نقش کشور، خطرات بازار منطقه ای و جهانی در پویایی میزان بازدهی آمریکای لاتین

We analyze the joint impact of country, regional and global market risks on daily changes in yield spreads of Mexico, Colombia and Brazil. In contrast to previous studies, we consider a homogenous set of liquid Eurobonds which are representative of current emerging bond markets. All risk-factor groups are significant but country-specific differences exist. Spread changes of all three countries are mainly driven by global risk. The second most important contributor to spread changes is country risk for Mexico and Brazil but regional risk for Colombia. The sensitivity of spread changes to risk factors varies with bond maturity.

Emerging market sovereign bonds have had a turbulent history over the past 20 years when periods of declining yield spreads often ended abruptly. A significant fall of yield spreads in the early 1990s ended in 1994 with the beginning of the Mexican Peso crisis and the tightening of U.S. monetary policy. The subsequent period of declining emerging market yield spreads lasted from 1995 to early 1997 and ended in a market-wide turmoil which was triggered by the crises in Asia, Russia and Brazil. The latest sustained significant fall in emerging market yield spreads occurred between mid 2002 and 2006; the yield spreads remained low throughout most of 2007. Unlike in previous periods, yield spreads continued to decline despite increasing U.S. short-term interest rates between April 2004 and August 2006. Emerging markets also weathered well the global financial crisis in 2008–2009. Several emerging countries considerably improved their credit quality and debt structure during 2002–2005 and, consequently, received higher credit ratings. The weighted average credit rating of emerging market borrowers in the EMBI Global bond market index increased from B+ in 2002 to BB in 2004 (IMF, 2004a). During this period, the emerging market Eurobond market also experienced a significant increase in the inflow of capital which was generally attributed to abundant global liquidity and low interest rates in major financial markets. Some analysts considered the emerging bond market oversaturated by 2004, claiming that yield spreads are too low to be justified by improvements in economic fundamentals (e.g., IMF, 2004b, IMF, 2006 and Sløk and Kennedy, 2004). An adverse change in global market conditions was predicted to cause an abrupt reversal of capital flows and lead to turmoil in emerging markets. At the onset of the global credit crisis in 2007–2008 IMF (2008) reported a sharp reduction in global liquidity and of short-term capital flows to emerging markets; a much higher risk aversion of investors; and a rising cost of credit for emerging markets. These risks, however, remain as emerging markets are still facing the consequences of the crisis: large financial and corporate writedowns, continuing deterioration of fundamentals in some countries and potentially limited credit availability in the coming years (IMF, 2009a and IMF, 2009b). The role of global market factors versus country risk in explaining yield spreads in emerging markets has been studied in Ferrucci (2003), Sløk and Kennedy (2004), Hartelius et al. (2008) and Ciarlone et al. (2009). Diaz Weigel and Gemmill (2006) point out that a third group of factors which is related to investors’ regional sentiment and common regional market co-movements might be more important in driving emerging market bond prices than the global and country-specific factors. They find country-specific factors to contribute only 8% to the explained variance of the distance-to-default of Brady bonds. The impact of regional investor sentiment is also emphasized by Kamin and von Kleist (1999) and Eichengreen and Mody (1998) who find marked differences between the pricing of debt issued in Latin America and Asia. This paper contributes to the debate on the importance of country-specific vis-à-vis regional and global market risk factors in driving emerging market yield spreads. To this end we study the joint impact and relative importance of these groups of factors by incorporating corresponding market-based measures in an empirical model of yield spread changes. We follow the methodology in Manzoni (2002) and Batten et al. (2006) to account for volatility clustering and autocorrelation effects which are present in daily spread changes. In contrast to the above literature, this paper analyzes changes in yield spreads on zero-coupon Eurobonds rather than coupon-paying bonds, Brady bonds (which are issued with collateral) or bond market indices (which include a range of debt instruments with different characteristics). Only straight liquid Eurobonds with no call options, collateral or other special features are employed in our analysis which allows avoiding the complexity that these additional features introduce. Most emerging market bond indices include a range of debt instruments with various characteristics which alter the bond pricing relationship. Eurobonds are representative of the current structure of the external emerging markets’ bonds market while Brady bonds, who dominated this market in the 1990s, have virtually disappeared. EMTA's Debt Trading Volume Survey (http://www.emta.org) shows that the share of Brady bonds within traded emerging market debt fell from 61% (U.S. $1.68 trillion) in 1994 to about 2% by 2005. Most Brady bonds were redeemed or exchanged for other instruments by most sovereign borrowers. We also contribute to the understanding of the question whether the bond maturity affects the exposure to both country-specific and external risk factors by studying the sensitivity of yield spread changes in three maturity groups (short, medium and long) to these factors. Yield spreads are likely to have different sensitivities to both country-specific and external factors depending on the bond maturity. Though the importance of the debt maturity profile for the vulnerability of emerging market debt to external shocks has been underlined by practitioners (e.g., Cassard and Folkerts-Landau, 1997), this issue has received little attention in the academic literature. Understanding the factors that drive emerging market yield spreads – as well as their interaction with bond maturity – are important issues for asset pricing, portfolio allocation and credit risk management applications. All of this makes the topic a central issue for financial market regulators while the re-occurring turmoil in emerging markets in the recent past (and its impact on global market stability) exposes our lack of understanding. The risks associated with the delayed reaction to the 2008–2009 global financial crisis persist, which highlights the relevance of our inquiry. We consider yield spreads changes during the period of May 2003–July 2009. This period spans the 2003–2006 episode of favorable economic conditions and financial stability, in which emerging market yield spreads declined considerably, and the latest peak of the global financial crisis in 2008. Sovereign U.S. dollar-denominated Eurobonds of Mexico, Colombia and Brazil are chosen for the purpose of our analysis because they have sufficiently large markets at different maturities. The remainder of this paper is structured as follows. Section 2 describes our dataset. Section 3 discusses the theoretical and empirical determinants of yield spread changes. Section 4 contains the model and the estimation results. Section 5 discusses and interprets our findings. Section 6 concludes.

