DISCUSSION PAPER MFM Global Practice No. 17 February 2017 Tobias Haque Jane Bogoev Greg Smith MFM DISCUSSION PAPER NO. 17 Abstract Over the past decade, a large number of low- and lower-middle income ‘frontier economies’ have begun to access international private capital markets to meet fiscal financing needs. In this paper we seek to identify drivers of this trend, identify associated risks, and present policy implications for frontier-market policy-makers. Through simple analysis of the characteristics of recent frontier market issuers, we show that smaller, poorer, and less well-governed economies are now accessing global credit markets. Through cross-country regression analysis, however, we demonstrate that the capacity of these countries to issue debt (and the cost of this debt) continues to be influenced by their macroeconomic performance and quality of governance. Drawing on evidence from Ghana and Zambia, we illustrate potential risks arising from recent expansions of access to global debt markets, where rapid debt accumulation of foreign-denominated debt in the context of lessened market discipline and following recent debt relief is now posing pronounced debt sustainability and refinancing risks. We conclude that increased access to international debt markets presents both opportunities and risks to frontier issuers. The new cohort of frontier issuing economies should: i) take careful account of debt risks and debt sustainability considerations when developing fiscal policy and debt strategies; ii) work to reduce the costs of ongoing external borrowing through adopting sound economic policies and protecting credit ratings; and iii) develop domestic debt markets as a potential alternative source of fiscal financing through which to reduce reliance on foreign- denominated Eurobond debt with its associated refinancing and currency risks. Corresponding author: thaque2@worldbank.org JEL Classification: G15, F65, H63, F34 Keywords: debt, Eurobonds, Africa, debt sustainability, macroeconomics, finance, frontier markets. ii This series is produced by the Macroeconomics and Fiscal Management (MFM) Global Practice of the World Bank. The papers in this series aim to provide a vehicle for publishing preliminary results on MFM topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For information regarding the MFM Discussion Paper Series, please contact the Editor, Ivailo Izvorski at iizvorski@worldbank.org. © 2017 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. iii Push and Pull: Emerging Risks in Frontier Economy Access to International Capital Markets Tobias Haque, Jane Bogoev, Greg Smith1 1. Introduction The number and range of ‘frontier’ economies issuing sovereign bonds and syndicated loans has increased over recent years.2 As East Asian and Latin American countries have graduated from frontier market status, bond issuance by Sub-Saharan African countries has accelerated rapidly. Sub-Saharan Africa now accounts for the majority of frontier market issuances, while the nominal volume of frontier market issuances and number of issuing countries continues to increase. What accounts for the changing global landscape for frontier market bond issuance? Does increased access for countries previously excluded from global capital markets reflect improvements in those countries’ economic fundamentals and capacity to manage debt-related risks? Or has the “search for yield” in post-crisis global capital markets driven increased risk appetite? What are the implications for the current cohort of frontier market economies in terms of managing macroeconomic risks associated with accessing private capital markets and their likely future capacity to meet fiscal financing needs through tapping global capital markets? In this paper we attempt to answer these questions. Based on recent trends in bond and syndicated loan issuance by frontier economies, we: i) examine changes in the country characteristics of frontier markets over the past fifteen years, showing how the “search for yield” has led to increased risk tolerance for investors with smaller, slower growing, and poorly-governed countries now able to access private capital markets to meet fiscal needs; ii) provide empirical analysis of the determinants of bond issuance and pricing among these recently issuing frontier economies, demonstrating that – while there has been an expansion of the range of countries able to access private capital markets – similar factors as those identified in the earlier and broader international literature continue to determine access and pricing; and iii) use case study evidence from two countries to illustrate how recently less-constrained access to private capital markets has led to heightened exchange rate, refinancing, and debt sustainability risks. Our analysis builds on a substantial existing literature examining the determinants of bond issuance and pricing. From these findings we distil two main policy messages. Firstly, frontier markets can no longer rely on ‘market discipline’ to ration access to private capital markets and therefore must ensure that adequate ‘policy discipline’ is applied. Countries with weaker fundamentals and governance can access international debt markets and must ensure that policies are adequate to prevent inappropriate or excessive borrowing including through ensuring sound debt management practices and carefully considering debt sustainability implications of fiscal policy decisions. Secondly, frontier economies stand the best chance of maintaining access to international capital markets if they are able to maintain strong growth performance and a strong sovereign risk rating. Policy reforms should be pursued with these objectives in mind. 1 All World Bank. Valuable research assistance was provided by Ying Li. The paper has benefited from comments from and conversations with Alex Sienaert, Jukka Strand, Mark Thomas, Ivailo Izvorski, and Ivana Ticha. All errors remain the responsibility of the authors. 2 For the purposes of this paper ‘frontier’ economies are those classified as low- or lower-middle income under the World Bank income group classification system. 1 In the second section of this paper we summarize recent literature regarding the factors determining sovereign access to private capital markets. In the third section, we discuss recent developments in frontier economy bond issuance. In the fourth section we present regression results regarding the factors influence bond issuance, coupon rates, and secondary market pricing for frontier economies. In the fifth section we present case study evidence regarding the risks of increased private capital market access from two Sub-Saharan African economies and lessons for mitigation of such risks. In the final section we present conclusions and policy recommendations. 2. Literature There is a very substantial existing literature regarding the determinants of global capital flows. An additional literature examines the determinants of private capital access for frontier markets. A more recent literature examines the specific very recent experience of frontier markets, especially in Sub-Saharan Africa, in tapping international credit markets through the issuance of global bonds. We briefly summarize these literatures in turn, and – based on this discussion - outline the novel contribution of this research. A large theoretical literature examines the factors constraining and enabling international capital flows. In a seminal paper, Lucas (1990) famously noted that, given relative scarcity and associated expectations of high returns, too little capital flows from rich countries to poor countries. Feldstein and Horioka (1980) are credited with the corollary empirical observation that savings increases with investment in most countries, with investors tending to allocate capital within national boundaries rather than optimizing investment based on global opportunities. Recent work has presented various theoretical explanations for the failure of neo-classical models predicting the flow of capital to developing countries. Kalemi-Ozcan et al. (2008), for example, noted that frictions at national borders may account for the divergence between theoretical predictions and empirical observation, showing that capital flows between richer US states to poorer states, conforming closely to the standard model. Verdier (2008) shows that, in presence of an international borrowing constraint and complementarity between domestic and foreign capital in production, foreign debt rises with domestic savings, a prediction consistent with data on capital flows. A large empirical literature examines the determinants of global capital flows and yields. Typically, regression analysis is employed using large cross-country datasets to determine the relative importance of ‘push’ and ‘pull’ factors in variables impacting the volume and pricing of international debt flows. Push factors refer to global economic conditions, including global risk and interest rates. Pull factors are country-specific, typically including – among others – growth rates, debt levels, reserve adequacy, and institutional performance. The conclusions of such work are often contradictory, reflecting in part the difficulty of establishing causal links, and clear messages for policy-makers are sometimes elusive. Studies reach different conclusions regarding the relative impact of global and country-specific factors. Results also vary regarding the importance of key macroeconomic variables, including levels of debt and reserve assets. Moreover, recent studies have suggested that the impact of specific factors may differ substantially over different periods of the credit cycle (Frances, Aykut, and Tereanu 2014; Fratzcher 2011). Table 1 provides a summary of recent literature. 2 Table 1: Summary of literature on push and pull factors Emerging and developing countries financing Dependent Probability of issuance Credit spread variable Type Driver Strong evidence of negative Weak evidence of Evidence of positive Evidence of No identified relationship (negative) relationship relationship negative or weak relationship relationship Push Global market Feyen, E., Ghosh, S., Kibuuka, K. Presbitero A. F., Ghura Presbitero A. F., Ghura D., Siklos, P., (2011). risk aversion and Farazi, S. (2015). D., Adedeji O. S., and Adedeji O. S., and Njie, L. World Bank (2014). Njie, L. (2015). (2015). IMF and WB (2015). Feyen, E., Ghosh, S., Kibuuka, K. and Farazi, S. (2015). Kennedy, M., and Palerm, A., (2014). Guscina, Pedras, and Presciuttini (2014) Mature economy Presbitero A. F., Ghura D., Feyen, E., Ghosh, S., Kibuuka, Presbitero A. F., Kamin, S.B., and interest rate (and Adedeji O. S., and Njie, L. (2015). K. and Farazi, S. (2015). Ghura D., Adedeji von Kleist, K., measures of Feyen, E., Ghosh, S., Kibuuka, K. Miklos, P., (2011). (Latin and O. S., and Njie, L. (1999). tightening of and Farazi, S. (2015). Asia region) (2015). liquidity) – proxy Eichengreen, B., and Mody, A., Min, H., Lee, D., Park, M., Eichengreen, B., of opportunity (2000). Nam, S. (2003). and Mody, A., cost and liquidity World Bank (2014). (2000). IMF and WB (2015). Money supply World Bank (2014) for bonds World Bank (2014) for (M2) capital flow Onset of crisis Fratzscher M. (2011). Siklos, P., (2011). (with Asia an events exception) Drivers Strong evidence of positive Weak or mixed Strong evidence of negative Weak evidence of negative relationship evidence of positive relationship relationship relationship Pull Macroeconomic Presbitero A. F., Ghura D., 1. Feyen, E., Ghosh, S., 3. Presbitero A. F., Ghura D., Feyen, E., Ghosh, S., Kibuuka, K. and fundamentals Adedeji O. S., and Njie, L. (2015). Kibuuka, K. and Farazi, Adedeji O. S., and Njie, L. Farazi, S. (2015). (fiscal status, Gelos, G.R., Sahay, R., and S. (2015). (2015). debt level, Sandleris, G., (2011). 