackets;
p < 0.01,
p < 0.05,
p < 0.1; multivariate logit, results reported as odd ratios. In Columns 1, 2, 4 and 5, we report odd ratios for US144A and Eurobond issuers relative to firms being able to issue in global bond market; in Columns 3 and 6, we report odd ratios of Eurobond issuers versus firms able to issue in US144A; multinomial logit; odd ratios
Marginal effects in multinomial logit
|Characteristics||Global bond||US144A||Eurobond market|
|Panel A: financial contracting features|
|Bond holder right||−2.36%||3.79%||−1.42%|
|No FS information||−2.01%||0.95%||1.06%|
|Baseline values (%)||5.47||11.79||82.75|
|Panel B: including firm information|
|Bond holder right||−5.25%||5.54%||−0.29%|
|Altman score < 1.21||2.09%||8.51%||−10.60%|
|Fixed assets to total assets||0.65%||−0.09%||−0.57%|
|Baseline values (%)||8.05||14.67||77.28|
Marginal effects computed on binary changes in categorical variables; and one-standard deviation in continuous variables. p-values of the test change is 0, reported below the marginal effects
|Variables||Model 1||Model 2||Model 3||Model 4|
|Amount issued||−0.338*** [0.050]||−0.405*** [0.057]||−0.405*** [0.057]||−0.354*** [0.052]|
|Term||−0.050*** [0.012]||−0.054*** [0.012]||−0.054*** [0.012]||−0.050*** [0.012]|
|Local GAAP||0.260* [0.149]||0.161 [0.160]||0.16 [0.163]||0.274* [0.153]|
|ADR||−0.427*** [0.147]||−0.336** [0.148]||−0.335** [0.148]||−0.418*** [0.147]|
|MOVE index||−0.010*** [0.002]||−0.010** [0.005]||−0.010** [0.005]||−0.009*** [0.002]|
|Rated security||−0.934*** [0.178]||−0.807*** [0.181]||−0.803*** [0.189]||−0.960*** [0.188]|
|Bond holder right||0.01 [0.153]||−0.066 [0.154]|
|No financial information||−0.005 [0.135]||0.124 [0.139]||0.124 [0.139]||0.02 [0.138]|
|First cut point||−6.454*** [0.321]||−5.817*** [0.569]||−5.813*** [0.570]||−6.386*** [0.334]|
|Second cut point||−5.008*** [0.308]||−4.274*** [0.568]||−4.270*** [0.569]||−4.938*** [0.320]|
|Test cut points||−11.46||−10.09||−10.08||−11.32|
|Probability > χ2||0||0||0||0|
Robust standard errors in brackets;
p < 0.01,
p < 0.05,
p < 0.1; ordered logit, base outcome is firm is able to issue in global bond market; 1 if it issues in 144A bond market; 2 if it issues in the Eurobond market; Eurodollar, US144A and global bond markets
|Independent variables||Model 1||Model 2||Model 3||Model 4||Model 5|
|Firm assets||−0.182*** [0.041]||−0.193*** [0.045]||−0.130*** [0.051]||−0.144*** [0.048]||−0.140*** [0.050]|
|Term||−0.061*** [0.016]||−0.067*** [0.017]||−0.064*** [0.018]||−0.056*** [0.017]||−0.056*** [0.017]|
|Local GAAP||0.193 [0.175]||0.118 [0.186]||0.108 [0.197]||0.187 [0.184]||0.189 [0.184]|
|ADR||−0.436*** [0.154]||−0.337** [0.155]||−0.345** [0.159]||−0.427*** [0.158]||−0.432*** [0.160]|
|MOVE index||−0.009*** [0.003]||−0.015*** [0.006]||−0.017*** [0.006]||−0.007** [0.003]||−0.007** [0.003]|
|Rated security||−1.397*** [0.221]||−1.444*** [0.231]||−1.392*** [0.233]||−1.391*** [0.225]||−1.380*** [0.227]|
|Bond holder right||−0.168 [0.201]||−0.045 [0.206]||0.097 [0.221]||−0.005 [0.225]||0 [0.225]|
|Market based||0.326* [0.189]||0.309 [0.192]|
|Altman score < 1.21||−0.352** [0.155]||−0.407*** [0.151]||−0.391** [0.157]|
|Fixed assets to total assets||−0.001 [0.004]|
|First cut point||−6.381*** [0.406]||−6.185*** [0.703]||−5.909*** [0.800]||−5.903*** [0.470]||−5.872*** [0.483]|
|Second cut point||−5.066*** [0.397]||−4.779*** [0.703]||−4.445*** [0.799]||−4.535*** [0.460]||−4.504*** [0.473]|
|Probability > χ2||0||0||0||0||0|
Robust standard errors in brackets;
p < 0.01,
p < 0.05,
p < 0.1;
ordered logit, base outcome is firm is able to issue in global bond market; 1 if it issues in 144A bond market; 2 if it issues in Eurobond market, including firm information: Eurodollar, US144A and global bond markets
Marginal effects in ordered logit
|Independent variables||Global bond||US144A||Eurobond market|
