International Finance - The Determinants of Sovereign Risk - Assignment Example

The rate that governments from emerging markets pay on their external debts is usually faced by extreme variations. The difference between the rate offered by the US Treasury on debt and rate that is paid on a country’s external debt, which is dominated in US dollar defines its yield spread – this is the common measure of the borrowing cost for the country, as per the international capital market. The macroeconomic fundamentals are known to influence the disparity in sovereign yield spreads, a factor that will be looked at in this paper. The explanatory power volatility of variables will be a very important factor to consider because this plays a crucial role in the pricing of defaultable debt. In a situation where a country experiences more volatile fundamentals, the probability of undergoing a rigorous deteriorating of fundamentals is so high that the probability of defaulting also becomes high. According to Merton’s (1974) seminal model of risky corporate debt, the risk of defaulting is featured in a superior yield spread on the country’s bonds. The price of a coutry’s exorts in comparison with its imports is memasured by terms of trade, a factor that Bulow and Rogoff (1989) noted that its change affects the country’s ability to raise dollars in revenue since the exports are relatively low – this makes it difficult for such a country to service its external debt, which is ideally denominated in dollars. In addition, the volatility of terms of trade is a factor to reckon with, as it is used to show how output volatility varies at business cycle frequencies in the overall economy – this has an adverse effect on the long run growth of the economy (Mendoza, 1995). A country-specific commodity price index will be created to address the terms of trade, which is essentially partly endogenous. Several authors have studied the impact of country-specific and global factors on emerging markets (Calvo, 2002; Herrera and Perry; 2002). In order to put into account the global factors, this paper will use 10-year US default yield spread – S&P500 index (VIX). This paper will augment the numerous studies that have touched on the empirical indictors of sovereign yield spreads. These studies have used a divergent of methodologies and variables such as reduced form regression of spreads on explanatory variables. Duffie et al. (2003) applied a flexible sovereign spread model of the reduced form. Pan and Singleton (2008) studied CDS spreads, terms of structure for Korea, Turkey, and Mexico, whereby they factored in the risk-neutral credit event loss rates and intensities that explains CDS data appropriately. These studies have related their spreads to factors such as foreign currency reserves, political factors, VIX, and oil prices – however, the pricing model that has been used in these studies does directly apply the macroeconomic fundamentals, which is in contrast with the current study where the impact of variation in macroeconomic on default probabilities abs spreads in fundamentals has been explored directly. Also, many studies have explored the role of macroeconomic fundamentals in forecasting currency and bank crises as well as defaults. For example, Sutton (2002) and Cat˜ao and Kapur (2006) uses hazard model to forecast sovereign default, and found that volatility of terms of trade is an essential indicator. Data description and choice of variables The prices of sovereign external debt over time and across countries will be the focus of this study. The denomination will be in form of dollar debt instruments, which are guaranteed or issued by the governments of the respective countries from South America. The Morgan’s Emerging Markets Bond Index Global (EMBI) will be the measure of the yield spread. The data includes Eurobonds, Brady bonds, and loans with provable prices and a 12-yeras average maturity. The data set includes 3 Latin American countries including …….. The variation across countries and over time seems to be significant. The indicators of spread will be examining by first finding out is there are some indicative global proof that terms of trade are a significant factor that is likely to cause changes in sovereign yield spreads. Table 1 presents data for the three Latin American countries including Argentina, Ecuador, and Venezuela. The Determinants of Sovereign Risk Taking risk neutrality of lenders for granted, Edwards (1986) has explained the rate of return on sovereign debt as an arbitrage situation, when putting into consideration that the capital is highly mobile. This situation can be explained by the following function: (1-p)(1+i’+s) = 1+i’……………………………………………….. (1) The possibility of default, from the above equation, is represented by p While the risk free rate is represented by i’ and the sovereign risk is s. The following function is consequently formed: s = [p/ (1-p)] k.