2. IMF and WB (2015). 4. Eichengreen, B., and Mody, current account Fratzscher M. (2011). A., (2000). balance, growth, World Bank (2014). 5. Kennedy, M., and Palerm, A., foreign reserve) Feyen, E., Ghosh, S., Kibuuka, K. (2014). and Farazi, S. (2015). 3 Emerging and developing countries financing Dependent Probability of issuance Credit spread variable Min, H., Lee, D., Park, M., Nam, 6. Guscina, Pedras, and S. (2003). Presciuttini (2014) 7. Min, H., Lee, D., Park, M., Nam, S. (2003). Institutions Presbitero A. F., Ghura D., Kennedy, M., and Palerm, A., Siklos, P., (2011). Adedeji O. S., and Njie, L. (2015). (2014). Faria, A., Mauro, P., and Zaklan, A., Faria, A., Mauro, P., and Zaklan, 1. Guscina, Pedras, and (2011). A., (2011). Presciuttini (2014) Gelos, G.R., Sahay, R., and Sandleris, G., (2011). World Bank (2014) Fratzscher M. (2011). Size of economy Presbitero A. F., Ghura D., Feyen, E., Ghosh, S., Kibuuka, K. and or income level Adedeji O. S., and Njie, L. (2015). Farazi, S. (2015). Feyen, E., Ghosh, S., Kibuuka, K. and Farazi, S. (2015). Faria, A., Mauro, P., and Zaklan, A., (2011). Gelos, G.R., Sahay, R., and Sandleris, G., (2011). IMF and WB (2015). Openness Faria, A., Mauro, P., and Faria, A., Mauro, P., and Zaklan, A., (2011). Zaklan, A., (2011). Gelos, G.R., Sahay, R., and Sandleris, G., (2011). Fratzscher M. (2011). David A. Grigorian, (2003). 4 A few researchers have addressed the determinants of market access specifically for frontier markets during particular previous periods. Grigorian (2003), for example, focusses on debut sovereign bond issuances. He identifies a range of country characteristics that impact on the likelihood of successful bond issuance, including size of the economy, the current account position, reserve adequacy, and inflation. He finds that the fiscal deficit is positively correlated with successful issuance, with the financing need effect outweighing any associated concerns regarding creditworthiness. Debt levels, trade openness, and the existence of a Fund program have no impact on the likelihood of a successful debut issuance. More recently, Das, Papioannou, and Polan (2008) present an analysis of 21 emerging and low-income debut issuances during the 1996-2008 period. They find that issuance size is determined by market conventions rather than macroeconomic fundamentals. They further find that the pricing of debut bonds (spread and coupon) is (unsurprisingly) strongly correlated with credit rating, with the adequacy of reserves playing no significant role. A recent literature addresses bond issuances by frontier market economies over the past decade, with a particular focus on the experience of Sub-Saharan Africa. Guscina, Pedras, and Pesciuttini (2014) provide an excellent analysis of 23 debut bond issuances since 2004. They identify low advanced economy interest rates and the ‘search for yield’, combined with investor appetite for diversification as important push factors. Strong economic fundamentals, often associated with high commodity prices, recent debt relief through HIPC, and limited inflation, are cited as important ‘pull’ factors. Through a formal analysis, the authors identify economic growth, quality of institutions, debt levels, the federal funds rate, and global volatility as exerting a significant impact on debut spreads. A recent joint World Bank/IMF Board paper includes formal analysis of the determinants of bond issuance, concluding that domestic factors are ‘most powerful’ (IMF/World Bank 2014). Several recent papers (Mecagni et al. 2014; Hou et al. 2014; Sy 2015; Tyson 2015; Tyson 2015a) provide narrative descriptions of the increase in global bond issuances by frontier markets and assesses the advantages, disadvantages and risks associated with such financing instruments. These authors typically ascribe roughly equal importance to global and country-specific factors in driving market access or do not attempt to determine their relative importance. Tyson (2015) highlights lack of conditionality as an important factor increasing the attractiveness of global bonds to frontier economies relative to more traditional sources of concessional financing. This literature emphasizes risks arising from increased reliance on global bond financing when countries face exchange rate volatility, difficulties in refinancing bullet-structure instruments, and challenges in ensuring the quality of investment of bond proceeds. This paper contributes to the existing literature by: i) providing the first quantitative analysis of recent changes in the characteristics of frontier market issuers and drawing conclusive findings regarding the relative importance of push and pull factors for frontier issuers; ii) providing an up-to-date analysis to date of the determinants of market access and pricing exclusively for frontier economies; and iii) providing specific policy messages based on the recent experience of frontier market issuers. 3. Characteristics of bond-issuing frontier economies In this section we describe changes in the characteristics of frontier market issuers over the past 15 years and assess the extent to which such changes have been driven by push or pull factors. We find that the broader range of frontier markets now accessing global capital markets reflects increased risk appetite by investors, rather than improved macroeconomic and governance performance of new issuers. 5 There have been substantial changes in the frontier countries issuing sovereign bonds over recent years. A larger number of frontier market countries are now issuing bonds. Sub-Saharan African countries are increasingly dominating frontier market issuance, both in terms of volumes and number of issuances. In 2000, Latin America accounted for three of five issuances and 70 percent of frontier market issuances by volume. As Latin American and East Asian countries have graduated from frontier market status, they have accounted for a steadily declining number and volume of issuances. The first Sub-Saharan frontier market bond issuance occurred in 2007. In 2015, Sub-Saharan Africa accounted for seven of nine frontier market issuances and around 80 percent of the total volume issued. Figure 1: Bond issuances by frontier economies by region (number) 12 10 8 6 4 2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 ECA LAC EAP SAR MENA SSA Source: Bloomberg Figure 2: Bond issuances by frontier economies by region (by volume – Bn US$) 14 US$ Bn 12 10 8 6 4 2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Source: Bloomberg What has driven the changing composition of countries issuing sovereign bonds? One possible explanation is that as Latin American and East Asian countries have graduated from frontier status a new cohort of countries, including those in Sub-Saharan Africa, have recently achieved a sufficient level of macroeconomic and governance performance to attract private sector lending. If the changing composition of issuing frontier economies is driven by “pull factors”, we would expect to see similar performance against key macroeconomic and institutional indicators of issuing frontier markets over time. New issuers would have achieved market access by meeting the performance level required by private markets. We find little evidence of this. 6 Firstly, it is clear that the average size of frontier issuing economies has declined over time. This has partly been driven by bond issuances by several very large frontier economies only over the 2000-2007 period (China, Brazil, and Indonesia, with China and Brazil both graduating from frontier status over the period). However, even with these outliers excluded, the average size of issuing frontier economies has declined significantly, with the proportion of frontier issuances by countries with economies of less than US$40 billion increasing from 28 percent over 2000-2007 to 61 percent over 2008-2015. Figure 3: GDP of issuing frontier markets (USD Billion) (Line is average for all charts) 200 180 160 Billions US$ 140 120 100 80 60 40 20 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: WDI, Bloomberg Frontier economies are now also issuing bonds at lower levels of income, as measured by GDP per capita. As shown in Figure 4, over 2000 to 2014, the average per capita income of issuing frontier market economies declined fairly steadily from US$2,800 to US$1,300 (before spiking in 2015 to US$2,600). The proportion of frontier issuances by countries with per capita incomes of less than US$2,500 increased from around 40 percent over 2000-2007 to nearly 90 percent over 2008-2015. Investors, however, may be concerned about economic growth, rather than only the income level. As shown in Figure 5, however, there also appears to have been a reduction in growth rates of issuing economies (over the five years preceding issuance), with the average issuing country having experienced a five-year annual average growth rate of greater than 6 percent in around 40 percent of cases between 2000 and 2007, but only 30 percent of cases over 2008-2015. Figure 4: GDP per capita of issuing frontier markets 5000 4500 4000 (Constant 2005 USD) 3500 GDP Per Capita 3000 2500 2000 1500 1000 500 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: WDI, Bloomberg 7 Figure 5: 5-year average GDP growth of issuing frontier markets 14 5-year Average GDP Growth 12 10 8 6 4 2 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: WDI, Bloomberg It is also clear that frontier economies with poorer governance are now issuing bonds. The average overall CPIA score for issuing frontier economies declined from 3.8 to 3.6 over the period 2000-2015.3 The proportion of issuances from countries with CPIA scores of less than 3.5 increased from around 12 percent over the period 2000-2007 to around 33 percent over the period 2007-2015. Analysis of World Governance Indicator scores shows a similar decline in average quality of governance among issuing frontier economies over the period. Figure 6: Overall CPIA score of issuing frontier markets 5 4.5 4 3.5 3 2.5 2 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: CPIAs, Bloomberg In contrast to the trends described above, the average level of indebtedness of issuing frontier economies has declined slightly since 2000, but recently begun to increase. As shown in Figure 7, the average public debt-to-GDP ratio of issuing frontier economies declined steadily from around 58 percent in 2000 to around 52 percent in 2007, before increasing steadily again to 56 percent by 2015. Low average debt levels over the period when Sub-Saharan African countries began to issue sovereign bonds in substantial volumes reflects the impacts of HIPC and MDRI debt relief, with Sub-Saharan African countries taking advantage of newly reduced debt burdens to access private capital markets. As Sub-Saharan African countries and other recipients of debt relief have begun to re-accumulate debt, debt-to-GDP ratios for issuing 3 The Country Policy and Institutional Assessment is a diagnostic tool that annually assesses the quality of policies and the performance of institutional frameworks in IDA countries. It measures the capacity of a country to support sustainable growth, poverty reduction, and the effective use of development assistance. 