|A: financial contracting features|
|Bond holder right||0.33%||0.46%||−0.79%|
|No financial statement||−0.76%||−1.07%||1.83%|
|Baseline values (%)||6||12||83|
|B: including financial ratios|
|Bond holder right||0.00%||0.00%||−0.01%|
|Altman score < 1.21||2.76%||3.07%||−5.83%|
|MOVE index 1.36%||1.43%||−2.79%|
|Fixed assets to total assets 0.18%||0.20%||−0.38%|
|Baseline values (%)||8||15||77|
Marginal effects computed on binary changes in categorical variables; and one-standard deviation in continuous variables. p-values of the test change is 0, reported below the marginal effects
Generalized ordered logit
|Independent variables||Model 1||Model 2|
|Global vs Eurobonds and US114A||Global and US144A vs Eurobonds||Global vs Eurobonds and US114A||Global and US144A vs Eurobonds|
|Amount issued||−0.237*** (0.000)||−0.237*** (0.000)|
|Maturity (years)||−0.0756*** (0.000)||−0.0526*** (0.000)||−0.0513*** (0.001)||−0.0513*** (0.001)|
|Local GAAP||1.347*** (0.000)||0.114 (0.471)||1.254** (0.004)||0.0121 (0.950)|
|ADR (cross-listed in the USA)||−0.962*** (0.000)||−0.363* (0.011)||−0.480** (0.002)||−0.480** (0.002)|
|MOVE index||−0.00895*** (0.000)||−0.00895*** (0.000)||−0.00539 (0.125)||−0.00539 (0.125)|
|Rated security||−1.241*** (0.000)||−1.241*** (0.000)||−1.396*** (0.000)||−1.396*** (0.000)|
|Bond holder rights||0.216 (0.127)||0.216 (0.127)||−0.165 (0.331)||−0.165 (0.331)|
|Post 2009||0.0691 (0.616)||0.0691 (0.616)||0.312 (0.090)||0.312 (0.090)|
|No financial information||−0.0167 (0.898)||−0.0167 (0.898)|
|Total assets||−0.311*** (0.000)||−0.127* (0.016)|
|Altman score < 1.23||−0.372* (0.014)||−0.372* (0.014)|
|Fixed assets to total assets||−0.00157 (0.671)||−0.00157 (0.671)|
|Constant||6.125*** (0.000)||4.373*** (0.000)||7.295*** (0.000)||4.464*** (0.000)|
|Probability > χ2||0||0|
In the first model, the coefficients for which the parallel regression assumption is not imposed are maturity, local GAAP and ADR; and in the second model, the coefficients for which the parallel regression assumption is not imposed are total assets and local GAAP. To identify the variables with the unrestricted parameters, we used the iterative process of testing logit models with restricted coefficients and testing p-values in parentheses,
p < 0.01,
p < 0.05,
p < 0.1
Generalized ordered logit
|Independent variables||Panel A: financial contract features||Panel B: including financial ratios|
|Bond holder right||−1.03%||−1.63%||2.66%||0.95%||1.20%||−2.15%|
|No financial statement||0.07%||0.11%||−0.18%|
|Altman score < 1.21||3.44%||4.27%||−7.71%|
|Fixed assets to total assets||0.67%||0.82%||−1.49%|
|Baseline values (%)||5.56||11.82||82.63||8.12||14.50||77.38|
In the first model, the coefficients for which the parallel regression assumption is not imposed are maturity, local GAAP and ADR; and in the second model, the coefficients for which the parallel regression assumption is not imposed are total assets and local GAAP; marginal effects
Description of firm-level variables
|Total assets||Logarithm of total assets|
|Altman score||We compute the Altman score for private companies: 0.717 × x1 + 0.847 × x2 + 3.107 × x3 + 0.42 × x4 + 0.998 × x5; where x1 is working capital to total assets, x2 is retained earnings to total assets, x3 is EBITDA to total assets, x4 is book value of equity to total debt and x5 is sales revenue to total assets|
|Dummy Altman < 1.21||Binary variable takes value 1 if the firm has an Altman score lower than 1.21|
|Fixed assets to total assets||Net property, plant and equipment to total assets. Winsorized at 1% and 99%|
|Leverage||Leverage is total debt to total assets. We winsorize it at percentiles 1% and 99%. Deviation with respect to the worldwide industry median; industry defined using Bloomberg Industries|