……………………………………………………. (2) Where k = 1+i’ Theoretically, as the probability of default approaches one, the risk premium moves towards infinity. Practically, the risk premium will differ based on the probability of default, for a particular i’. It is, however important to note that it is extremely difficult for a default to approach one, because in such an occurrence the borrower would not be blocked from accessing the bond market because a mounting likelihood of default is a pointer of financial predicament. P can be explained using a logistic function as follows: p = (exp ΣβiXi)/ (1 + exp ΣβiXi) …………………………………………….. (3) In this function, βi are the respective coefficients while Xi are the factors that establish sovereign risk premium. Using natural logs and combining (2) and (3) produces the following function: Log s = α + log k + ΣβiXi…………………………………………………….. (4) The manner in which agencies are rated can, for the purpose of this illustration be as in Nogues and Grandes (2001), which helps in finding out the appropriate Xi variables and in the measure of country risk. A country’s risk can be calculated through its government’s bonds average EMBI spreads. These spreads are used to measure the US Treasury bonds’ risk premium. The influential factors can be determined using the following four classes of variables: 1. Contagion variables to establish the impact of volatility in other financial markets on the respective country; 2. Macroeconomic essentials associated to the probability of debt repayment; 3. Political variables to establish the consequence of disturbances on financial market; and 4. Structural reform variables. A study by Abrego, Flores, Pivovarsky, and Rother (2006) inspires integration of macroeconomic variables in a sovereign risk model that is explained in this paper. These variables include the public external debt to GDP ratio and Debt Service to Export ratio. The contagion variable, which in this case is EMBI index for foreign countries, is used to establish the contagion impact of the foreign countries – this variable is important because it helps explain the headline risk in foreign markets. The purpose of using US Treasury rate is to establish the effect of the activities in the US market in respect to the country’s sovereign risk. It is also expected that the terms of trade for the three emerging markets will get better because the three countries are commodity exporters and hence expecting an improvement in their terms of trade – this will also lead to a decline in spreads as the commodity prices surge. Figure 1 illustrates the way the commodity prices are measured in accordance with the commodity index. The trend that is evident from this graph is that spreads and commodity prices are strongly and negatively correlated. Literally, this means that the commodity exporters are more likely to service their loans when the commodity prices are increasing – this tends to reduce their spread. For a case in point, there was significant decrease in spread from 2002 to 2007, when the significant surge in commodity prices was experienced. Since different countries have different types of export commodities, it will be somewhat erroneous to generalize from this perspective. The leading exports commodities for each country have been reflected in Table 1. For instance, Ecuador is mostly known for crude petroleum, Argentina for feeding stuff for animals and crude petroleum, and Venezuela for crude oil. Country Beginning of EMBI sequence EMBI Spread Volatility of terms of trade Number of defaults Debt/GDP Two exports Argentina 1994 707 9.2 2 42.4 Feeding stuff for animals (10%), Crude petroleum (10%) Ecuador 1995 824 12.4 3 60.9 Fruit, nuts, fresh, dried (18%), Crude petroleum (41%), Venezuela 1994 601 21.8 3 40.3 Petroleum products, refined (25%), Crude petroleum (59%), Table 1: Summary statistics for Argentina, Ecuador, and Venezuela Figure 1: Commodity prices and EMBI Spread Country variables The possibility of default is the variable that defines country’s ability to service its external debts. In turn, this also affects the ability to pay in the international market, or in other words the spread. The annual cross-county data is collected and used to analyze variation over time and across countries. For each of the country’s variables that has been selected, the attention is paid on the measures of fundamentals, which are most likely to be influenced by the risk of default. The ability of a country to earn revenue in form of dollars, which could be used to service the external debt, would probably be affected by factors such as the price of exports compared with imports, and the country’s terms of trade. This metric is very important in the pricing of debt particularly in the oil-exporting countries. In such a country, dollar revenue is earned by exporting oil, and in turn the dollars are used to pay for imports. The oil producing nation stands at a better financial position when the prices of oil goes up, since they are able to generate more revenue dollars and use them to service their debt which is denominated in dollars. The fact that term of trade is a relative price sequence makes it essential to estimate each country’s terms of trade by building a percentage change for the past five years. If a country’s export is dearer, a positive number is expected in that particular country. It is also important to note that the bondholders are concerned about the risk of significant and unfavorable shocks in the future, besides the recent adjustments in term of trade. An example of a future shock in this case could be a significant decline of oil prices, a situation that is characterized by a positive correlation between volatility of terms of trade and spreads. The volatility of terms of trade is calculated as the standard deviation of terms of trade’s percentage change in relation to the last ten years. Figure 2 is the graph of the terms of trade volatility against EMBI spreads – the two variables are evidently positively correlated. Also capture in Table 1 is a measure of each of the three country’s recent default history. This variable is affirmed by Reinhart et al. (2003), who provides that the past default can be used to forecast the future default. It is notable that the three countries that are being studied are predominantly commodity exporters, depending on the world market to determine their export prices. Control and Global Variables The global variables include world interest rates, aggregate risk aversion, and liquidity. The US default world and the VIX index will be included are