8 countries have again increased since 2008, and some counties with very high levels of public debt (exceeding 80 percent of GDP) continue to issue sovereign bonds. Figure 7: Public debt to GDP ratio of issuing frontier markets 160 140 120 100 80 60 40 20 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: WEO, Bloomberg Given that smaller frontier markets with lower incomes, lower rates of economic growth, and poorer governance are now issuing bonds, we might expect to see an increase in risk premiums and therefore higher coupon rates and spreads. In other words, the new cohort of frontier economies may have achieved market access despite higher risks by offering appropriate compensation for those risks. As shown in Figure 8 and Figure 9, there is clear evidence for this. The average coupon rate declined steadily between 2000 and 2012, except for a peak during the immediate post-crisis period, and has begun to move up again over the past four years. There is, however, a clear upward trend in the spread between sovereign bond coupon rates and the Federal Funds Rate (FFR), which increased significantly post-crisis. The average spread declined consistently between 2002 and 2007, but exceeded 2002 levels from 2010. Figure 8: Coupon rate for frontier market bonds 14 12 10 8 6 4 2 0 2000 2002 2004 2006 2008 2010 2012 2014 Source: Bloomberg 9 Figure 9: Coupon spread on federal reserve rate for frontier market bonds (%) 14 12 10 8 6 4 2 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 -2 -4 Source: WDI, Bloomberg Overall, this analysis strongly suggests that changes in the composition of frontier issuing economies over recent years have been driven, on average, more by push than pull factors. New issuers have not improved to meet a threshold level of macroeconomic and institutional performance required necessary for issuance by private capital markets. Rather, the standards for investment applied by private capital markets appear to have declined. Investors, seeking a given return on investment (relative to the zero-risk federal funds rate) have moved into riskier investments. 4. Determinants of frontier market issuance and pricing The analysis above might be taken to suggest that international capital markets are increasingly unconcerned about market and institutional fundamentals of issuing countries, responding instead to global push factors, including low global interest rates. In this section we present formal analysis to further examine the relative importance of push and pull factors in determining market issuance and pricing of frontier economies over the past ten years. We find that the pull factors such as macroeconomic performance and quality of governance continue to exert an important impact on issuance and pricing for the new cohort of frontier- market issuers. Model specification The quantitative analysis is based on two different regression models and estimators. The first regression model builds on the analysis above and estimates the propensity of issuance of a Eurobond or a syndicated loan determined by the fundamental push and pull factors. The sample includes 110 countries classified as low- and lower-middle income in 2005.4 Push and ‘pull’ factors included in the analysis are those commonly cited in the existing literature, including the federal funds rate, the VIX risk index, growth rates and various other macroeconomic variables, and selected indicators of institutional quality. The full set of variables included in the analysis is summarized in the table below (summary statistics are provided in Annex 1). The model is unbalanced panel that incorporates the cross country and 4 More details about the World Bank Atlas methodology and country classifications can be obtained from: https://datahelpdesk.worldbank.org/knowledgebase/articles/378833-how-are-the-income-group-thresholds-determined 10 time variation of the data set as a logistic function. Thus, a pooled unbalanced logit model is employed, presented as follows: Issuancejt = α2 + β3X't + β4w'jt + u2jt (1) where: Issuance is a binary dependent variable having values of 1 in the years when the Eurobond or syndicated loan was taken and 0 otherwise, α2 is a constant, X is a vector of exogenous push factors, w is a vector of domestic pull factors, β3 and β4 are parameters to be estimated, u2 are the residuals and j and t are a country specific and time specific subscripts, respectively. The second regression model estimates the relationship between the coupon rates of the issued Eurobonds and the major push and pull factors. It employs a simple cross sectional ordinary least squares (OLS) estimator with robust standard errors presented with the following formula: Coupon_ratei = α1 + β2X'j + β3w'j + u1j (2) where: coupon_rate is the rate by which the Eurobond was issued, α is a constant, X is a vector of exogenous push factors, w is a vector of domestic pull factors, β2 and β3 are parameters to be estimated, u1 is the white noise error term and j is a country specific subscript. The models are specified according to variables identified as significant in the existing empirical literature. We employ an estimation strategy moving from general to specific model estimations. In the preliminary search for the optimal specification, all variables that were identified in the broad survey of the literature (Section 2), were provisionally considered for inclusion in the regressions (data description is provided in Table 2). 11 Table 2: Variables and data sources Dependent variables Variable: Source: Units: Bonds coupon rate Bloomberg Percent Secondary bond pricing Bloomberg Index Eurobond and syndicated Bloomberg Dummy variable: value of 1 in the loan issuance dummy year when the bond/loan was issued and 0 otherwise Global push factors Variable: Source: Units: Federal Funds Rate (FFR) Bloomberg Percent VIX index Bloomberg Index Local pull factors GDP growth WDI Percentage change GDP per capita WDI Constant US$ GDP per capita growth WDI Percentage change Total gross debt IMF-WEO Percent of GDP Total external debt WDI Percent of GDP PPG external debt WDI Percent of GDP Debt services of total WDI Percent of exports external debt Debt services of PPG WDI Percent of exports external debt Current account balance IMF-WEO Percent of GDP Budget balance IMF-WEO Percent of GDP FDIs WDI Percent of GDP Trade openness WDI Percent of GDP Foreign reserves WDI Percent of GDP Inflation WDI Percent change Net official development WDI Percent of GDP/GNI assistance received Real exchange rate WDI Index IMF dummy IMF website Dummy variable: 1 if the country has IMF arrangement, 0 otherwise Commodity exporter UNCTAID Dummy variable: 1 if the country is a commodity exporter, 0 otherwise Sovereign risk rating Fitch Government WB governance effectiveness indicators Political stability WB governance indicators Regulatory quality WB governance indicators Rule of law WB governance indicators CPIA WDI 12 Data quality and availability have imposed some constraints on modeling. Data with large gaps or missing recent observations were omitted from the modeling. Variables discarded due to data quality and coverage considerations included:  Foreign reserves (no data available for many SSA countries);  Real exchange rate (no data available for majority of the countries in the sample);  Debt service of total external and external PPG debt (data not available for many countries in the sample, especially from SSA region);  Trade openness that is calculated as a sum of exports plus imports over GDP (the data for exports and imports start from 2005 and data for many countries is missing or is available only until 2012);  Fitch sovereign rating (no data available for majority of the countries that have not yet accessed international markets); and  CPIA as an alternative indicator for institutional quality (data starts from 2005 and no data is available for the upper middle income countries as CPIA assessments are only carried out for LICs). Starting with the most general model that includes most of the variables commonly included in the empirical literature, the specification was refined based on progressively dropping variables with insignificant coefficients. 13 Regression results for issuance Main results for determinants of issuance, based on model 1 above, are presented in Table 3. Table 3: Regression results for estimating the determinants of propensity to issue Eurobonds/syndicated loans Dependent variable: Eurobond and Regression 1 Regression 2 Regression 3 Regression 4 Regression 5 Regression 6 Regression 7 syndicated loan issuance Regressors: log GDP per capita 0.0543 (0.378) debt_weo_diff 0.00802 0.0104 (0.0248) (0.0291) FDI_GDP -0.0744 -0.0857 -0.0326 (0.0532) (0.0556) (0.0422) IMF dummy -0.369 -0.374 -0.134 -0.109 (0.301) (0.302) (0.280) (0.282) Inflation 0.0761* 0.0457** 0.00617* 0.00593* 0.00762** 0.00737** 0.00248 (0.0419) (0.0205) (0.00333) (0.00319) (0.00319) (0.00313) (0.00294) Commodity exp. dum. 0.601 0.687* 0.555 0.461 0.776** 0.781** 0.694* (0.392) (0.393) (0.374) (0.370) (0.353) (0.352) (0.374) GDP growth 0.0276 0.0293 0.0368 0.0319 0.0459** 0.0457** 0.0397* (0.0377) (0.0354) (0.0248) (0.0241) (0.0220) (0.0219) (0.0223) odanet_gni -0.295*** -0.289*** -0.294*** -0.290*** -0.307*** -0.308*** -0.373*** (0.0746) (0.0605) (0.0588) (0.0590) (0.0596) (0.0594) (0.0613) budget_bal_gdp -0.0116 -0.0176 -0.0493 -0.0544 -0.0907*** -0.0925*** -0.0698*** (0.0474) (0.0477) (0.0416) (0.0413) (0.0337) (0.0334) (0.0267) CAB_GDP 0.00664 0.00718 0.0219 0.0328 0.0378** 0.0373** 0.00979 (0.0257) (0.0256) (0.0233) (0.0209) (0.0174) (0.0173) (0.0141) Regulatory quality 1.778*** 1.802*** 1.662*** 1.624*** 2.100*** 2.085*** (0.502) (0.460) (0.409) (0.401) (0.353) (0.348) Rule of law 1.270*** (0.283) lvix -1.608*** -1.536*** -1.311*** -1.314*** -1.296*** -1.261*** -1.122*** (0.564) (0.532) (0.452) (0.453) (0.443) (0.417) (0.388) ffr -0.0927 -0.0698 -0.0426 -0.0437 -0.0183 (0.106) (0.0959) (0.0708) (0.0712) (0.0685) Constant 4.416 4.752*** 3.974*** 3.904*** 3.482** 3.334*** 3.127*** (2.942) (1.731) (1.447) (1.447) (1.393) (1.249) (1.154) Observations 567 568 656 656 886 886 886 Number of countries: 61 61 68 68 77 77 77 Pseudo R squared 0.32 0.31 0.31 0.31 0.33 0.33 0.29 Joint F test for testing the statistical signifcance of the excluded 5 variables from Regression 1 to Regression 6: chi2 = 5.75; p-value=0.33 Robust standard errors in parentheses. Logistic panel data regression. *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations. The most general specification (Regression 1, Table 3) includes most of the important variables identified in the empirical literature. This model includes a battery of pull factors such as: log of GDP per capita, change in the stock of debt to GDP ratio, net FDI inflows, inflation, IMF and commodity export dummy variables, GDP growth, inflow of net official development assistance, budget balance, current account balance and some of the factors that measure the institutional quality in the countries (regulatory quality or rule of law). In addition, two push factors are included: federal funds rate and the VIX index measuring the volatility of the international financial markets. 