|ROA||ROA is EBITDA to assets. We winsorize it at 1% and 99%. Deviation with respect to the worldwide industry median; industry defined using Bloomberg Industries|
|Current ratio||Current ratio is the ratio of liquid assets to Winsorized at 1% and 99%. Deviation with respect to the worldwide industry median; industry defined using Bloomberg Industries|
|Interest coverage ratio||Winsorized at 1% and 99%. Deviation with respect to the worldwide industry median; industry defined using Bloomberg Industries|
|Accounting standard||Categorical variable with firm accounting standard. We construct a binary variable takes value 1 if the firm reports using a local GAAP (all GAAPs excluding US GAAP and IFRS are considered local)|
|US ADR||Categorical variable which describes the US exchanges in which a firm is cross-listed. We construct a binary variable taking value 1 if the firm is cross-listed in a US exchange (excluding OTC exchanges)|
|Listing status||Binary variables take the value of 1 if the firm is listed in a local exchange|
Variables are accessed using Bloomberg; ultimate source are financial statements filed by firms
Source: Own elaboration
Description of deal-level variables
|Amount issued (million dollars)||Logarithm of the amount issued|
|Term||Maturity of the debt security in years|
|Sinking fund||Binary variable takes value 1 if the debt security has a sinking fund|
|Convertible rights||Binary variable takes value 1 if the debt security grants convertible rights|
|Call option||Binary variable takes value 1 if the debt security has an embedded call option|
|Put option||Binary variable takes value 1 if the debt security has an embedded put option|
|Bond holder rights||Binary variable takes value 1 if the debt security protects bond holders by having any of the following features: convertible rights, sinking fund; any combination of options including at least convertible rights or sinking funds|
|Rating||Binary variable for securities which are rated|
|Country of risk||Categorical variable: ISO code of the issuer’s country of risk, computed using four factors: management location, country of primary listing, country of revenue and reporting currency|
|Country of incorporation||Categorical variable: ISO code of the country of incorporation of the issuer|
|Country of ultimate parent company||Categorical variable: ISO code of the country of domicile of the ultimate parent company|
|Market of issuance||Categorical variable taking values Eurodollar, US144A, Global, Yankee, Samurai, Bulldog, Shogun, US domestic|
|Currency of denomination||Currency in which the debt security is issued|
|Ticker parent||Fundamental company ticker underlying each debt security. This ticker identifies the firm guaranteeing the debt security|
Variables are accessed using Bloomberg; ultimate sources are debt securities prospectus
Source: Own elaboration
Debt guaranteed by emerging economies firms in the period of 2000-2014. Break down by country (million dollars)
Sources: Bloomberg; own elaboration
Firms with access to global bond markets: switchers and non-switchers
|A: Firm-level information|
|Fixed assets to total assets||53.32||55.17|
|Interest coverage ratio||4.65||3.38|
|Local GAAP (%)||6.1||8.0|
|US ADR (%)||57.6||49.3|
|B: financing conditions|
|Sinking fund (%)||0||2.7|
|Convertible rights (%)||0||2.7|
|Call option (%)||18||30.7|
|Put option (%)||0||0.0|
|Bond holder rights (%)||0||16|
Firms with access to global bond markets are broken down in the groups: switchers are firms which also issue in any private debt securities markets; non-switchers are firms which only issue in the global bond market
Traditional placements in the US domestic market by non-US firms, known as Yankees, have become rare, to some extent due to the enforcement of securities regulation since the early 2000s.
