Determinant Factors of Brazilian Country Risk: An Empirical Analysis of Specific Country Risk

Many studies in international finance try to investigate to w hat extent domestic and external economic factors constitute significant determinant fa ctors of international country risk. This article tries to analyze, for the period 1992-2003, Bra zili n country risk from the point of view of three empirical models: i) First, the internal eco nomic determinants of the country risk; ii) the second has the same purpose as the first, with the difference that the variable “intensity of global risk aversion”, that serves as proxy fo r the external component of the risk, is included in the group of explanatory variables; iii ) in the last model the emphasis is on the relation between specific country risk (country risk m inus the external component) and the internal and external economic determinants.


Introduction
Unlike the 1980s, which where characterized by the inaccessibility of international capital markets for most emerging nations, the 1990s saw a reversion in this picture by the intensification of international capital flows to emerging economies, mostly by acquisition of shares, bonds or by making direct investments.Motivated by this movement, many researchers draw their attention to the investigation of the most important factors behind this dynamic.Among the most important studies are those of Calvo et al. (1993), Fernandez-Arias (1996), Taylor and Sarno (1997) and Chuchan et al. (1998), which distinguish two important types of factors as determinants of these capital flows: internal and external factors.
The internal factors are related to the own opportunities and risks that a specific market offer to international investors.In this case the country's internal economical factors are often used to represent this risk.
On the other side, the external factors are associated with external opportunities and risks, such as the macroeconomic situation in the industrial countries, international interest rates and the intensity of risk aversion of international investors.
An alternative to the analysis of international capital flows is the investigation of the internal and external determinant factors of international emerging market bond prices: international movements of those bond prices, from the country risk's point of view, are not only consequence of the perception of macroeconomic conditions, but also of the perception and valuation of global risk.
The nineties and the beginning of the twenty-first century give an interesting example of the impact of external factors on the price of sovereign debt bonds, given the high frequency of crises and turbulences occurred during this period, such as the Mexican Peso Crisis (1995), the Asian Crises (1997), the Russian Crisis (1998), the Brazilian Crisis (1999), the Argentine Crisis (2001), and the terrorist attacks of September 11, 2001.Given the existence of some studies that already analyzed the relationship between internal and external factors and country risk, like Canton and Packer (1996), Min (1998) and Nogués and Grandes (2001), the main objective of this study is to extend the existing research by estimating three key models and applying these models to the analysis of Brazilian country risk in the period ranging from 1992 to 2003: i) in the first model the domestic determinants of the country risk are tested; ii) in the second model the variable "intensity of global risk aversion" is included in the model as proxy variable for the external county risk factor; and iii) in the third model the dependent variable is modified to measure the "specific country risk", defined as country risk less the estimated external component.
The most important difference of this study, if compared to similar studies, is that it uses a new approach to the analysis of country risk, called ARDL (Autoregressive Distributed Lag) Model.In this context, the main advantage of this technique is that it can be applied to models irrespective of whether the regressors are I(0) or I(1) or mutually cointegrated.It avoids the conventional pre-testing procedure of unit roots associated with cointegration analysis and has the advantage of be easily understood within the framework of traditional error correction modeling approaches.