14 The results in Regression 1 indicate that GDP per capita is insignificant (p-value of 0.9) and part of its impact is captured by GDP growth. Therefore, we exclude it from the model and proceed with more parsimonious specification (Regression 2, Table 3). In Regression 2, the first difference of debt to GDP ratio remains insignificant (p-value of above 0.7) and is correlated with the budget balance. In the more restricted model (Regression 3), the FDI variable remains insignificant. The IMF dummy variable and the FFR remain insignificant in Regressions 4 and 5, respectively, and are therefore omitted from the most parsimonious model presented with Regression 6. The joint statistical significance of all the excluded variables (starting from Regression 1 to Regression 6), was tested and their joint insignificance confirmed, supporting their exclusion from the model. Assessing the goodness of fit with the Pseudo R squared coefficient, we observe no deterioration in goodness of fit from exclusion of variables (0.32 Regression 1 to 0.33 Regression 5). The results from the most parsimonious model presented in Regression 6 are based on a large number of observations (i.e. 886 over 76 LICs). The model was specified with only one indicator of institutional quality to avoid problems of multicollinearity with the other institutional indicators given their strong correlation. Results are generally intuitive and consistent with theoretical predictions, although with some exceptions. In terms of push factors, there is a robust statistical evidence based on 7 different regressions (Table 4) in which only the VIX index has a statistically significant impact on countries’ propensity to raise financial resources from the international financial markets. When the VIX index increases, there is higher expected volatility on the financial markets which appears to increase risk aversion among investors, leading to lower issuance. However, contrary to most theoretical predictions, the FFR rate has no statistically significant impact on frontier markets’ propensity to issue. In terms of ‘pull’ factors, our results show robust statistical evidence for the impact of the official development assistance and quality of the institutions. Countries with higher levels of official development assistance are less likely to access the international financial markets, potentially reflecting either reduced financing needs in the context of large aid inflows, or higher risk perceptions in the context of high aid dependence. Institutional quality is also an important determinant for counties’ access to international financial markets, with this result robust across different indicators of institutional quality: regulatory quality (Regression 6) and rule of law (Regression 7). Less robust statistical evidence is found for the impact of GDP growth, the budget deficit, the current account balance, inflation and the commodity exporting dummy variable because the significance of these variables depends on the model specification. In that sense, most of them have become significant only after the model was restricted by dropping the other redundant variables. This may imply that part of the impact of this variables has been captured by the redundant variables that were causing multicollinearity explained previously. The size of the budget balance is negatively correlated with issuance, reflecting increased likelihood of accessing international capital markets when financing needs are higher. In general, however, stronger fundamentals are associated with greater likelihood of issuance. Countries are more likely to issue when experiencing a lower current account deficit, stronger GDP growth, or if they are commodity exporters. 15 Regression results for coupon rates Regression results regarding the determinants of bonds’ coupon rates, based on model 2 (above), are summarized in Table 4. Table 4: Regression results for the determinants of coupon rates Dependent variable: Regression 1 Regression 2 Regression 3 Regression 4 Regression 5 Regression 6 Regression 7 Regression 8 Coupon rate Regressors: Rule of law 0.167 (1.735) debt_weo 0.00513 0.000677 (0.0256) (0.0214) CAB_GDP -0.0112 (0.104) FDI_GDP 0.138 0.178 0.00904 (0.191) (0.178) (0.0488) Commodity exp. dum. -0.963 -1.236 -0.826 -0.788 (1.176) (0.899) (0.906) (0.859) budget_bal_gdp 0.237 0.210 0.191 0.187 0.169 (0.208) (0.150) (0.130) (0.121) (0.115) Inflation 0.107 0.115 0.104 0.105 0.0956 0.0444 (0.0814) (0.0781) (0.0751) (0.0756) (0.0750) (0.0625) odanet_gni -0.202 -0.200** -0.205** -0.205** -0.269*** -0.291** -0.315** -0.333** (0.172) (0.0968) (0.0958) (0.0934) (0.0898) (0.115) (0.128) (0.132) IMF dummy 1.233 1.283* 1.379** 1.396** 1.508** 1.458** 1.579** 1.741*** (0.833) (0.674) (0.682) (0.664) (0.619) (0.716) (0.665) (0.612) Fitch rating -0.649*** -0.677*** -0.662*** -0.660*** -0.620*** -0.628*** -0.574*** -0.625*** (0.208) (0.200) (0.182) (0.183) (0.163) (0.198) (0.108) (0.106) GDP growth -0.154 -0.144 -0.161 -0.145 -0.183 -0.194 -0.110 (0.177) (0.178) (0.180) (0.131) (0.116) (0.147) (0.104) lvix 2.947** 2.801** 2.746** 2.756** 2.651** 2.328** 2.196** 2.417*** (1.300) (1.223) (1.188) (1.177) (1.152) (1.117) (0.904) (0.865) ffr 0.279* 0.274* 0.307** 0.310** 0.340*** 0.394*** 0.375*** 0.403*** (0.161) (0.143) (0.135) (0.133) (0.123) (0.123) (0.128) (0.122) Constant 13.23* 14.17** 14.44** 14.27** 13.71*** 14.61** 13.77*** 13.62*** (6.691) (5.964) (5.374) (5.350) (5.051) (5.932) (3.800) (3.659) Observations 46 50 52 52 52 52 59 59 R squared 0.54 0.60 0.61 0.61 0.61 0.60 0.60 0.60 Joint F test for testing the statistical signifcance of the excluded 7 variables from Regression 1 to Regression 8: p-value=0.66 Robust standard errors in parentheses. Logistic panel data regression. *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations. The most general (unrestricted model) includes battery of pull and push factors (Regression 1, Table 4). Institutional quality (the rule of law) and the current account balance as a share of GDP were insignificant (p value above 0.9) and therefore dropped from the model. In the more parsimonious specification (Regression 2), the debt to GDP ratio becomes insignificant with a p value above 0.95 and is therefore omitted from subsequent regressions. The final regression is presented as Regression 8.5 The joint significance of the excluded variables was tested with the joint F test, which supported our decision to exclude these variables from the 5 In the regressions we substituted GDP growth with GDP per capita growth and also the institutional variable – rule of law was substituted with others from the same group such as political stability, government effectiveness etc. and we obtained consistent results as the ones shown in Table 4. These results are available from the authors’ upon request. 16 model (see Table 4). The goodness of fit as measured by R squared coefficient improves marginally with exclusion of the redundant variables. The most restricted model (Regression 8) includes 59 coupon rates from frontier market issuances and the goodness of fit is relatively good at 0.60. The estimated results are generally consistent with the findings from the existing empirical literature (Section 2). In terms of push factors, the FFR and the VIX index have positive impact on the coupon rate; a higher federal funds rate increases the opportunity cost faced by investors and therefore leads to higher coupon rate of the issued Eurobonds. In a similar manner, higher expected volatility on the international financial markets increases the risk aversion of investors, driving higher coupon rates. Regarding domestic pull factors, there is robust evidence that the existence of an IMF program increases the coupon rate, indicating that countries that are under an IMF arrangement are considered riskier than those without an arrangement. Higher levels of overseas development assistance also reduce the coupon rate, potentially reflecting increased availability of foreign exchange or perceptions that a high level of international assistance represents a form of implicit guarantee against default. A better risk rating assigned from the credit rating agencies reduces the coupon rate, presumably due to lower risk perceptions. Variance decomposition of the coupon rates (Figure 10) indicates that the largest contribution in determining variations in the coupon rates is the sovereign risk rating followed by the federal funds rate. The rest of the variables make almost equal contributions. Figure 10: Variance decomposition of the coupon rate of issued international sovereign bonds 60% 50% 40% 30% 20% 10% 0% -10% -20% -30% -40% -50% odanet_gni imf rating lvix ffr Source: Authors’ calculations. Discussion of regression results Overall, the results above show that pull factors still matter. A larger group of countries is now considered ‘eligible’ for market access, and the economic fundamentals and institutional quality of these countries appears to be lower than those of earlier frontier issuers. But the relative performance of countries among the group of eligible economies matters in determining market access and pricing. Further, while low global post-crisis interest rates may have initially spurred investors to consider a broader range of countries as potential borrowers, interest rates are unlikely to, on their own, ensure a continued broadening of 17 access to capital markets or even sustained access for current frontier issuers given the continued importance of other macroeconomic and governance factors. 5. Case study analysis Over recent years there has been a structural shift in global bond issuance, with smaller, poorer, and less well-governed economies increasingly accessing global capital markets. Pull factors still seem to exert an important influence, with domestic macroeconomic and governance factors playing an important role in determining access and pricing. In this section we seek to verify these findings with reference to the experience of two case study countries. We examine the experiences of Ghana and Zambia - two Sub-Saharan African frontier markets that have repeatedly issued global bonds to meet fiscal financing needs over recent years. We explore the conditions in which these frontier issuers were able to access global credit markets, the subsequent deterioration of pull factor conditions, and the subsequent increases in financing costs and emergence of repayment risks. Case study countries are not selected on the basis that they are representative, but rather because they illustrate particular debt sustainability risks associated with broadened frontier market access. Bond issuance Ghana issued its first Eurobond (US$750 million with a coupon of 8.5%) in October 2007 (Table 5). In issuing, Ghana become one of only four African countries to have issued prior to the global financial crisis6. After a pause of almost six years, Ghana issued its second Eurobond of US$ 1 billion in August 2013, then a further US$ 1 billion in both 2014 and 2015, and US$ 750 million in September 2016. Zambia issued its first Eurobond later than Ghana, in September 2012, in a wave where many other African countries also took up the opportunity to raise capital in this manner of the first time7. Zambia’s first issue was for US$ 750 million and was 15 times over-subscribed (Table 5). Zambia issued a second time in 2014 for US$ 1 billion, and again for US$ 1.25 billion in 2015. Table 5: Ghana and Zambia's Eurobond Issuance Year Amount Maturity Coupon GDP Public Issued (US$ growth (%) Sector Debt m) (% GDP) Ghana 2007 750 2017 (bullet) 8.50% 4.3 31.0% 2013 1,000 2023 (bullet) 7.88% 4.8 55.9% 2014 1,000 2026 (bullet) 8.125% 3.9 69.3% 2015 1,000 2030 (bullet) 10.75% 3.9 71.6% 2016 750 2000-22 9.25% 3.6 66.9% (3 annual installments) Zambia 2012 750 2022 (bullet) 5.375% 6.7 25.5% 2014 1,000 2024 (bullet) 8.5% 4.9 35.2% 2015 1,250 2025-27 8.97% 3.0 52.9% (3 annual installments) Source: World Bank Development Indicators and Bloomberg. 6 At the end of 2008 South Africa, Seychelles, Gabon, Republic of Congo and Ghana were the only countries to have issued Eurobonds from the SSA region. 7 By the end of 2012 eleven countries from the SSA region had issues Eurobonds: South Africa, Seychelles, Gabon, Republic of Congo, Ghana, Senegal, Cote D’Ivoire, Nigeria, Namibia, Angola and Zambia. 