These costs include all the expenses that borrowers need to pay to issue the debt security, and include underwriter compensation,legal fees, accountants’ fees, costs of fulfilling the regulatory requirements at the time of the issuance and ongoing basis.
Foreign bonds are issuances by non-resident firms in a given local market. These deals are referred to with nicknames related to the corresponding local market of issuance (Yankee for the USA, Samurai for Japan, etc.).
Our paper does not compare firms’ borrowing in all domestic and international markets, which is analyzed in Gozzi et al. (2015). Their findings underscore that international issuances are larger, of shorter maturity, denominated in foreign currency, include more fixed rate contracts and entail lower yields. They treat international bonds as a single asset class, while we investigate the differences across markets. Our paper does not analyze either the capital structure of public and private firms (Brav, 2009); but we examine if being listed in the USA through ADRs matters for firms’ choice.
Eurodollar bonds are Eurobonds denominated in US dollars.
For instance, a Mexican firm issuing a bond in the USA. In contrast, a firm domiciled in the USA issuing a debt security in the US, is carrying out a US domestic placement. The popular distinction between domestic and international debt markets reflects the relationship between the residence of debt issuer and the market of issuance location. Following this convention, domestic debt issuances are placements by resident issuers in their home-countries.
Registration is a process in which firms provide a description of the company, of the security offered for sale, the management of the company and the firm financial statements. The firm will need to file every year the form 20-F, in which are requested to provide standardized financial information. Other local legislations are deemed to share similar features.
Frequently they are not listed in the US, and firms need to provide additional information, or adapt their financial statements to the US GAAP. This contrasts with the much softer requirements of information disclosure of 144A private placements.
For some firms there is an additional factor, not discussed in this note: the desire to diversify funding costs.
Due to risk-shifting or asset substitution.
We follow previous research stressing that firms cross-listed in the US through ADRs already comply with regulatory requirements and accordingly are more likely to issue bonds in regulated markets (i.e. Yankee, global bonds) ; see Gao (2011), Miller and Puthenpurackal (2002); in contrast, being listed in a local exchange is likely unimportant.
At this point, we are interested in the sheer existence of a rating; we will control for the credit quality using different financial ratios-.
There is debate about the relative importance of each of these two factors, being the empirical evidence about the underlying cause mixed (see Dutordoir et al. (2014) for a review). Overall, there is support for the hypotheses that firms issue debt-securities with convertible rights to mitigate adverse selection; and less so, to mitigate moral hazard. Recent research suggests that the conversion features (date of conversion, callability of the bond, term) can make these debt-securities more akin to an equity-security or to a debt-security; the former are better suited to mitigate adverse selection, the latter to mitigate agency costs.
More specifically, we use the Altman score for private companies; this way we are able to compute it for firms which are not listed in equity markets. This score is equal to 0.717*x1 + 0.847*x2 + 3.107*x3 + 0.42*x4 + 0.998*x5; where x1 is working capital to total assets, x2 retained earnings to total assets, x3 is EBITDA to total assets, x4 is book value of equity to total debt, and x5 sales revenue to total assets.
Following standard credit risk techniques, we compare each firm with its peers. Hence, we construct industry-adjusted financial ratios, defined as deviations with respect to its industry median. These annual industry medians are obtained using the worldwide population of firms of each industry (see Appendix 1 for details).
MOVE (Merrill Lynch Option Volatility Estimate) is a yield curve weighted index of the implied volatility on 1-month treasury options.
Latin America includes Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Venezuela; emerging Europe includes Bulgaria, Belarus, Bosnia, Croatia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Serbia, Slovakia, Slovenia, Turkey and Ukraine; Africa and Middle East includes Egypt, Morocco, Nigeria, Saudi Arabia, South Africa and UAE. Emerging Asia includes India, Indonesia, Malaysia, Philippines, Thailand and South Korea.
Because of the small volume of foreign bonds and “private placements” that are not Rule 144A bonds, we do not include these bonds in our analysis.
There are two options: it can be an emerging economy firm, and be in our sample; or alternatively, it can be an advanced economy firm, and be excluded from it.