Review of Studies of Country Risk
As already mentioned, the nineties experienced expressive international capital flows to emerging markets, mostly in form of investments in bonds, shares and direct investment.This reverted the tendency of the eighties, characterized by a continuous fall in capital transfers to these countries, as consequence of the default crises during this period.In this context, one key variable reveals itself as one important determinant in the explanation of the international capital flows dynamic: the country risk.
From the definition used by Claessens and Embrechts (2002), country risk is a measure associated with a country's default probability and is caused by events which can be at least to some degree under the control of the government but definitely not under the control of a private enterprise or individual.In quantitative terms, country risk is represented by the yield difference between a risky and a non-risky asset, which is, in turn, dependent on the general liquidity conditions in international markets and the behavior of international investors, their degree of risk aversion and the risk attributed by them to single assets (Canuto and Santos, 2003).
Basically country risk has two components: domestic and external risk.Domestic risk refers to the specific country risk determinants, which are related to the economic fundamentals, such as for example the fiscal and balance of payments situation, the stock of international reserves, the real growth rate of the economy and the inflation rate.The external risk, on the other side, includes all the global factors, which are mainly represented by the risk free interest rate, by the contagion effects of financial crises and by the degree of international investor risk aversion Calvo et al. (1993), Fernandez-Arias (1996), Taylor and Sarno (1997), Kim (2000).
In a recent study Fiess (2003) investigated the relation between country risk and capital flows to Argentina, Brazil and Mexico in the period 1990-2001, and Venezuela, in the period 1996-2001.Starting from the idea that capital flow dynamics can be explained both by domestic and external factors, the author decomposed country risk in two components: i) specific country risk, which comprises opportunities and risks of domestic investment; ii) global risk, which reflects investment risk in industrial countries.Using a VAR methodology, the author found strong evidences that specific country risk exercises a strong influence over capital flows.In the case of global risk there was also a strong statistical representative effect, except for the case of Argentina.
Departing from this relation between international capital flows and country risk, we can conclude that the analysis of the determinants of both variables is quite similar.In both cases the risk-return perception of foreign investors results in changes in the risk premium and in the direction of international capital flows.The advantage in the analysis of country risk lies in the empirical finding that the asset prices adjust themselves more quickly to a change in domestic and external factors than the quantities of capital flows (Oks and Padilla, 2000).
Based on this discussion following question is posed: what are the most important determinant factors of country risk?Many of the studies cited below have been developed about this topic and most of them come to the conclusion that country risk is basically explained by both domestic and external factors.Among the domestic factors most frequently cited as being statistically significant are: public debt, inflation, real economic growth rate, external debt, international reserves and current-account surplus.As for the external factors, two variables are constantly tested: the U.S. Treasury Notes interest rate and the degree of risk aversion of international investors.While there seems to be a positive influence of the latter one over the country risk, the effect of the U.S. Treasury Notes interest rate is still subject to strong controversy.
The high intensity of international emerging markets bond trading during the nineties lead to the growing interest of many researchers in the relation between yield spreads and economic fundamentals.In this category we can include Min (1998), which studied this relation for the US dollar denominated, fixed-income securities of emerging markets issued during the period from phonelstrans1991-1995.The study was realized by the decomposition of the explaining variables in four groups: i) liquidity and solvency: external debt/GDP ratio (+) 1 , debt service/exports ratio (+), international reserves/GDP ratio (-), current account surplus/GDP ratio (-), imports growth rate (+), exports growth rate (-), real GDP growth rate (-) and net foreign assets (-); ii) macroeconomic fundamentals: domestic inflation rate (+), terms of trade (-) and real foreign exchange rate (+); iii) external shocks: international oil price and three-month U.S. Treasury bill rate; and iv) dummy variables: whether the issue was made before or after Mexican Crisis (1995); the emission was public or private; the emission was made by a Latin-American or Asiatic country.Besides that, the author included two other variables: maturity of the asset and amount issued.The results of the panel regression were very satisfactory since all the variables of liquidity and solvency had statistically significant coefficients and their signs were in conformity with the expected ones.In the variable group "external shocks", none of the variables presented a coefficient which was statistically different from zero.Finally the results for the dummies lead to following results: i) private issues have a tendency to show higher yield spreads than public ones; ii) there is significant difference between the yield spread before and after the Mexican Crisis; iii) the region of origin of the is-sues is not a relevant factor to explain differences in the yield spread.In relation to the amount issued and the maturity, this last variable showed a negative signal, as expected.
Concerning articles analyzing exclusively the time series analysis of Brazilian country risk, Andrade and Teles (2003) analyzed the effect of macroeconomic policies on the Brazilian country risk in the period from January 1991 to December 2002.They used a model called Country Beta Market Model, where the countryrisk is a time varying coefficient.The model starts from the idea that the beta of a CAPM model that measures the covariance of the Brazilian Stock Exchange with the market return of the rest of the world is a proxy for the Brazilian country risk.In order to capture the effect of the macroeconomic policies, the following variables were selected: i) budgetary necessity of the public sector -primary concept; ii) oil prices; iii) international reserves; and iv) the interest rate, measured by the Brazilian Selic Rate.The initial estimation of the model showed the variable referring to the fiscal situation of Brazil as not being statistically significant, which was then excluded from the model.For the following estimations, the Kalman filter was applied to the coefficients in order to test their long run stability.The study showed that during the observed period monetary policy played a relevant role, the interest rate exerting a negative influence over the Brazilian country risk.It could also be observed that international reserves had a negative effect over the country risk.However their effect seemed to be more important during the fixed exchange-rate regime.Grandes (2002) changed the focus of the study from that of Andrade and Teles (2003), which emphasized the effect of monetary policy over the country risk, and tried to identify the permanent and transitory impacts of macroeconomic fundamental variables and of variables related to external monetary policy on the Brazilian, Argentine and Mexican sovereign risk during the 1993-2001 period.Based on the empirical analysis of Kharas (1984), the following macroeconomic variables were considered: current account deficit/GDP ratio (+/-), debt service/GDP ratio (+) and GDP growth rate (-).Grandes selected, as variables representing the external factors: i) the FED funds rate (+) as proxy for monetary policy; ii) the 30-year U.S. Treasury rate (+/-), which may capture the so called portfolio substitution effect (+) and the "fly to quality" effect (-); and iii) dummy variables respectively for the Mexican, Russian and Brazilian crises.In relation to the effect of the 30-year U.S. Treasury rate it should be mentioned that through a substitution effect, it was expected that increases in the 30-year bond interest rate made investment in these bonds more attractive, so that the supply of loanable funds for emerging countries would diminish and, therefore, the country risk would increase.On the other hand, Grandes (2002) argumented that in periods of extreme crises -which was a frequent observation over the period under his analysis-the flight-to-quality effect seemed to prevail.A more risk-averse behaviour could push U.S. bond rates down (excess demand of USTB bonds) and increase country risk of emerging markets.Hence, the expected sign of the 30 year U.S. Treasury rate was ambiguous.
For each country, Grandes (2002) estimated individual equations using the SUR-SYS (Seemingly Unrelated Regressor System Equation) technique.The estimations showed that the permanent effects of the debt service/GDP and current account deficit/GDP ratios are statistically significant for Argentina and Brazil.Concerning the variable GDP growth rate a statistically significant negative permanent effect was observed for the three countries.The analysis of the dummy variables demonstrated that the Mexican and Russian crises propagated to Argentina and Brazil and that the Brazilian crisis affected the Argentine economy.Finally the FED funds rate was significant for the Brazilian (positive sign) and Mexican (negative sign) country risk.The impact of the 30 year U.S. Treasury rate was statistically significant only for Brazil, indicating that a rise in that rate reduces the Brazilian country risk.
The main difference between the present study and the studies discussed above is in the use of a new methodology, called ARDL approach.The use of this statistical method allows testing the influence of selected factors on Brazilian country risk both in the short and in the long run, making the analysis more dynamic as compared to previous studies.