18 Debut issuance for both countries followed periods of robust growth. But in both cases, issuance proceeded over a period in which important macroeconomic imbalances were beginning to emerge. Ghana’s debut Eurobond came at a time when the country had been growing strongly. In the five years up to its issuance average economic growth was 5.5% (2002-06) and inflation was in single digits (8.4% in 2007). Ghana’s debt to GDP ratio was also low at this point following rounds of debt relief, including reaching the HIPC completion point in 2004. Between its first and second Eurobonds, Ghana revised its GDP series in 2010 (including rebasing) leading to an upward shift in its GDP of 60% and it being classified as a lower-middle income country. This provided a boost for Ghana’s credentials as an investment destination, although the shift also meant that it would gradually lose access to concessional financing. The higher level of growth and prospects for future gas and oil revenues attracted investors to Ghana. However, the period following initial issuance is also characterized by high and twin deficits and growing debt levels as Ghana not only tapped Eurobonds, but also scaled-up its infrastructure borrowing (especially from China) resulting in its debt-to-GDP ratio increasing from 31% in 2007 to 72% in 2015. The large imbalances led to Ghana signing a three-year IMF program in April 2015. Continued strong performance against some macroeconomic indicators (growth, new exports and international support) and deteriorating performance against others (macroeconomic imbalances and growing debt levels) led to Eurobond pricing fluctuating between 7.88% and 10.75% across its first five issues8. Much like Ghana, Zambia experienced robust economic growth from the mid-2000s. Zambia’s average growth was 8.25% in the five years leading up to its debut issue in 2012 buoyed by an increase in copper mining production (its main export) and prices. On account of this growth Zambia was reclassified as a lower-middle income country in 2011. Furthermore, the macroeconomic situation had improved throughout the 2000s and inflation was 6.8% in 2012. Zambia’s debt levels were also low in 2012 given the conservative fiscal stance since debt relief, including HIPC completion, in 2005, which saw debt levels plummet from pre-debt relief highs of 124% of GDP in 2004. These factors helped Zambia issue its first Eurobond at 5.4%, a much lower costs than subsequent issuances. Interest in the second and third issues was helped by the growth story continuing despite copper prices falling from their 2011 peak, and a revision (including rebasing) of the level of GDP upwards by 25% in 2014. However, balanced against this was a growing debt burden and increasing financing needs driven by rising and repeated fiscal deficits, all while the copper price continued to decline. The cost of Zambia’s second Eurobond was over 3 percentage points higher than the first and the third cost nearly 9% as signs of an economic slowdown and an external financing deficit emerged. Debt risks Over the period of issuance, deteriorating macroeconomic performance posed growing risks to debt sustainability for both countries. As commodity prices fell, growth and the external position weakened. At first, the impact was mainly on the composition of Zambia’s and 8 Care needs to be taken when comparing the costs of Eurobonds as the market environment is different at each issue, some have different tenors (for example the 2015 paper in 15 years) and repayment agreements. 19 Ghana’s debt portfolios, as opposed to their levels, but with later issuance their debt to GDP ratios have also increased rapidly. Foreign currency denomination of Eurobonds meant that currency risks resided with the issuer. Both countries recently suffered large depreciations of their local currency against the US$ (in which the Eurobonds are denominated). Zambia’s Kwacha depreciated by 40% in 2015 and the Ghanian Cedi by 50% in 2014. Such foreign exchange risks needs to be managed and assessed carefully by sovereign borrowers. Depreciation had immediate and substantial impacts on the fiscal position through the growing cost of debt service in local currency. Zambia’s interest costs, as a percentage of domestic revenue climbed from 6.7% in 2014 to an estimated 18.4% in 2016 (Smith, Davies, Chinzara 2016). Mounting repayment risks reflected not only large payments in single years, but also the bunching of repayment years. The redemption profiles of both Ghana and Zambia show a significant concentration of payments over coming years and efforts will likely be needed to buy-back some of this debt in in the hope of smoothing repayments. Growing concerns about debt sustainability are evident in sovereign ratings and debt sustainability assessments. Ghana’s debt to GDP reached 72% in 2015 from 31% in 2007 while Zambia’s reached 53% in 2015 from 26% in 2012. From its first issue in 2007 to 2013 Ghana had been regarded by the World Bank and IMF Debt Sustainability Assessments (DSA) as being at ‘moderate risk of debt distress’. This status was elevated in to ‘high risk’ in 2014. Zambia was deemed to be at ‘low risk’ from the time of its first Eurobond issue in 2007 to mid- 2015 when its status was elevated to ‘moderate risk’. Equally the long-term foreign currency sovereign rating of Ghana, assessed by Fitch Ratings, remained at ‘B’ between 2006 and 2015, although the outlook switched from positive in 2006 to flipping between negative or stable from 2008 to 2014. Fitch rated Zambia ‘B+’ in 2011 and 2012, but downgraded the country to ‘B’ in 2013 where the rating remained with stable or positive outlook in 2014 and 2015. Growing risks are also reflected in pricing. Over the period Zambia and Ghana have issued market appetite for frontier bonds has waxed and waned with global trends and the availability and prospects for returns from wider emerging in relation to OECD economies. This has been clear in the pricing of the Ghanaian and Zambian paper in the secondary market (Figure 11). Eurobond paper from both Ghana and Zambia has traded at a premium when compared to other SSA issuers in 2014-16, although there has was some narrowing of this gap between June 2016 and September 2016 as investors’ outlooks for both countries became more positive. 20 Figure 11: Ghana and Zambia's secondary market pricing Source: Bloomberg Policy measures are being explored to manage emerging debt sustainability risks, but it is not clear that such measures will be effective. Both countries have announced the intention to establish sinking funds as a means of reducing repayment risks. However, both countries are far from running budget surpluses and any contributions therefore needs to be borrowed. There are benefits of sinking funds, including lower costs of future borrowing linked to a sinking fund, but borrowing at nearly 10% to place money in low yielding safe financial assets is both expensive and means that sinking fund assets are accumulating only very slowly. This and Ghana’s recent efforts to buy-back most of the Eurobonds maturing in 2017 suggests that refinancing will more likely be the tool selected to ensure funds are available to repay the bonds as they mature. 9 Also reflecting recognition of repayment risks, Ghana and Zambia switched to Eurobonds that are amortized over three years, instead of a single year, in 2015. This change will smooth the debt profile and hence manage repayment risks better. Conclusions from case studies Ghana and Zambia were able to access global credit markets on the basis of strong growth performance and low levels of debt following HIPC relief. Both countries retained access to global credit markets and continued to issue Eurobonds as growth slowed with commodity price declines, fiscal balances deteriorated, exports slowed, and governance remained weak. The use of Eurobond instruments magnified risks, with US$ denomination increasing exchange rate exposure and the bullet structure posing repayment risks. As a result of these risks, debt sustainability and credit ratings declined, and risk premiums on the coupon rates for new issuances and in the secondary market grew. Both countries now face difficult adjustments to reduce debt costs and risks and avoid continued dependence on future issuance of relatively expensive Eurobonds to roll over maturing bonds. 9 As of 26 October 2016 only US$ 199 million or 27% of Ghana’s paper maturing in 2017 was outstanding. 21 6. Conclusions and policy implications This paper has explored the recent trends in bond and syndicated loan issuances by frontier markets and the implications for debt risks. Through simple analysis of the characteristics of issuing frontier markets we have shown that there has been a structural shift in the composition of issuing countries, with smaller, poorer, and less well-governed economies now accessing global credit markets. The need for risk diversification by international investors and the search for yield have enabled a wider range of frontier markets to access international financial markets. Through cross-country regression analysis, however, we have shown that the same factors as those identified in previous literature continue to exert an important influence on issuance and pricing. While a broader range of frontier markets are now enjoying access to global credit markets, pull factors continue to influence the likelihood of issuance and the pricing of bonds. Frontier countries with strong GDP growth, prudent fiscal policy, good external positions and sound institutions are more likely to be able to access global credit markets. Countries with good credit ratings are likely to face substantially lower prices for private sector debt. Through analysis of case studies we have illustrated the country-level implications of these findings. Both Ghana and Zambia are among the new cohort of frontier market economies that have successfully and repeatedly accessed global credit markets. In the lead-up to debut issuance, both countries experienced strong growth and reductions in debt levels through HIPC relief. In the context of the search for yield, both countries were able to issue Eurobonds repeatedly to meet fiscal financing needs, with markets interpreting strong growth performance as evidence of creditworthiness. Over the period of issuance, however, underlying vulnerabilities, including emerging macroeconomic imbalances, reliance on commodity exports, and weaknesses in policies and governance surfaced, leading to deteriorations of debt risk and credit ratings. Risks were heightened by the US$ denomination and bullet structure of Eurobond debt. Both countries now face difficult adjustments to manage and reduce debt risks. These findings, overall, demonstrate a general heightening of risk due to a broader range of countries now issuing on international capital markets and the need for increasing caution on behalf of frontier-market policy makers in managing these risks. Firstly, there is an increased need for ‘policy discipline’ among frontier market countries when accessing global capital markets. ‘Market discipline’ has recently played a weaker role in rationing access for countries facing macroeconomic or governance vulnerabilities. Policy responses need to reflect debt sustainability considerations and the reality that market conditions might be much more restrictive when current and recent borrowing must be refinanced. Such policy discipline should involve: 1. Taking full account of debt sustainability risks and vulnerabilities in fiscal policy decision- making. This means both restricting borrowing within sustainable limits and ensuring that borrowing is used to finance viable public investments that can support future growth. An important additional element of managing debt sustainability risks is taking account of the pro-cyclicality in market access illustrated in our results. Countries that may have easy access when growth is strong and budget deficits modest may find their access is restricted and more costly during periods of downturn. Changing market access and borrowing costs as macroeconomic conditions evolve should be taken carefully into account when making fiscal policy decisions. 