This paper investigates the joint impact of country-specific, regional and global risk factors on daily yield spread changes of Eurobonds issued by Mexico, Colombia and Brazil. Our dataset covers May 2003–July 2009 which include the period of favorable economic conditions that ended in 2006 and an episode of recent financial crisis in 2008–2009. Our study expands the existing literature in a number of ways. First, we study a joint impact of three groups of factors that have been only partially considered by other studies of emerging market yield spreads. Second, we employ yield spreads on zero-coupon Eurobonds which are estimated from the market prices of a homogenous set of liquid Eurobonds. Unlike Brady bonds, Eurobonds are representative of the current structure of sovereign debt. Earlier studies used yield spreads on coupon-paying bonds (mostly Brady bonds) or emerging market bond indices (e.g., EMBI, EMBI Plus and EMBI Global) which include a range of debt instruments with various characteristics which are priced differently. Third, the effect of Eurobond maturity on the sensitivity of the yield spread dynamics to the country-specific, regional and global factors is studied here, which has been overlooked in the literature on emerging market bonds. We find that all three types of risk factors, country-specific, regional and global, are important drivers of sovereign emerging market yield spread changes. The relative importance of these factors, however, varies for countries in our sample. Global market risk factors have the largest impact on daily spread changes. The effect of global factors was at least as important as the combined effect of regional and country-specific factors at most maturity horizons for all three countries. Their relative importance is the strongest for Mexico. Country-specific factors are the second most important contributor for Mexico and Brazil whereas the impact of regional market risk was less important for these countries (marginally so for Mexico and approximately three times less important for Brazil). For Colombia, common regional market factors were the second largest contributor and country-specific factors were the least important. Our findings support the conjecture that a country's global economic positioning, the size of its Eurobond market and its credit quality (investment versus non-investment grade) are among the factors influencing the relative impact of country, regional and global factors on the yield spread changes. However, a broader set of countries would be needed to obtain fully conclusive evidence. Since the response of yield spread changes to country-specific, regional and global factors varies across different countries, pooling country bond data together and not allowing for such differences may lead to misleading conclusions. Our analysis also shows that the sensitivity of yield spread changes to a number of country-specific and external factors varies across the bond maturity rage. This implies that the factor impact can differ markedly for bonds of the same country but with different maturity or duration. One should therefore exercise care when using and interpreting empirical results which are based on a single bond or on a country's aggregate market index as these results may depend on the maturity (duration).