Modeling and Estimating Shadow Sovereign Ratings

This paper describes and evaluates “shadow” sovereign credit ratings, which represent the credit ratings of countries that are not rated by credit rating agencies. Credit ratings represent the creditworthiness of companies or governments. They are important in attracting foreign capital. Countries without credit ratings can face greater difficulties than countries with low credit ratings, for example paying a higher price for capital. This paper has two objectives. The primary objective of this paper was to estimate a rating prediction model to the assess credit ratings of countries that are not yet rated. Large numbers of potential determinants were tested, and nine variables were selected that play a key role in assessing credit ratings. According to the chosen determinants, a highly precise model was calculated (80% of the estimated ratings were identical to the corresponding actual ratings or only one notch different). The purpose of this analysis was to estimate credit ratings for a sample of 31 unrated countries. The results are statistically significant and explained in detail. The second objective of this paper was to demonstrate that countries that are not ranked would not necessarily receive the lowest rating, and the results supported that hypothesis.


Introduction
A sovereign credit rating can be defined as a "ticket that provides access to the international capital market".
Sovereign ratings are assessments that measure the capability and willingness to pay off debts. Investors and fund managers, make their own investment decisions but base them on the decisions of credit rating agencies.
Changes in credit ratings may be the primary motive for buying or selling a particular security. While credit rat-ings have benefits, authors such as Bolton, Freixas, and Shapiro (2012) caution that credit rating agencies also have negative effects, which result from two situations.
In the first situation, because the main goal of credit rating agencies is to obtain profits, competition among agencies can reduce efficiency, as it facilitates ratings shopping. Second, ratings are more likely to be inflated during booms and when investors are more trusting.
the official website of Standard & Poor's, sovereign ratings have increased dramatically over the last twenty years. In 1993, approximately 40 countries were ranked; since then, that figure has risen to nearly 126 countries, but a large number of developing countries have yet to be rated. According to Cantor (2004), "credit risk has been one of the most active areas of recent financial research" (p. 2565). Credit ratings determine the cost of capital and reduce the information asymmetry between investors and debt issuers. There is a strong connection between government borrowing and credit ratings. Afonso, Furceri and Gomes (2012) discovered that credit ratings and outlook changes have a significant influence on government bond yields. According to Bhatia (2002), "sovereign ratings are fundamental building blocks for a global credit risk architecture" (p. 3). Canuto, Santos and Porto (2012) defined "sovereign risk as a credit risk associated with operations involving credit for sovereign states" (p. 4). Sovereign credit ratings play an important role in capital markets, as the country's rating serves as a ceiling for the ratings of corporations and other entities within that country's borders (Borensztein, Cowan, & Valenzuela, 2013). Williams, Alsakka and Gwilym (2013) found that sovereign rating upgrades and downgrades have substantial impacts on bank rating upgrades or downgrades. A sovereign risk assessment is an evaluation of a government's capacity for debt repayment. Why are credit ratings important?
According to Hooper, Hume and Kim (2008), impact of a rating change is experienced in both the capital and foreign exchange market, indicating that rating changes may contribute to capital movement. Credit ratings play an especially important role in the emerging markets, and there are numerous papers on the subject. According to Larraín, Reisen, and Von Maltzan (1997), the "sovereign rating industry has the potential to help dampen excessive private capital inflows into the emerging markets with negative rating announcements" (p. 5). Reisen and Von Maltzan (1998; reported that credit ratings can intensify or attenuate boom-bust cycles in emerging markets. Brooks et al., (2004) found no evidence that emerging markets are particularly sensitive to rating changes; however, the results of an empirical study by Kraüssl (2005) show that credit rating agencies influence the size and volatility of lending in emerging markets.
Kraüssl found that downgrades of government ratings have a stronger impact than do rating upgrades. For further details on emerging markets and credit rating agencies, see Kaminsky andSchmukler (2002), Sy (2002), Kim and Wu (2008), Jaramillo andTejada (2011), andErdem andVarli (2014). Sovereign debt ratings can spill over, even into international stock markets, and Ferreira and Gama (2007) show that sovereign ratings and outlook changes affect the stock market returns of other countries. Gande and Parsley (2005) also confirmed the existence of an international spillover effect in sovereign debt markets. More about role, interests and critics of credit rating agencies in Baresa, Bogdan and Ivanovic (2012). Cantor and Packer (1996)  ings by mapping the probability that the debt-to-GDP ratio might exceed a maximum debt limit at some point in the future. Such a debt limit can be determined ad hoc or based on the financial capacity of a government. Polito and Wickens (2015) also constructed a modelbased measure of sovereign credit ratings derived solely from the fiscal position of a country for calculating the credit ratings of 14 European countries.
The main contribution of the present study is based on the calculation of a highly accurate model that can assign ratings to unrated countries. These unrated countries are mostly low-and middle-income countries. The importance of assigning credit ratings is that investors will always prefer financial instruments that are rated to those that are not.