Data
As already mentioned, this study covers the period 1992-2003.All the data are quarterly and were extracted from IPEA-Data, with the exception of the proxy used for the Brazilian country risk, obtained from JP-Morgan.
In this study, the dependent variable on which we will analyze the influence of selected economic indicators is the specific country risk.Obtaining the specific country risk is directly related to the isolation of the external risk component.A variable indicated in the literature as representative for this risk is the so called high yield U.S. bond spread over the U.S. Treasury rate, which acts as proxy for global risk aversion.
High yield bonds are debt instruments issued by firms to finance diverse activities and are considered being of speculative grade by most of the rating agencies.The most important characteristic of these bonds, also known as junk bonds, is that they offer a high expected return as function of their high default risk.In our work, in order to extract the external component of the country risk we regressed the EMBI+ spread (defined below) against the high yield spread.The residual obtained in this regression, which characterizes the part of the country risk not explained by the external risk, was given the name "specific country risk" or "domestic country risk".
The most well known indicator for measuring the emerging markets risk premium is the Emerging Market Bond Index Plus (EMBI +), calculated by J.P. Morgan.The index is composed by external federal debt instruments of emerging countries, denominated in foreign currency and negotiated in secondary markets.
A large portion of these instruments is composed by Brady Bonds, followed by negotiated loans, Eurobonds and domestic bonds.In quantitative terms the EMBI+ Index is published by J.P. Morgan in two ways: in absolute level and sovereign spread, the last one being the most well known measure for country risk.In level, the EMBI+ is calculated on the basis of the price average of the bonds that compose the basket, weighted by their traded volume on the secondary markets.On the other hand, the sovereign spread is calculated on the basis of the difference between the returns of sovereign bonds of a given country and the returns of the U.S. Treasury Notes, considered to be free of risk.
Knowing that the dependent variable is the domestic or specific Brazilian country risk, we then defined which economic variables would be included in the econometric model.Based on the academic works discussed above, the selection of the models took two requisites into account: i) economic relevance and ii) the performance of the variables in previous studies.For better understanding of the model each of the selected variables is defined and its supposed relation with the country risk will be explained.
1. GDP growth [GROWTH]: this variable is of great importance, being a synthesis of the general economic conditions in a country.We use here the real GDP growth.Furthermore, higher economic growth is associated with a better relation between public debt and GDP, implying a smaller country risk.We expect a negative correlation between GDP growth and country risk.
2. Fiscal Surplus as percent of GDP [FISCSUR]: the amount of resources that the public sector of a country can generate is directly related to the perception that the economic agents have about the solvency of a country.The bigger the fiscal surplus, the greater are the conditions of maintenance or even reduction of the public debt/GDP ratio and consequently the smaller is the default probability.We can therefore expect a negative correlation between the fiscal surplus and the country risk.
3. Public Debt as percent of GDP [PUBDEBT]: this variable is often used as an indicator of the accumulated fiscal performance of a country.A larger debt stock usually results in more difficulty of the public sector to honor the debt service, which increases the default risk (Guardia, 2004).We thus expect a positive correlation between public debt and country risk.