Standalone affiliates of advanced/emerging economies companies incorporated in an emerging/advanced economy are treated as emerging/advanced economies firms. Had we used the country of incorporation of the debt-issuer or the ultimate parent company, we would have tracked incorrectly emerging economies funding patterns. For example, Jaguar-Land Rover is a standalone affiliate incorporated in an advance economy (UK) with an emerging economy firm as its parent company (Tata Motors from India). Jaguar-Land Rover guarantees its own debt and its issuances are not included in our data base. On the contrary, Kansas City Southern de México is a standalone affiliate incorporated in Mexico with its parent being a US company. We then include issuances from Kansas City Southern de México in our data base.
However, we acknowledge there is probably not a single best criterion to track firms’ international activity. This decision depends on the purpose of the analysis: what it is useful for analyses on taxation or revenue diversification can be misleading for financial stability analysis, and the other way around. For instance, unconsolidated analyses might be useful for research on geographical diversification of income revenues, or the impact of taxation on firms’ organizational structure.
Overall, the generalized ordered model estimates 13 parameters; it is more parsimonious than the multinomial model (which estimates 21), and less so than the ordinal model (which has just 11, at the cost of imposing restrictive relations between them).
We cannot estimate the multinomial logistic model leads to quasi-perfect separation. This problem is less severe in the ordered logit –although the number of observations lost hovers around 60-, since the number of estimated parameters is smaller. Besides, we cannot include firm fixed effects, since the number of companies covered is very high and many only issue once. Results not reported for the sake of brevity.
Appendix 1. Data appendix
Description of variables
Table AI describes the firm-level variables. The information is obtained using Bloomberg; the ultimate sources are the financial statements filed by firms.
Table AII summarizes information about the deal-level variables. A key variable is the ticker parent: it is the fundamental company ticker associated to each security and identifies the company guaranteeing it. It need not be the issuer company or the ultimate parent company of the issuer. All firm-level variables listed in Table AI are obtained for this company.
Table AIII provides a country breakdown of debt issued in international markets in the period of 2000-2014 and guaranteed by emerging economy firms, thus complementing Figure 3 in the main text. Brazil and Mexico are the countries with the largest amount issued, followed by Russia and South Korea.
Allen, F., Brealey, R. and Myers, S. (2013), Principles of Corporate Finance, Mcgraw-Hill, New York, NY.
Altunbas, Y., Kara, A. and Marques-Ibanez, D. (2010), “Large debt financing: syndicated loans versus corporate bonds”, European Journal of Finance, Vol. 16 No. 5, pp. 437-445.
Arena, M. (2011), “The corporate choice between public debt, bank loans, traditional private debt placements, and 144A debt issues”, Review of Quantitative Finance and Accounting, Vol. 36 No. 3.
Avdjiev, S., Chui, M. and Shin, H. (2014), “Non-financial corporations from emerging market economies and capital flows”, BIS Quarterly Review, December 2014.
Avdjiev, S., McCauley, R. and Shin, H. (2015), “Breaking free of the triple coincidence in international finance”, BIS Working Paper No.524.
Ayala, D., Nedeljkovi, M. and Saborowski, C. (2015), “What slice of the pie? the corporate bond market boom in emerging economies”, IMF Working Paper No. 15/148.
Barrett, G. and Donald, S.G. (2003), “Consistent tests for stochastic dominance”, Econometrica, Vol. 71 No. 1, pp. 71-104.
Berlin, M. and Loyes, J. (1988), “Bond covenants and delegated monitoring”, Journal of Finance, Vol. 43 No. 2, pp. 397-412.
BIS-ECB-IMF (2015), Handbook of Securities Statistics, IMF, Washington, DC.
Blackwell, D.W. and Kidwell, D.S. (1988), “An invertigation of cost differences between public sales and private placements of debt”, Journal of Financial Economics, Vol. 22 No. 2, pp. 253-278.
Brav, O. (2009), “Access to capital, capital structure, and the funding of the firm”, The Journal of Finance, Vol. 64 No. 1, pp. 263-308.
Brennan, M. and Swartz, E. (1988), “The case for convertibles”, Journal of Applied Corporated Finance, Vol. 1 No. 2, pp. 55-64.
Chaplinsky, S. and Ramchand, L. (2004), “The impact of SEC Rule 144A on corporate debt issuance by international firms”, The Journal of Business, Vol. 77 No. 4.
Chemmanur, T. and Fulghieri, P. (1994), “Reputation, renegotiation, and the choice between bank loans and publicly traded debt”, Review of Financial Studies, Vol. 7 No. 3, pp. 475-506.