Data and methodology
The study was conducted based on a full sample that all macroeconomic data used as determinants in the estimated credit rating model were available for those countries. Of the remaining 73 (unrated) countries, 31 were selected for which data were available. Table   1 shows the statistics of the selected sample of rated and unrated countries. Table 1 (Gaillard, 2011). Rating watches or outlooks indicate the probability of a rating change and the direction of that change; however the issuance of a watch or outlook does not necessarily mean that there will be a change in the credit rating. According to Cavallo, Powell and Rigóbon (2008), the outlook was altered at least one year before most rating changes.
Watchlist and outlook will not be considered here, as this study is focused on identifying the determinants that affect the credit rating. According to Figures 1 and 2 In an attempt to capture the most relevant determinants, this paper examines the criteria that form the basis of the sovereign ratings. From the overall group of determinants, nine are considered key for assigning ratings. Determinants that are relevant in forecasting credit ratings are described detail in  Inflation: Inflation is given as average growth in consumer prices average over the last 3 years (%). Inflation is observed over the last three years because it is a macro variable that is considered volatile to the extent that observing it for a single year can be misleading. Inflation is most commonly defined as the rise in the general price level. Purchasing power declines when the general level of prices for goods and services rises. Factors that affect aggregate supply and demand also affect inflation.
A high inflation rate is indicator of economy in which the demand for goods and services exceeds productive capacity, thereby exerting greater price pressures. High inflation can cause political instability because of public discontent. There must be an inverse relationship between inflation and credit ratings. The Laspeyres formula is generally used to produce the inflation indicator.
Investments: Investments are expressed as a ratio of total investment in current local currency to GDP   vices, the quality of the civil service and the extent to which it is independent of political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. A higher government effectiveness score will also result in a better credit rating.

Results and discussion
Financial indicators cannot determine credit ratings when considered individually. It is necessary to observe them as a group to determine the economic situation of a country and its future potential. According to Bissoondoyal-Bheenick (2005), economic variables do not have the same significance for low-ranking countries as they do for high-ranking countries. In this study, countries have been classified into two groups with respect to GNI per capita: up to $12,615 and over $12,615.
This section analyzes the individual impact and significance of the variables described above. FC and LC credit ratings were collected from the Standard & Poor's official website. Credit ratings agencies use information from the past to describe the present and the future status of a country, corporation or security.
Economic determinants were collected to calculate the relationships among them and between these determinants and the assigned ratings. Table 5 shows the regression results for significant variables employed in the allocation of FC and LC ratings. Based on the results of the analysis reported in Table 5, that the coefficient of determination between the FC and LC credit rating with respect to the nine independent variables is 0.83, meaning that 83% of the variation in the dependent variable is caused by variations in the selected independent variables. All FC and LC variables are statistically significant at the 1% or 5% level.
According to the classical linear regression model: Y i = α+β 1 X 1 +β 2 X 2 -β 3 X 3 -β 4 X 4 + β 5 X 5 -β 6 X 6 -β 7 X 7 -β 8 X 8 + -β 9 X 9 + e i (1) Credit ratings models are calculated as follows:      The second test was for the normality of residuals, and we used the Jarque-Bera test for thus purpose. The next test checks for the presence of heteroskedasticity. If heteroskedasticity is not present, the data are homoskedastic. As Table 6 shows, Prob. Chi-Square (for the FCR data) is larger than α (0.29>0.05), and thus it can be concluded that there is no heteroskedasticity and the data are homoskedastic. Prob. Chi-Square for the LCR data is larger than α (0.50>0.05), and hence it can be concluded that there is no heteroskedasticity; the data are also homoskedastic in the LCR model.
Multicollinearity is present if two regression variables are dependent or approximately linearly dependent (Bahovec & Erjavec, 2009). The standard indicator of multicollinearity is the variance inflation factor (VIF). The results of the multicollinearity test are shown in tables 7 and 8: The VIF for each of explanatory variables is quite small, suggesting that the null hypothesis, which assumes the presence of multicollinearity, should be reject for the FCR and LCR models. When the empirical VIF values are less than five (VIF<5), it can be concluded that there is no multicollinearity in the observed sample.

Forecasting
In Table 9, the ratings are calculated according to formulas (2) and (3) Therefore, these two models are very precise.
After evaluating credit rating models within the sample, we estimate FC-and LC-denominated credit ratings for 38 countries out of the sample. The selected countries represent unrated countries. As   Issue 3 367-384 2015 with high predictive power; the second aim was to demonstrate that unrated countries are not necessarily at the bottom of rating scale. After testing a number of possible determinants, nine were found to have a significant impact on credit ratings (at confidence levels of 95% and 99%). These determinants are: HIPC (dummy variable), external debt, GDP per capita, government deficit/surplus, inflation, investments, legal rights, total reserves and government effectiveness. Using a sample of 50 countries, two rating prediction models were constructed to estimate ratings: an FC model and a LC model. Both models have large coefficients of determination (0.83), and after comparing the within-sample ratings estimates, 40% were correct for FC and 30% for LC; 40% were +/-1 notch from the actual rating for FC and 48% were +/-1 notch from the actual rating for LC. Only 2% were +/-3 notches from the actual rating for FC; in the LC model only 4% of estimates were +/-3 notches from the actual ratings and only 2% in were by +/-4 notches from the actual rating. This analysis confirms the high precision of these models. Note that all of the assumptions required for a linear regression model are satisfied.
Among the unrated countries, the best estimated rating is exhibited by Mauritius: A in LC and A-in FC.
There are five countries that have investment-grade FC ratings. These countries are: Mauritius, Dominica, St.
Lucia, Tonga and Algeria. According to the FC rating, 20 countries would classify as speculative or highly speculative. Only 7 countries are considered to have high or very high credit risk. The secondary hypothesis of this paper is also confirmed-unrated low-and middle-income countries need not occupy the bottom of the ratings scale. It is important to further emphasize that the credit rating agencies, beyond the objective factors, also include subjective factors, which makes it difficult to exactly quantify ratings. For this reason, it is not possible to construct a model capable of reflecting current credit ratings with 100% accuracy.