Inflation rate [INFL]
: according to Min (1998), the inflation rate serves as reference for the quality of the economic policy of a country.The same argument is found in Pinheiro (2004), who argues that without a stable economy and correct prices there is a rise in the economic risk coupled with a fall in productivity and in investment.In other words: the higher the inflation, the higher the uncertainty about the economic environment and consequently the higher the risk perception of the economic agents.We can therefore expect a positive correlation between the inflation rate and country risk.
5. Current account surplus (deficit) as percent of GDP [CURRAC]: as discussed in Pinheiro (2004), one way of reducing the country risk is raising its current account surplus, as this increases the liquidity of a country, reducing the default probability.In this sense there we can expect a negative correlation between the current account surplus and country risk.

External Debt as percent of Exports [EXTDEB]
: according to Underwood (1990) and Cohen (1996), the ratio of external debt to exports is an important indicator of a country's solvency and the probability of default is very high when this ratio assumes values in the interval between 200% and 250%.As a consequence, the impact of this variable on the country risk is often taken into account, being one of the few variables considered in the evaluation of the country risk rating of the Institutional Investor Magazine together with the default history and the country's macroeconomic stability.Therefore, we can expect a positive correlation between external debt and country risk.

International Reserves as percent of GDP [INTRES]
: the volume of international reserves has a direct relationship to a country's degree of international liquidity.Starting with a definition by Williamson (1973) that international liquidity measures the ability of a country to honor a current-account deficit, without having to recur to restrictive adjustment measures, we can conclude that the higher the ratio of international reserves to GDP, the higher the country's adjustment flexibility in case of economic shocks.We can expect therefore a negative correlation between international reserves and country risk.
Using these independent variables, we can define the three models that will be studied in this work: in the first model we will test the relation between total country risk and fundamental domestic economic variables; in the second model the analysis will be exactly the same, with the only difference that the external component will be added to the group of explanatory variables; and the third model, on its turn, will test the relation between specific country risk and the economic fundamentals.
Starting with the estimation of these three models applied to the Brazilian case, it will be possible to evaluate to what degree there is relationship between the economic determinants of the country risk and the specific country risk.