22 2. Carefully monitoring and responding to debt management risks. Case study countries illustrate that frontier markets are taking borrowing opportunities when offered with little account taken of portfolio composition and associated debt risks. These risks can be particularly pronounced given the bullet structure of most Eurobonds. There remains an active need to ensure all decisions regarding the composition of borrowing are guided by a strategy that takes currency and repayment risks into account. Building debt management capacity and developing and using medium-term debt strategies to guide debt portfolio composition is vital. Secondly, frontier markets need to ensure that policies support continued access to private credit markets at reasonable cost. Empirical evidence from cross-country analysis and case studies suggests that private capital markets remain highly sensitive to a small number of indicators when allocating and pricing capital among frontier markets. Firstly, the focus of international investors on growth performance and the budget balance suggests the need for continued efforts to pursue growth-enhancing structural reforms and prudent fiscal management. Secondly, the importance of international credit ratings in determining access and pricing suggests that countries should invest in: i) improving and sustaining performance against policy dimensions taken into account by credit rating agencies; and ii) ensuring that information of relevance to credit rating agencies is quickly and transparently communicated, including through adequate debt recording and reporting practices and open dialogue with rating agencies. Finally, frontier markets should continue to work towards the development of domestic debt markets. Recent experience has shown that under appropriate institutional settings larger frontier economies are capable of fostering relatively deep and efficient domestic debt markets. Domestic private debt can provide an important alternative source of fiscal financing through which to reduce reliance on foreign-denominated Eurobond debt with its associated refinancing and currency risks. 23 References Antzoulatos, A., 2000. On the determinants and resilience of bond flows to LDCs,1990– 1995. J. Int. Money Finance 19 (3), 399–418. Ahmed, Shaghil and Andrei Zlate (2013). "Capital Flows to Emerging Market Economies: A Brave New World?" International Finance Discussion Papers 1081, Board of Governors of the Federal Reserve System. Washington, DC. Alfaro, Laura, Sebnem Kalemli-Ozcan and Vadym Volosovych (2008). "Why Doesn't Capital Flow from Rich to Poor Countries? An Empirical Investigation". Review of Economics and Statistics 90(2) (May): 347-68. Bellas, Dimitri, Papaioannou, Michael G., Petrova, Iva, (2010), “Determinants of Emerging Market Sovereign Bond Spreads: Fundamentals vs Financial Stress” Working Paper 10/281 (Washington: International Monetary Fund). Blanchard, Oliver and Francesco Giavazzi (2002) “Current Account Deficits in the Euro Area: The End of the Feldstein-Horioka Puzzle?” Brookings Papers on Economic Activity 33(2), 147-210. Caballero, Ricardo J., Emmanuel Farhi, and Pierre-Olivier Gourinchas (2008) "An Equilibrium Model of `Global Imbalances’ and Low Interest Rates," American Economic Review 98(1), 358-393. Chuhan, Punam, Stijn Claessens, and Nlandu Mamingi (1998). "Equity and Bond Flows to Latin America and Asia: The Role of Global and Country Factors". Journal of Development Economics 55 (2) (April): 439-63. Contessi, Silvio, De Pace, Pierangelo, 2011. The (non)-resiliency of foreign direct investment in the United States during the 2007–2009 financial crisis. Pac. Econ. Rev. 17 (3), 368–390. Coeurdacier, Nicolas and Philippe Martin (2009) “The geography of asset trade and the euro: Insiders and outsiders,” Journal of the Japanese and International Economies 23(2), 90-113. Das, Udaibir S., Michael G. Papaioannou, and Magdalena Polan, 2008a, “Debut Sovereign Bond Issues: Strategic Considerations and Determinants of Characteristics,” in Debt Relief and Beyond: Lessons Learned and Challenges Ahead , ed. By Carlos A. Primo Braga and Dörte Dömeland (Washington: The World Bank). Das, Udaibir S., Michael G. Papaioannou, and Magdalena Polan, 2008b, “Strategic Considerations for First-Time Sovereign Bond Issuers,” IMF Working Paper 08/261 (Washington, November). Edwards, S., 1984. LDC foreign borrowing and default risk: an empirical investigation: 1976–1980. Am. Econ. Rev. 74 (4), 726–734. Feldstein, M. and C. Horioka (1980) ‘Saving and International Capital Flows’ The Economic Journal, vol. 90, No. 358. 24 Francis, Aykut, and Tereanu (2014) ‘The cost of private debt in the credit cycle’, Journal of International Money and Finance Volume 41, pp146-181. Fratzscher, M. (2011) ‘Capital flows, push versus pull factors, and the global financial crisis’ European Central Bank Working Paper No. 1364, European Central Bank, Brussels. Gelos, G. R.Sahay, and G. Sandleris (2004) ‘Sovereign Borrowing by Developing Countries: What determines market access? IMF Working Paper WP/04/221, IMF. David A. Grigorian, (2003), "On The Determinants of First-Time Sovereign Bond Issues," Working Paper 03/184 (Washington: International Monetary Fund). Guscina, Pedras, and Presciuttini (2014) ‘First-time international bond issuance – new opportunities and emerging risks’ IMF Working Paper, 14/127. Hale, G., 2007. Bonds or loans? The effect of macroeconomic fundamentals. Econ. J. 117 (516), 196–215. Hostland, D. (2009), ‘Low-Income Countries’ Access to Private Debt Markets’, World Bank Policy Research Working Paper 4829. Hou, Z., Keane,J., Kennan,J., Massa, I, and te Velde D. (2014). “Global monetary shocks: Impacts and policy responses in Sub-Saharan Africa”. ODI Shockwatch Bulletin. Available at http://www.odi.org.uk/. IMF and World Bank (2015) Public Debt Vulnerabilities in Low-Income Countries: The Evolving Landscape, Joint Board Paper, IMF and World Bank, Washington DC. Jeanneau, S., Micu, M., 2002. Determinants of International Bank Lending to Emerging Market Countries. Discussion Paper No.112, BIS Working Paper. Kalemli-Ozcan, Sebnem, Bent E. S¿rensen, and Oved Yosha (2001), “Economic Integration, Industrial Specialization and the Asymmetry of Macroeconomic Fluctuations," Journal of International Economics 55, 107. Kamin, S., von Kleist, K., 1999. The Evolution and Determinants of Emerging Market Credit Spreads in the 1990s. Discussion Paper No.68, BIS Working Paper. Lane, Philip R. and Gian Maria Milesi-Ferretti (2008) "The Drivers of Financial Globalization," American Economic Review 98(2), 327-32. Lucas, Robert E. (1990) “Why doesn't Capital Flow from Rich to Poor Countries?" American Economic Review 80(2), 92-96. Mauro Mecagni, Jorge Ivan Canales Kriljenko, Cheikh Anta Gueye, Yibin Mu, Masafumi Yabara, and Sebastian Weber (2014), ‘Issuing international sovereign bonds: Opportunities and challenges for sub-Saharan Africa’, IMF Working Paper 14/02, IMF, Washington DC. Mecagni, M. J. Kriljenko, C. Gueye, Y. Mu, M. Yabara, and S. Weber (2014) ‘Issuing International Sovereign Bonds: Opportunities and Challenges for Sub-Saharan Africa, IMF Working Paper 14/02, IMF, Washington DC. Mendoza, Enrique, Vincenzo Quadrini, and José-Victor Ríos-Rull (2008) “Financial Integration, Financial Development and Global Imbalances”, forthcoming, Journal of Political Economy. 25 Min, H., Lee, D., Park, M., Nam, S., 2003. Determinants of emerging market bond spreads: cross-country evidence. Global Finance J. 14 (3), 271–286. Reinhardt, D., L. Ricci, and T. Tressel (2010) ‘International Capital Flows and Development: Financial Openness Matters’ IMF Working Paper WP/10/235. Sandri, Damiano (2010) "Growth and Capital Flows with Risky Entrepreneurship," IMF Working Papers No. 10/37. Smith. G., F. Davies, and Z. Chinzara (2016), ‘Beating the Slowdown in Zambia: Reducing Fiscal Vulnerabilities for Economic Recovery’, World Bank Policy Note September 2016. Song, Zheng M., Kjetil Storesletten, and Fabrizio Zilibotti (2009) “Growing like China,” CEPR Discussion Paper No. 7149. Sy. A (2015), ‘Trends and Developments in African Frontier Bond Markets’, Africa Growth Initiative Publications, March 2015. te Vilde, D. (2014) ‘Sovereign Bonds in Sub-Saharan Africa: Good for Growth or Ahead of Time?’ Overseas Development Institute Briefing #87. Thomas, M., T. Stucka and D. Merotto (2015), ‘African debt since HIPC: how clean is the slate?’ World Bank MFM Discussion Paper No. 2/ 97567. Tyson, J. (2015) ‘Sub-Saharan Africa International Sovereign Bonds: Part 1 – Investor and Issuer Perspectives’ EPS PEAKS Report. Available from: http://www.odi.org/sites/odi.org.uk/files/odi-assets/publications-opinion-files/9435.pdf Tyson, J. (2015b) ‘Sub-Saharan Africa International Sovereign Bonds: Part 2 – Risks for Issuers’ EPS PEAKS Report. Available from: http://www.odi.org/sites/odi.org.uk/files/odi- assets/publications-opinion-files/9434.pdf Tyson, Griffith-Jones, and te Velde (2014) “Post-crisis trends in private capital flows to developing countries” Overseas Development Institute Report, Overseas Development Institute, London. Ul Haque, N., Kumar, M., Mark, N., Mathieson, D., 1996. The Economic Contents of Indicators of Developing Country Creditworthiness. Discussion Paper No. 96/9, IMF Working Paper. Verdier, Geneviève (2008). “What drives long-term capital flows? A theoretical and empirical investigation,” Journal of International Economics 74, 120-142. World Bank (2014) Global Economic Prospects: Coping with Policy Normalization in High-Income countries, World Bank, Washington DC. World Bank. (2009). “World Bank Group Guarantee Instruments: 1990-2007: An Independent Evaluation”. Washington D.C. Yin, R. (1984) Case Study Research: Design and Methods, Sage Publications, Washington DC. 26 Annex 1: Summary statistics Number of Variable: Mean Std. Dev. Min Max observations: ltradeprice 623 4.73 0.49 1.98 7.57 tradepricechange 530 1.01 15.19 -92.08 204.44 coupon rate 92 7.06 2.30 1.72 12.75 Bond/loan issuance dum 1,646 0.07 0.26 0.00 1.00 ffr 1,504 1.87 2.03 0.10 6.33 lvix 1,504 2.99 0.30 2.48 3.47 gdpgrowth 1,398 5.17 3.70 -33.10 54.16 lgdppercapita 1,396 7.39 0.78 4.90 8.68 gdpcapitagrowth 1,398 3.60 3.77 -34.89 50.12 debt_weo 1,457 50.74 25.47 7.45 344.32 externaldebt_gni 1,361 42.54 29.01 4.29 186.17 external_ppg_gdp 1,364 26.42 20.83 0.80 125.32 debtservice_external_exp 869 13.94 9.91 0.42 64.87 debtservice_ext_ppg_exp 869 6.76 4.69 0.17 26.15 cab_weo 1,500 -1.76 5.65 -27.35 25.58 budget_balnce_gdp 1,482 -2.86 3.68 -35.40 23.38 fdi_gdp 1,405 3.27 3.79 -5.50 45.27 tradeopen 895 62.45 26.13 21.35 137.47 reserves_gdp 823 1.24 5.40 -51.89 26.13 infl 1,326 8.53 20.98 -10.07 513.91 odanet_gni 1,270 3.04 4.89 -0.07 62.19 lreer2010 960 4.54 0.19 4.01 6.72 imf 1,504 0.37 0.48 0.00 1.00 commodity 1,504 0.20 0.40 0.00 1.00 lrating 453 3.15 0.12 2.83 3.40 goveff 1,316 -0.31 0.41 -1.96 0.63 polstab 1,316 -0.73 0.84 -3.18 1.19 regquality 1,316 -0.26 0.42 -2.15 0.69 ruleoflaw 1,316 -0.49 0.42 -1.95 0.37 cpia 1,396 3.72 0.42 2.40 4.50 Source: authors’ calculations on the data obtained from various sources: WDI, IMF -WEO database, Bloomberg etc. 27 Annex: 2: Summary of relevant literature Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) Presbitero A. F., Ghura Emerging markets and Sovereign bond 1. Global market condition: 10yr T- 1.Global market conditions matter for market D., Adedeji O. S., and developing economies issuance and bond yield and volatility (VIX); access. And higher global financial market volatility Njie, L. (2015). (EMDE10) with some spread 2. Domestic macroeconomic factors: and lower yield increases the spread. International Sovereign are LIDCs during the per capita GDP, growth, external debt 2. Strong fiscal positions as measured by fiscal Bonds by Emerging period 1995-2013 level, fiscal balance, current account, balance and debt level are the most important Markets and Developing reserve and existence of IMF program predictor of market access. They are also associated Economies: Drivers of in last 3 years; with lower spread. Issuance and Spreads, 3. Country size (population) 3. Strong government effectiveness increases the IMF Working Paper, WP probability of bond issuance. 