عوامل گسترش چرخه اعتباری وابسته در بازارهای در حال رشد بدهی های مستقل

We address credit cycle dependent sovereign credit risk determinants. In our model, the spread determinants' magnitude is conditional on an unobservable endogenous sovereign credit cycle as represented by the underlying state of a Markov regime switching process. Our explanatory variables are motivated in the tradition of structural credit risk models and include changes in asset prices, interest rates, implied market volatility, gold price changes and foreign exchange rates. We examine daily frequency variations of U.S. dollar denominated Eurobond credit spreads of four major Latin American sovereign bond issuers (Brazil, Colombia, Mexico and Venezuela) with liquid bond markets during March 2000 to June 2011. We find that spread determinants are statistically significant and consistent with theory, while their magnitude remarkably varies with the state of the credit cycle. Crisis states are characterized by high spread change uncertainty and high sensitivities with respect to the spread change determinants. We further document that not only changes of local currencies, but also changes of the Euro with respect to the U.S. dollar are significant spread drivers and argue that this is consistent with the sovereigns' ability to pay.

The behavior of emerging market sovereign spreads—e.g. during the 1982 and 1995 Mexican crises, the 1997 Asian, the 1998 Russian and the 2001 Argentine crisis—repeatedly raised considerable concerns among financial institutions, central bankers, as well as investors worldwide. While various sovereign spread determinants were proposed in the literature, the magnitude and the impact of financial risk factors on emerging market spreads are not yet fully understood. Most of the previous work on the determinants of credit spreads addresses unconditional determinants.1 However, there is evidence that spread determinants exhibit varying significance during time, which may relate to the state of the underlying credit cycle.2 Due to institutional shifts and other instabilities in emerging markets (see e.g. Bekaert and Harvey, 2003) and the particular features of emerging markets bonds (see Erb et al., 2000) it is sovereign debt markets in particular, which are prone to time-variation. The highly indebted Latin American economies, which are dominant representatives of emerging market sovereign debt, therefore appear as natural candidates for an examination of the stability of spread determinants. Given the setting above, we provide further insight into the determinants of sovereign spread changes in Latin American emerging market economies. To this aim, we address spread changes conditional on an unobservable state of an endogenous credit cycle. In our model, sovereign spread determinants may exhibit differing significance and sign conditional on the state of this cycle. The feature is captured by modeling daily spread changes in a Markov regime switching framework given a set of explanatory variables (see e.g. Ang and Bekaert, 2002 and Hamilton, 1990). Our modeling approach considers financial market spread determinants, which are derived in the tradition of the Merton (1974) structural model of default. Structural models predict that asset and interest rate factors are the key determinants of credit spreads. Of course, the causes and consequences of default in the sovereign setting differ from those in the corporate setting. Institutional arrangements and bankruptcy laws are typically ambiguous and political issues may govern defaults in sovereign settings. Still, understanding which variables relate to changes in sovereign spreads and understanding the underlying credit cycle dynamics remains a challenging task in the pricing process of sovereign bonds. Previous studies of sovereign bonds, which address this area include for example Claessens and Pennacchi (1996), Westphalen (2001), Chan-Lau and Kim (2004), Keswani (2005), Batten et al. (2006), Diaz Weigel and Gemmill (2006), Gray et al. (2007), Gapen et al. (2008) and Galai et al. (2011).3 As Bekaert and Harvey (2003) point out, emerging equity markets have gained much more research attention than emerging bond markets. This is somewhat surprising given today's economic relevancy of Latin American sovereign issuers and the fact that Latin American lending had already become widespread in the mid 19th century. The region continues to be the dominant issuer of international bonds among emerging market issuers. As such, e.g. according to BIS (2011) statistics, Latin America accounts for about 35% of the total of emerging markets outstanding international bonds and notes. The region's reliance on external debt financing makes its debt markets highly sensitive to international capital flows and to credit events in particular. For example, following the 2001 Argentine default up to mid 2002, Latin American countries including Brazil suffered from limited debt market access. More recently, IMF (2011) (p. 35–40) reports that emerging market economies including Latin America face new challenges associated with rapid credit growth and large capital inflows. We examine bond issues by four leading Latin American sovereign issuers, namely Brazil, Colombia, Mexico and Venezuela. These countries maintain the most liquid bond markets in the region and—given BIS (2011) December 2010 bond market statistics—represent nearly 70% of the outstanding Latin American debt of 405.7 billion U.S. dollars. We use the structural model as the starting point of our analysis and examine the importance of our explanatory variables as determinants of changes of sovereign credit spreads, including an asset price proxy of the country's capacity index, the risk-less short rate, the steepness of the term structure, implied U.S. equity market volatility, changes in the gold price as indicator of global uncertainty and finally the local exchange rate as well as the U.S. dollar/Euro exchange rate. In contrast to previous sovereign bond studies, we condition the economic magnitude of the explanatory variables on the unobservable state of the sovereign credit cycle. Our sample includes 2853 daily logarithmic credit spread changes of eight liquid U.S. dollar denominated Eurobond issues during the period March 1, 2000 to June 30, 2011. Our results underline that the determinants of sovereign credit spread changes behave as predicted by theory. However, explained spread variation in individual bond series appears low. Performing a principal component analysis (PCA) of the residual series we identify an omitted component which influences all credit spread changes with approximately the same magnitude. This component can be interpreted as an unobservable endogenous sovereign credit cycle variable, which is modeled by our two-state Markov regime switching framework. We identify a low spread change volatility non-crisis and a high spread change volatility crisis state. The approach has several implications. First, it improves model fit. Second, the results reveal that spread change determinants generally have a higher magnitude during the crisis regime. Finally, the framework allows for detecting breaks in the behavior of the relevant credit risk determinants. Here, the Argentine default is an ideal event given that there were no other major defaults in the Latin American region, or in other emerging markets, after the beginning of our sample period in March 2000. The remainder of this paper is organized as follows. In Section 2 we outline our model of sovereign spread changes and introduce the explanatory variables. In Section 3 we explain the features of our Latin American Eurobond issues and our dataset. We discuss the empirical results in Section 4. Section 5 concludes.

We investigate important pricing questions for emerging markets sovereign bonds based on a set of liquid Latin American Eurobond issues. The determinants of daily sovereign credit spread changes generally follow the ex-ante theoretical predictions as initially summarized. Our results show that a common factor is missing in the conventional analysis of the spread changes in our sample. The incorporation of an unobservable endogenous credit cycle variable improves model fit in spread change explanation. In the crisis regime, spread determinants exhibit an even higher predicted impact than otherwise. Our empirical results confirm in several directions that sovereign spread determinants exhibit regime dependent behavior. The given findings raise some doubt about unconditional sovereign credit ratings. When, as demonstrated, spread change determinants depend on a persistent credit cycle, slowly following changes in such cycle may induce rating momentum (see e.g. Al-Sakka and ap Gwilym, 2009), which is in conflict with ratings as efficient predictors of future credit risk. Additional research could also address the question whether the inclusion of additional regimes may improve rating accuracy and default prediction.