ARDL model
In this work, the relationship between country risk and their determinants will be examined in the long and short run using the methodology developed by Pesaran and Shin (1999), which has its emphasis on the ARDL (Autoregressive Distributed Lag) Model.Later this model was also used by Pesaran et al. (2001) in order to develop new approaches, called bounds testing, to the analysis of level relationships between variables with no need to conduct unit root tests.In other words, the main advantage of this procedure is that it eliminates the need for pretesting the variables for the order of integration (and cointegration) and it does not require that the variables be of the same order of integration.It can be used whether the variables are I(0) or I(1) irrespective of whether they are cointegrated or not.As is known from time-series analysis the use of first-differenced (stationary) variables in regression models is necessary to reduce the spurious results that are likely to arise when the variables are specified in their level (non-stationary) form.However, when variables are used in their differenced form, (long-run) information from the data is removed, resulting in a model that can only provide partial (shortrun) information on the relationship between the variables.Further, by not taking into account the potential long-run relationship among the variables, models constructed using only differenced data may be misspecified if there exist such longrun influences, resulting in biased parameter estimates.To avoid such problems, one must test if there is a long-run relationship between the variables in the model.Therefore, in the present context, application of cointegration technique would enable us to examine the long-run equilibrium relationship between country risk and its determinants.The technique would also enable us to trace out the long-run and short-run response of country risk to changes on its determinants.
The ARDL method of testing the long run relationship between variables is based on the estimation, by ordinary least squares, of an unrestricted Error Correction Model (ECM) in following form: where y t is the dependent variable and x t is a vector of independent variables of rank k.
In this case, if φ = 0 and δ = 0 there is a long run relationship between the levels of y t and x t which is given by: where θ 0 ≡ −α0 φ , θ 1 ≡ −α1 φ , θ 2 ≡ −σ φ and v t is zero mean stationary process.If φ < 0 , the long run relationship between y t and x t is stable.In this case, equation (1) can be expressed as an ECM of following form: Following equation ( 3), testing the null hypothesis that φ = 0 can be interpreted as a test of existence of a long run relationship between y t and x t .In the ARDL methodology, this test is made following the joint hypothesis that φ = 0 and δ = 0 based on the ECM defined in equation ( 1).
where, Zt = (LREM BI, GROW T H, EXT DEB, P U BDEBT, F ISC-SU R, IN F L, CU RRAC, IN T RES) = (LREM BI, W t); LREM BI = log of the residuals from the regression of EM BI+ spread against the high yield spread, defined as the specific country risk and W t−1 and W t = all the variables with exception of LREMBI in t − 1 and t.
Before presenting the results of the long and short run estimation of the three models presented above, we test the existence of a long run relation between the dependent variable and the independent ones.
Departing from an optimal lag number of 4, selected on the basis of the AIC -Akaike Criterion -and HQC -Hannan-Quinn Criterion, Table 1 presents the values found for the F-statistics.According with the test from Pesaran et al. (2001), if the value of the F-Statistics, obtained after the estimation of the regression equation by OLS, is out of the range of critical values, we can reject the null hypothesis of no long run relation between the variables.If contrary, we accept the null hypothesis.
For the null hypothesis H 0 := λ = γ = 0 we can accept, using a statistical significant level of 5%, for the three models the alternative hypothesis of a long run relation between the variables.After confirming the existence of this long run relation between the variables in the three models, the next step consists of estimating the long run coefficients and the short run dynamics coefficients from the ARDL model.In this case, for a number of variables (7 in model 1 and 3) and (8 in model 2), with a maximal number of lags corresponding to 4, we processed a total of (4 + 1) (7+1) = 390.625 and (4 + 1) (8+1) = 1.953.125ARDL regressions for the models.
In the following tables the results of the estimated models are presented.Tables 2 and 3 show, respectively, the long and short run domestic determinants of the country risk.Following both selection criteria, the chosen ARDL model was of type (2, 0, 0, 1, 0, 0, 1, 0).
In Table 2, it can be observed that none of the estimated coefficients is statistically significant.In other words, this table shows that the Brazilian country risk has no long run relation with their domestic determinants.In the short run (Table 3), however, three variables were identified as determinants of the Brazilian country risk: public debt/GDP ratio (+), primary fiscal surplus/GDP ratio (-) and the international reserves/GDP ratio (−) 1 .
Going to the second model, Tables 4 and 5 present the domestic and external determinants of the Brazilian country risk, the external risk being represented by the high yield spread.Based on both selection criteria, the ARDL model chosen is of type (2, 0, 1, 0, 0, 0, 0, 1, 0).
In Table 4 it can be seen that the domestic determinants are not long run explanatory factors for the Brazilian country risk.A contrary result can be observed for the external risk component.The regressions lead to the conclusion that there is a statistically significant long run relation between the high yield spread and the Brazilian country risk.The short run analysis (Table 5) shows that the determinants are the same ones identified in the first model: public debt, primary surplus and international reserves.Beside those, the high yield variable, when included in the second model, also turns out to be statistically significant.Finally in Tables 6 and 7 the long run and short run determinants of the specific country risk are illustrated.According to the selection criteria AIC and HQC, the chosen ARDL models for the long run and short run differ from one another.They are respectively of following type: (3, 0, 0, 1, 0, 0, 1, 0) e (2, 0, 0, 1, 0, 0, 1, 0).Similar to the first 2 models, the domestic determinants do not exercise any long run effect over specific country risk, as can be observed in Table 6.
However, as it can be seen on Table 7 in the short run, two economic variables presented coefficients statistically different from zero and with the expected signs: public debt/GDP ratio and international reserves/GDP ratio.It should be emphasized that the error correction term had, for all three models, a statistically significant coefficient at the 1% level, indicating that disequilibria which occurred in a period were corrected in the subsequent one.
Resuming the discussion above, we come to following interesting conclusions: first, in all three estimated models, the long run relation between the domestic variables and the measures of risk (country risk and specific country risk) is not statistically significant.However, when the external risk component is included in the group of explanatory variables, its coefficient turns out to be statistically different from zero, which means that what really matters on the long run is the relation between the external factors and Brazilian country risk.In this study, the indicator used as proxy variable for the external factors was the high yield spread, which is a measure of the degree of risk aversion of the international investor.Second, it could be verified that, in the short run not only the external component affects country risk, but also the domestic components, such as external debt, primary surplus and international reserves.With the exception of the primary fiscal surplus, these same domestic variables revealed themselves as explanatory variables of the specific Brazilian country risk in the short run, leading to the conclusion that there is a consistency among the models.
Briefly, the main idea defended in our current work is: even when the deviations of the domestic variables from their long run tendencies affect the country risk in the long run, if the external conditions are favorable, it is more probable that this deviations will get smaller along time, making the country risk smaller as well.On the other hand, if the long run perspectives about the external scenario are unfavorable, the long run deviations of the domestic variables will be greater, with the consequences of a greater country risk.That means that the external scenario reveals itself as the most important explanatory variable of country risk.