15/275 4. Increase in the size and income level increases the market access. Feyen, E., Ghosh, S., 71 emerging markets Sovereign and 1. Global push factors: VIX index, The paper focuses on the impact of global factors Kibuuka, K. and Farazi, S. and developing corporate bond Libor-OIS spread , the corporate on bond issuance decision, pricing and maturity. (2015). Global Liquidity economies during credit spread, the size of federal The findings are and External Bond 2010-14 reserve balance sheet (mortgage- Issuance in Emerging backed securities + US treasuries), US 1. Post-crisis period has seen an unprecedented Markets and Developing 10-year Treasury yield. surge in external bond issuance and stocks across Economies. The World 2. Domestic pull factors: GDP per emerging and developing economies (EMDEs). Bank Policy Research capita, growth, current account Bond yields (and spreads) at the time of issuance Working Paper, balance, external debt as percentage have fallen to record lows. WPS7363. of GDP, total bank credit to the private sector as % of GDP. 2. Global factors have a powerful impact on primary activity. A decrease in U.S. expected equity market (or interest rate) volatility, U.S. corporate credit spreads, and U.S. interbank funding costs and an increase in the Federal Reserve’s balance sheet (i) raise the odds that the monthly issuance volume of a country-industry is above its historical aver- age; (ii) decrease individual bond yields and spreads; 10 By classification of WEO. 28 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) and (iii) raise bond maturities, after controlling for country pull factors, bond characteristics Fratzscher M. (2011). Micro-level data of Portfolio 1. Push factors: crisis events, liquidity 1.Common factors (push) – such as specific crisis “Capital Flows Push 13830 equity funds investment flow (TED spread), risk (VIX), US macro events, changes to global liquidity and risk Versus Pull Factors and and 6837 bond funds shock, US equity return conditions – are the main drivers on global capital the Global Financial in 50 EMEs and AEs 2. Pull factors: macroeconomic flows. Moreover, the effects of such global factors Crisis”. ECB Working fundamentals (growth, FX reserve, have changed markedly during the crisis. Rise in risk Paper No. 1364. fiscal balance, CA, inflation, debt, and important crisis events triggered the flow from bank credit growth); institutions EMEs to AEs during crisis, while the effects are (sovereign rating, rating change, opposite before and after crisis. political risk index and financial risk index); policy intervention (capital 2.The heterogeneity of the above effects can be control, credit intervention, explained by the domestic pull factors, and in deposition guarantee, debt particular countries’ macroeconomic guarantee, asset relief and capital fundamentals, institutions and policies, which in injection); exposure (trade, capital fact have been the dominant drivers of capital flows openness etc.) in the 2009-10 recovery for emerging market. And openness plays little or no role. Eichengreen, B., and Emerging economies Sovereign, public risk free rate (10-yr US T-bond rate) Higher US rate decrease the odd of issuance and Mody, A., (2000). “What from 1991 to 1996 and private bond and country characteristics as proxy reduce the spread but less than proportionally; the Explains Changing issuance of credit quality including debt ratio, signs of countries characteristics variable all come Spreads on Emerging debt restructure dummy, FX reserve, as expected, i.e., the factor increase the probability Market Debt?.” growth and budget deficit. of issuance reduces the launch spread. Kennedy, M., and Emerging economies EMBI spread Risk (weighted average of HY bond From mid 2002 to mid 2007, the model suggests Palerm, A., (2014). from 2002 to 2011 covering premium and US equity premium), that just over two thirds of the decline in these “Emerging market bond sovereign and foreign debt as share of GDP, spreads on average reflected improved spreads: The role of quasi-sovereign risk*debt, fiscal balance, ln(political fundamentals, with the rest due to easy credit global and domestic bonds risk index), debt servicing, term conditions. Countries with viable fiscal positions, factors from 2002 to structure, US treasury rate low external debt levels, low political risk and 2011”, Journal of importantly healthy foreign exchange reserves faired better. During the 2008 crisis, virtually all of 29 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) International Money and the run-up in emerging market spreads was due to Finance, 43: 70-87 the large increase in our measure of risk aversion. Siklos, P., (2011). Emerging market from Sovereign bond Domestic factors include exchange The study finds the heterogeneity across regions. “Emerging Market Yield 1997 to 2009 yield spread rate regime, inflation gap, domestic There are few common determinants of yield Spreads: Domestic, (EMBI+ or CDS) credit gap, forecast of inflation, spreads across all the markets considered such as External Determinants, growth and current account balance, volatility, proxied by the VIX indicator, central bank and Volatility Spillovers” institution indicator, and oil producer transparency, and the onset of the global financial dummy. External include change in crisis. US fed fund rate, external debt position, global oil price inflation, FDI, FX reserve, openness, dummy for global financial crisis. Kamin, S.B., and von Emerging market from International Characteristic of bond issues or The study does not identify the relationship Kleist, K., (1999). “The 1991 to 1997 bond issuance issuers including credit rating, between industrial country interest rate with EM evolution and and loans maturity, currency denomination, credit spread. determinants of and period dummy, industrial country emerging market credit interest rate spreads in the 1990s” Faria, A., Mauro, P., and Emerging market and Sovereign bond (for modern time) Risk, population, (for modern time) Large country size and good Zaklan, A., (2011). “The developing countries institution quality, openness, GDP per institution quality increases the odd of market external financing of in 1870-1905 and capita, past default access; if selection bias is controlled in the equation emerging markets modern time of spread, the coefficient of quality of institution Evidence from two becomes insignificant; and openness reduces the waves of financial spread while risk increases spread. globalization” Bruno, V. and H.S. Shin Emerging/developing Capital flow Global leverage and growth, local The leverage cycle is the main driver of the capital (2015). “Cross-Border and advanced through global leverage and growth, change in RER, flow through banks before 2008. Banking and Global economies where the banks M2, GDP and debt ratio, inflation Liquidity” foreign banking penetration are high Guscina, Pedras, and First time sovereign Sovereign bond Growth, quality of institutions, The paper confirms signs of the explanatory Presciuttini (2014) ‘First- bond issuance since issuance inflation, money supply, CA balance, variables in prior studies. time international bond 2003 debt to GDP, federal fund rate, VIX 30 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) issuance – new opportunities and emerging risks’ Gelos, G.R., Sahay, R., 139 developing Bond issuance Size of the country measured by Country size and perceived quality of institution and Sandleris, G., (2011). countries and syndicated population; income volatility and matter substantially; the countries vulnerable to “Sovereign borrowing by loan from 1980 to vulnerability to shocks measured by shocks are less likely to tap international capital developing 2000 by std of GDP growth and term of trade; market; openness does not matter. countries:What sovereign or openness; political instability; quality determines market guaranteed by of institution; country liquidity; IMF access?” sovereign program; defaults; and control variable: time dummy etc. Calvo, Guillermo A., Latin economies from Monthly capital Use component analysis to identify Factor analysis reveals that the foreign factors that Leonardo Leiderman, Jan 1988 to Dec 1991 flow proxied by global factor capture the US interest rates, economic activity as and Carmen M. Reinhart official reserve well as swing in investment return in US market, (1993) “Capital Inflows and rate account for sizable share of variance of real and Real Exchange Rate exchange rate exchange rate and reserve. Appreciation in Latin America: the Role of External Factors.” World Bank (2014) Developing countries Private capital QE episodes; The global factors accounted for about 60 percent “Global Economic 2009 to 2013 flow including Global financial conditions: US fed of the increase in capital inflow to developing Prospects: Coping with bond and equity fund rate, US M2, yield curve, VIX; countries, with the remainder explained by Policy Normalization in portfolio flows, Real side global condition: GDP country-specific factors. High-Income countries”, FDI and cross- growth, global PMI; World Bank, Washington border bank Domestic pull factor: GDP level, Global financial conditions play important role in DC. lending. institutional rating, GDP growth determining the level of capital flow and are signed differential (vs. US), int rate consistent with priori (negative); real side factors differential (vs. US) have modest effect; QE episode has positive and significant effect; in terms of domestic pull factors, the quality of institutional quality is the most significant, followed by growth differential. By breaking down the dependent variables into portfolio, loans and FDI, it is found that portfolio 31 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) flow is most sensitive to external factors associated with monetary conditions, FDI is insensitive to global push factors but more sensitive to country specific characteristics, while bank loans are most sensitive to QE but less to liquidity or portfolio rebalancing factors. Francis, Aykut, and Private sector Micro-level cross- For bond level cross-sectional The paper studies the determinants of credit Tereanu (2014) ‘The cost borrowing in emerging border estimation: spread of private debt. The findings are as follow: of private debt in the and developing syndicated loans Bond attributes: sector, debt First, borrower characteristics associated with credit cycle’ countries in three and international collateral and investor grade; loan lower loan spreads are not necessarily associated cycles: early 1990s to bond issuance maturity and value; regional and time with lower bond spreads. Second, we find Asian financial crisis, dummies. differential effects of borrower characteristics 1998 to 2001 and 2002 For panel estimation: between cycle phases for loans and bonds to 2008. Macro variables: reserves to GDP, separately. Third, we find strong reductions in the investment to GDP, imports plus cost of debt finance during periods when inter- exports as a ratio to GDP, the growth national debt flows are more than one standard rate of real GDP, the change in the deviation above