Sovereign spreads are key for emerging countries for many reasons. Perhaps the most important is that they constitute the floor for the cost of external capital. Emerging economies have used external bond financing extensively in the last few decades. Latin American countries, in particular, currently account for over 60 percent of the outstanding bonds in the Emerging Market Bond Index (EMBI) constructed by J. P. Morgan. The region's economic growth appears to be closely associated with the amount of net capital inflows received.1
Given the importance of external financing and the volatility of Latin American sovereign spreads, the region's economic authorities need to identify the main driving forces of sovereign spreads. No consensus has yet emerged, despite much debate. One strand of the literature argues that domestic factors—namely, economic fundamentals—are particularly relevant in determining sovereign spreads. Another strand emphasizes external factors, such as international interest rates, global economic growth, and contagion. In this study, we focus on one external factor, namely global investors' attitude toward risk, which we refer to as global risk aversion. This factor has recently received much attention from practitioners and policymakers, though it has not yet come under much scrutiny by academics.
The traditional literature identifies the U.S. risk-free interest rates as the main external factor affecting sovereign spreads. While this is clearly relevant, investors' attitude toward risk should also have a bearing on high-risk markets, [End Page 125] including emerging countries' sovereign bond markets. Risk issues are becoming increasingly relevant owing to the sophistication of financial markets, and the impact on emerging markets of international investors' appetite for risk is clearly recognized today.2
Our paper contributes to this literature by analyzing how investors' attitude toward risks affects Latin American sovereign spreads. An important issue in this endeavor is how to measure the degree of global risk aversion. The measure chosen should be unrelated to default risk, while reflecting investors' pure risk appetite. A broader definition may include the financial position of international investors, which basically depends on the economic cycle and the cost of capital in the investors' country of origin. When the economy is booming and liquidity is ample and cheap, global risk aversion should, in principle, be low. This is exactly what has happened in the past few years. On this basis, this paper not only studies the impact of global risk aversion on Latin American sovereign spreads, but also that of its main driving forces, such as U.S. economic growth and the U.S. risk-free rate. The paper thus contributes more generally to the literature on the external determinants of sovereign spreads.
The rest of the paper is divided into six sections. We start by reviewing the existing literature and setting out the paper's objective. The subsequent section describes our empirical strategy. We then report on the variables and data used. The following two sections offer some stylized fact and present our empirical results. The final section concludes.
Literature Review and Paper Objective
Empirical work on the determinants of emerging countries' sovereign risk has grown markedly in the last few years. This has triggered a lively debate on whether external or domestic factors are most relevant in explaining sovereign spreads. Our literature review focuses on external factors, in particular the cost of capital in the United States.
Calvo, Leiderman, and Reinhart were probably the first authors to point out the importance of external factors, although they concentrate on capital inflows to Latin American countries rather than on sovereign spreads.3 They find evidence that increases in U.S. short-term interest rates are responsible for the [End Page 126] reduction in capital inflows to the region, based on a sample of ten Latin American countries. In turn, Min and Kamin and von Kleist report that the relation between the two variables is statistically...