Denis, D. and Mihov, V. (2003), “The choice among bank-debt, non-bank private debt, and public debt: evidence from new corporate borrowing”, Journal of Financial Economics, Vol. 70 No. 1, pp. 3-28.
Dutordoir, M., Lewis, C., Seward, J. and Veld, C. (2014), “What we do and do not know about convertible bond financing”, Journal of Corporate Finance, Vol. 24, pp. 3-20.
Esho, N., Lam, Y. and Sharpe, G. (2001), “Choice of financing source in international debt markets”, Journal of Financial Intermediation, Vol. 10 Nos 3/4, pp. 276-305.
Fenn, G.W. (2000), “Speed of issuance and the adequacy of disclosure in the 144A high-yield debt market”, Journal of Financial Economics, Vol. 56 No. 3, pp. 383-405.
Feyen, E., Ghosh, S., Kibuuka, K. and Farazi, S. (2015), “Global liquidity and external bond issuance in emerging markets and developing economies”, Worldbank Policy Research Working Paper 7363.
Fuertes, A. and Serena, J.M. (2014), “Firms financial soundness and access to capital markets”, Financial Stability Review Banco de España n° 27.
Gao, Y. (2011), “The Sarbanes-Oxley act and the choice of bond market by foreign firms”, Journal of Accounting Research, Vol. 49 No. 4.
Gomes, A. and Phillips, G. (2012), “Why do public firms issue private and public securities?”, Journal of Financial Intermediation, Vol. 21 No. 4.
Gozzi, J.C., Levine, R., Martinez Peria, M.S. and Schmukler, S.L. (2015), “How firms use corporate bond markets under financial globalization”, Journal of Banking & Finance, Vol. 58, pp. 532-551.
Green, R. (1984), “Investment incentives, debt, and warrants”, Journal of Financial Economics, Vol. 13 No. 1, pp. 115-136.
Green, R. and Talmor, E. (1986), “Asset substitution and the agency costs of debt financing”, Journal of Banking and Finance, Vol. 10 No. 3, pp. 391-399.
IOSCO (2010), “Objectives and principles of securities regulation”, Public Report of IOSCO.
Krishnaswami, S. and Yaman, D. (2008), “The role of convertible bonds in alleviating contracting costs”, The Quarterly Review of Economics and Finance, Vol. 48 No. 4, pp. 792-816.
Krishnaswami, S., Sprindt, P.A. and Subramanian, V. (1999), “Information asymmetry, monitoring and the placement structure of corporate debt”, Journal of Financial Economics, Vol. 51 No. 3, pp. 407-434.
Kwan, S. and Carleton, W. (2010), “Financial contracting and the choice between private placements and publicly offered bonds”, Journal of Money, Credit and Finance, Vol. 42 No. 5, pp. 907-929.
Lewis, C., Rogalski, R. and Seward, J. (1998), “Understanding the design of convertible debt”, Journal of Applied Corporate Finance, Vol. 11 No. 1, pp. 45-53.
Lo Duca, M., Nicoletti, G. and Vidal Martinez, A. (2014), “Global bond corporate issuance: what role for US quantitative easing?”, ECB Working Paper N1 1649 March 2014.
Miller, D. and Puthenpurackal, J.J. (2002), “The costs, wealth effects, and determinants of international capital raising: evidence from public yankee bonds”, Journal of Financial Intermediation, Vol. 11 No. 4.
Myers, S. (1977), “Determinants of corporate borrowing”, Journal of Financial Economics, Vol. 5 No. 2, pp. 147-175.
Myers, S. and Majluf, N. (1984), “Corporate financing and investment decisions when firms have information that investors do not have”, Journal of Financial Economics, Vol. 13 No. 2, pp. 187-221.
Resnick, B. (2012), “Investor yield and gross underwriting spread comparisons among US dollar domestic, yankee, eurobond, and global bonds”, Journal of International Money and Finance, Vol. 31 No. 2.
Fuertes, A. and Serena, J.M. (2016), “How firms borrow in international bond markets: securities regulation and market segmentation”, Banco de España Working Paper 1603.
Pagan, A. and Ullah, A. (1999), Nonparametric Econometrics, Cambridge University Press, Cambridge.