Conclusion
The objective of this article was to identify the determinants of Brazilian country risk and the specific Brazilian country risk during the 1992-2003 period.To achieve this goal, three models were estimated: i) country risk versus domestic variables; ii) country risk versus domestic and external variables; iii) specific country risk versus domestic variables.
As most studies divide country risk into domestic and external components, the following domestic explanatory variables were selected for the analysis of country risk in this work: growth rate, external debt/exports ratio, public debt/GDP ratio, primary surplus/GDP ratio, international reserves/GDP ratio and the current account/GDP ratio.As indicator of the external risk component, the variable high yield spread was used, as it can serve as proxy for the analysis of the international investor's degree of global risk aversion.
In the short run it could be verified that, in addition to the external component, three domestic factors affect the country risk.These are: public debt/GDP ratio, primary surplus/GDP ratio and international reserves/GDP.These same variables, with exception of the primary surplus, are also related with the specific country risk.
In the long run, however, the country risk is explained only by the external component, as the domestic determinants are not statistically significant.In the same way, the specific country risk does not have a long run relation with the domestic variables.
In other words, the results obtained indicate that the country risk is influenced by the deviations of the domestic economic variables from their long run tendencies in different points of time.However, if we assume that the intensity and direction of those deviations depend on external conditions, the results show that, in the long run, the external scenario has the greatest influence over the country risk.
For example, in a favorable external scenario the expectative is that a variable like public debt, which empirically affects country risk, shows a declining tendency line over time.That means, on the long run we have the direct relation between external scenario and the country risk.On the other hand, on the short run, we should expect that the deviations from public debt from its long run tendency line will be smaller under these conditions, having therefore a favorable impact over country risk.In this case, on the short run, we have an association between the external and domestic scenarios and country risk.
However, it should be noted that the findings of this study are based on the results from Brazil.In order to generalize them more studies of specific country risk are needed.In the case of Brazil, one another suggestion for future studies would be to consider narrower periods of time.In our case, for example, the period analyzed (1992 to 2003) includes several breaks that should be considered, such as periods of fixed and floating exchange rate, hyperinflation and stabilization, external shocks, capital market and economic reforms.

Table 1
Test for the long run relation

Table 2
Long run domestic determinants of the country-risk Notes: See Table2.

Table 4
Long run domestic and external determinants of the country risk

Table 5
Short run domestic and external determinants of the country risk Notes: See Table2.

Table 6
Long run domestic determinants of specific country risk Notes: See Table2.