their mean, but not for real effective exchange rate, and expansionary periods, when the growth rate of inflation; debt flows is increasing. Time-specific fixed effects. In the panel estimate, it is found that higher trade ratios in the borrower’s home country raise loan spreads more in periods of high credit flows but have no effect on bond spreads. At the same time, borrowers residing in countries with high investment ratios pay lower spreads on bond issuance particularly during periods of high credit flows, but we find no similar effect for loan spreads. Inflation rates, real exchange rates and previous banking crises have small impacts on loan and bond Min, H., Lee, D., Park, M., 11 emerging Bond issuance Liquidity and solvency variables: Emerging economy specific liquidity related Nam, S. (2003). economies in Latin and external debt to GDP, FX reserve to variable play very important role using the test of Determinants of Asia 1991-1999 GDP, CA to GDP, debt service to zero restriction; macro fundamentals are 32 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) emerging market bond exports, GDP growth, import growth, important; US T-bill rate has a positive effect on spreads: cross-country NFA; spread, while oil price has no significant effect. evidence. Global Macro fundamentals: terms of trade, Finance J. 14 (3), 271– CPI, Exrate; 286. External shocks: oil price US T-bill rate; Debt related: maturity and value; Dummies Antzoulatos, A., 2000. On Latin countries from Monthly public By country regression; On demand The paper examines the determinants of bond the determinants and Jan 1990 to Dec 1995 and private Bond side: cost of foreign borrowing, flows to Mexico, Argentina and Brazil. resilience of bond flows flow excluding lagged domestic int rate, lagged FX The positive effect of global bond flow is found on to LDCs, 1990–1995. J. debt res, CA, lagged bond flow; on supply the bond flow to Mexico and Argentina; the Int. Money Finance 19 restructuring side: global bond issuance, US decreasing developed country interest rate induced (3), 399–418. flows interest rate, lagged flow to region, the capital flow to Argentina and Brazil. variance of inflation, credit rating. Ul Haque, N., Kumar, M., 60 developing Country GDP growth, inflation, CA/GDP, terms The most domestic factors that have influenced Mark, N., Mathieson, D., countries creditworthiness of trade, reserves/import, external credit rating are country’s reserve holdings and 1996. The Economic indicators from debt, RER, variability in TOT, income lagged CA. Contents of Indicators of three rating invariability, FX reserve, Developing Country agencies from imports/GNP, growth of exports, All developing countries rating were adversely Creditworthiness. 1980 to 1993 variability in CA affected by an increase in international interest Discussion Paper No. rate independent domestic factors. 96/9, IMF Working Paper. Uribe, Martin, Yue, 7 emerging economies EM spread VAR model of GDP, domestic The paper attempts to disentangle the interrelation Vivian, 2006. Country from 1994 to 2001 investment, trade balance/GDP, US between country spread, world interest rate and spreads and emerging interest rate, EM spread=EMBI+ domestic business cycle. countries: who drives minus US rate whom? J. Int. Econ. 69 Country spread shock explains 12 percent of (1), 6–36. movement in domestic economic activities, and in turn, macro fundamentals explain 12% of change in spread. US interest rate explains significant fraction of 33 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) movement in output and domestic activities, however mostly due to the domestic spread’s response to world interest rate. Jeanneau, S., Micu, M., Emerging market International Push factors: economic cycles/output Both push and pull factors had significant impact on 2002. Determinants of economies 1986-2000 bank lending gap in lending countries, excess aggregate international bank lending. International Bank liquidity (real interest rate), risk Lending to Emerging appetite /aversion (yield difference Short-term lending is explained by limited number Market Countries. between BBB-rated corporate bond indicators, related mainly to creditworthiness, Discussion Paper No. and US t-bond); exchange rate risk and financial market 112, BIS Working Paper. Pull factors: international trade and performance, while long-term lending is affected regional bias in lending (trade flows), by a broader set of factors. cyclical conditions, exchange rate volatility, external debt and creditworthiness (export/GDP), dummy to control Brady operations, rate of return (IFC rate of return) Nini, Greg, 2004. The Emerging market Syndicated loan Local bank inclusion variables: The paper examines the participation of local bank Value of Financial economies in Latin to private sector number of bank in the syndicate and in foreign arranged syndicated loans and its effect Intermediaries: Empirical America and Eastern bank liquid liability/GDP; on borrowing cost. Evidence from Europe from 1995 to Credit risk variables (linear and Syndicated Loans to 2002 square); type of borrowers; type and The local bank is more likely to be included in the Emerging Market purpose of loans syndicates for riskier borrower. Borrowers. Discussion Paper No.820, Federal The participation of local bank reduced the Reserve Board borrowing cost by around 10 percent of average International Finance spread. Discussion Paper. Contessi, Silvio, De Pace, US 2006-2010 FDI flow to US Industry characteristic: financial FDI flows are divided into four components, which Pierangelo, 2011. The vulnerability measures (External are equity, intercompany debt, reinvested earnings (non)-resiliency of financial vulnerability (share of capital and valuation adjustment. foreign direct investment not financed by internal cash flow), or in the United States trade credit, or tangibility (share of Variations in the cost of finance in the source during the countries have little or no effect on the total FDI 34 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) 2007–2009 financial liquid assets)), factor intensity inflows. But the effect varies depending on the crisis. Pac. Econ. Rev. 17 measure; component of FDI under question, industry (3), 368–390. Cost of finance: immediate rate or ST financial vulnerability and whether in crisis. rate; Country level: factor abundance, GDP, US retail sales, total commercial loans; fixed effect Das, Udaibir S., Michael 31 emerging market Private sector Sovereign risk measures (rating, It is found that sovereign default to private G. Papaioannou, and economies from 1980- external spread or debt-crisis episodes and its creditors has a strong negative impact on corporate Christoph Trebesch, 2004 and 1993-1997 borrowing in characteristics); other control external borrowing for period 1980-2004; Delays in 2008, “Sovereign Default bond issuance or variables including GDP, inflation, debt renegotiations have an additional negative risk and Prviate Sector syndicated loans GDP per capita, RER, total capital flow spillover effect, but not intercreditor disputes and Access to Capital in and equity to EM, political stability, VIX, volume creditor litigation against the sovereign. Emerging Maarkets.” issuance of sovereign issuance For the period 1993–2007, deterioration in sovereign risk perceptions negatively affects corporate access to capital, in particular, the volume of corporate external borrowing. Economic development (per capita GDP) and external factors such as total capital flows to emerging markets are additional main determinants of corporate access to external credit and equity. There is no evidence for close co-movement between public and corporate access to capital. Das, Udaibir S., Michael 21 Emerging market Debut sovereign Regression on size: (1) rating, growth, Only small part of the paper is dedicated to G. Papaioannou, and and low income bond issuance inflation and deficit; (2) VIX and Libor; empirical study on the determinants of the size and Magdalena Polan, 2008a, countries from 1996 to On spread at issue: rating, reserve cost of debut bond issues. “Debut Sovereign Bond 2008 and EMBIG spread Issues: Strategic The size of the issuance doesn’t seem to be affected Considerations and by most of the macro or market condition variables. Determinants of The issuance size depends more on market Characteristics.” 35 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) convention than on macroeconomic variables. The cost of the issue (measured by spread and coupon) depends on countries credit rating and global market conditions. The spread increase with low rating and higher EMBIG spread. IMF and World Bank 50 LICs that were both Sovereign bond Global factors: US 10-year T-note Accommodative global conditions such as looser (2015) “Public Debt eligible for IDA and issuance yield, VIX; domestic factors: GDP per global liquidity and low global volatility, as well as Vulnerabilities in Low- PRGT as of end-2014 capita, GDP growth inflation, CA/GDP, strong domestic fundamentals increase the Income Countries: The including 14 frontier fiscal balance/GDP, debt/GDP, likelihood of bond issuance. Regression shows that Evolving Landscape.” LICs over 1995 to 2014 private credit/GDP, inflation, higher GDP per capita, lower dependence on government effectiveness, aid/GDP, foreign aid and past issuance increase the dummy for previous issuance probability of issuance, while the coefficients of debt level, inflation, GDP growth, CA and private credit have expected signs but are not statistically significant. David A. Grigorian, 38 emerging market First time and Global: US T-bill yield, US growth; Lower US interest rate and higher growth are (2003), "On The economies that issued subsequent Domestic factors: GDP, share of world associated with higher probability of first and Determinants of First- emerging market sovereign bond GDP, reserve, CA/GDP, ext subsequent market access, while the effects are Time Sovereign Bond bonds during 1980- issue debt/exports, fiscal balance/GDP, more pronounced for first time issuance. Issues," Working Paper 2002 GDP per capita, growth, ToT, trade 03/184 (Washington: openness, FDI/GDP, IMF program Other results are as follow: GDP (+ but less International Monetary dummy pronounced for subsequent), CA (+ for first-timer Fund). but not others), foreign reserve (+ first time not others), income level (+), fiscal balance (- showing demand overweighs fiscal discipline), inflation (-), 36 Citation Country coverage Instrument Explanatory variables tested Main findings coverage (i.e. corporate bonds, sovereign bonds, bank lending, etc.) openness (no effect), IMF program (no effect), ToT (+) The paper concludes differential treatment for first- timer and repeater is needed. Reinhardt, D., L. Ricci, All countries with Capital flow Initial GDP, Index of capital account When accounting for the degree of capital account and T. Tressel (2010) population larger than (current openness, cross term, fiscal openness, the prediction of the neoclassical theory ‘International Capital 1 million (109 account/GDP in balance/GDP, old age dependency is confirmed. For countries with open capital Flows and Development: countries included in most ratio, population growth, NFA/GDP, accounts, less developed countries tend to Financial Openness the study based on specifications, oil trade balance/GDP, Per capita GDP experience net capital inflows and more developed Matters’ IMF Working data availability) 1980- others are FDI, growth countries tend to experience net capital outflows, Paper WP/10/235 2006 portfolio equity, conditional of various countries’ characteristics. portfolio debt and other investment as well as reserve accumulation) 37