included to define the spread between ratings of Baa and Aaa. The 10-year US Treasury rate is used as a substitute for the world interest rate, while TED is included to take into account the changes in aggregate liquidity. Other variables include ratio of reserves to GDP and the external debt to GDP, both of which are country-specific variables. As explained in a number of literatures, for example by Edwards (1986), the ratio of debt in spread regressions is described as having a significant positive coefficient. These variables are possibly endogenous, in which case the developments of debt are potentially non-linear and endogenous. In addition, Uribe and Yue (2006) have associated a country’s GDP to its spread. Again, none of these variables can be considered to be a perfect measure of sustainability – this means that the reserves to GDP specifically are likely to be an indicator of liquidity and not solvency. To make the results more comparable, data is presented with and without these variables. 3. Spreads and Macroeconomic Fundamentals Regarding macroeconomic fundamentals and spreads, the will be interesting to establish the level of one variable that can be explained by the other variable. To achieve this, a linear regressions of yield spread, will be run on explanatory variables. This involves making of end-of-year spread sequence, which is equivalent to the yearly incidence of the metrics of fundamentals. The median spread, which is borrowed from Morgan for the month of December, is the measure of end-of-year spread. The spread observation, when the country is not undergoing default, will be the focus of these analyses given that the aim is to examine the indicators of the risk of foreign default and its effect on debt prices. Availability of the full variable is the requirement for an observation to be included. The study is concentrated on the sample that is similar in all specifications to enhance evaluation of significance levels, point estimates, and changing R2 across specification. The summary statistics for the spread regression sample is illustrated s shown in table 2 below. Table 2 provides a summary statistics for the sample of spread regression. As discussed earlier, EMBI stands for spreads on J.P. Morgan’s Emerging Market Bond Index Global. EMBI spread Volatility of terms of trade Change in terms of trade Years since last default Debt/GDP Reserves/GDP Mean 375 5.6 3.5 8.5 46.5 18.1 Median 264 4.3 0.1 9.9 44.4 14.7 St. Dev. 327 4.4 21.1 3.3 21.7 13.7 Min 23 0.6 -31.8 1 11.7 1.4 P5 59.9 1.4 -15.9 1 15.7 5.1 P95 1071 13.5 55.2 10.9 86.9 45.3 Max 1802 25.4 162.3 10.9 114.1 83.4 For the regression to include an observation, all the independent variables must be available. Evidently, there is a significant discrepancy in spreads as well as all the independent variables. The results of the regressions of spreads on a variety of independent variables are presented in table 3. There are three categories for the independent variables, including global time series variables, the main country-specific variables, and control variables. Colum 1 consists of a baseline regression with global and country-specific variables. Changes and volatility of terms of trade are significant at % and they sign has its own significance. This means that those countries with higher spreads will have higher terms of trade volatilities and their terms of trade tend to be deteriorating an observation that affirmed by Reinhart et al. (2003). While the coefficients on the other time sequence variables are insignificant, the VIX index coefficient is significant at 1% and positive. The global factors and the country-specific fundamentals’ regression of spreads produces an adjusted R squared of 0.49 , while the regressions in the 7th column produces a significantly lower adjusted R squared of 0.18, but this time round including the global factors only. Regression Equation: Spread = α + β1vol tot + β2chg tot + β3 ytd + β4VIX + β5DEF + β6r 10year + β7TED + β8 debt GDP+ β9 reserves GDP + dummies + ε 1 2 3 4 5 6 7 8 9 10 COUNTRY VARIABLES Volatility of terms of trade 36.07 (9.68) 46.67 (7.65) 35.50 (9.89) 46.01 (8.54) 31.25 (8.64) 32.53 (8.70) 19.20 (2.65) 20.81 (3.14) Change in terms of trade -5.03 (5.41) -6.15 (4.52) -4.22 (5.33) -5.43 (4.41) -3.75 (4.87) -3.70 (4.77) -2.32 (2.10) -2.17 (2.11) Years since last default -34.11 (5.60) -28.51 (4.21) -24.08 (3.32) -18.23 (2.49) -28.23 (3.76) -16.62 (3.32) -22.10 (3.54) -16.95 (2.35) GLOBAL VARIABLES VIX index (VIX) 9.22 (3.01) 9.81 (3.07) 7.55 (2.20) 8.16 (2.26) 7.35 (2.01) 15.13 (3.56) 15.06 (3.52) 11.38 (3.41) 10.20 (3.07) Default yield spread (DEF) 0.62 (1.30) 0.47 (1.10) 1.04 (1.78)+ 0.91 (1.58) 1.06 (1.89)+ 1.46 (1.54) 1.27 (1.40) 0.84 (1.32) 1.05 (1.54) Treasury 10-year yield 0.08 0.08 0.10 0.10 0.15 0.15 0.17 0.17 0.13 0.13 0.67 0.57 0.80 0.70 0.27 0.27 0.28 0.28 TED spread (TED) 0.19 (0.33) 0.16 (0.29) 0.55 (1.09) 0.52 (1.04) 0.43 (0.87) -0.63 (1.11) -0.70 (1.20) -0.15 (0.42) 0.06 (0.12) CONTROL VARIABLES Debt/GDP 4.15 (4.60) 4.08 (4.13) 4.02 (4.63) 4.14 (4.63) 2.49 (3.54) Reserves/GDP -4.48 -4.37 -3.13 -2.44 -3.55 Adjusted R2 51.7% 50.4% 57.5% 56.2% 60.5% 65.8% 12.1% 51.2% 61.2% 64.0% Table 3. Regressions of sovereign spreads The table 3 above shows the regression results of EMBI spreads on a set of independent variables, including time sequence variables, macroeconomic variables, regional dummy variables, and control variables. In columns 2 and 4, change and volatility in the commodity price index and change of terms of trade is used respectively. In column 8 to 10, dummy variables, which are equivalent to letter ratings, are included. Standard errors are used to calculate the numbers reported in parentheses representing the t-statistics. Read More