International Business Project: Marketing Sovereign Bonds in Russia Completed by University of Abstract This paper builds and offers the plan for providing financial advisory services, mainly of advising clients interested in investing in sovereign bond market, in Russia. The “Brukovski and Partners” company (a fictional one), based in the city of New York, has been providing financial advisory services in the US for the last 22 years. After facing an increased competition on the domestic market and after carefully analyzing the financial markets in Eastern Europe and Asia, “Brukovski and Partners” decided to set up its consulting operations in Russia. The company aims for high returns that are connected with the high growth of the Russian bond market, especially the sovereign bond one. This work provides an in-depth financial, economic, and market analysis -- including the determinants of changes in the yield spread (credit spread) -- of sovereign bonds in the emerging market of Russia over a period of 2000 – 2005 and offers the most suitable entry strategy for “Brukovski and Partners”. The paper also examines macroeconomic situation of Russia and according to such analysis constructs regression model taking into account the specific features of this country. Estimations of the model show that the most significant factors in determining the strategy of advising clients on investing in the sovereign bond market in Russia are credit spreads, affected by oil price, US inflation and Libor, which are consistent with the macroeconomic analysis provided in this paper. Contents 1. Executive Summary ....................................................................................................................4 2. Theoretical determinants of Sovereign spreads in Russia….......................................................6 3. Economic Outlook of Russia ....................................................................................................10 4. Macroeconomic situation in Russia ..........................................................................................11 5. Regression Analysis...................................................................................................................14 6. Specification of macroeconomic variables ...............................................................................16 7. Econometric methodology ........................................................................................................18 8. Regression analysis of Russia 2030 sovereign bond.................................................................19 9. Conclusion.................................................................................................................................26 Bibliographical References............................................................................................................29 APPENDIX A................................................................................................................................32 APPENDIX B ...............................................................................................................................34 APPENDIX C …...........................................................................................................................46 Executive Summary The past decade has seen a substantial growth in international bond issuance by emerging market economies. As appears, it is now one of the fastest growing sources of external financing. Exhibit 1 Net international bond issuance by EMEs. Source: World Bank According to World Bank data, foreign investors bought more than US$300 billion (net) of emerging markets bonds during the 1990s, compared with US$20 billion during the 1980s. In addition the terms of these economies have improved due to increased investor participation in emerging market securities. However, emerging market sovereign spreads also have experienced turbulence. Therefore, these movements have created attractive opportunities over time, both for buyers as well as for sellers of emerging markets economies such as Russian in our case. This paper examines macroeconomic factors that were not a main focus of empirical studies, such as inflation and unemployment rates for US and Russia. Our main finding is that spreads on sovereign bonds broadly reflect fundamentals, but non-fundamental factors also take place in those spreads. The models point out to a set of variables which represent three categories: liquidity, solvency and external shocks. The regression analysis can explain 35% of credit spread for Russia. During our investigation we find that selected macroeconomic fundamentals are highly cross-correlated. We therefore rerun regressions in order to find optimal variables that are least correlated in order to include in our model. This paper is designed as follows: Section 1 discusses various measures of emerging market sovereign bonds’ specifics, which have been examined in different empirical studies. Section 2 presents the strategy for working on the Russian sovereign bond market from theoretical perspective. What is more, this section introduces macroeconomic overview for Russia, outlying specific features and conditions for this country. Section 3 comments regression analysis results and offers the best as well as most suitable entry strategy. The purpose of examining the determinants of corporate and sovereign bond spreads is to have a broader look at macroeconomic variables which were employed in different studies and select the most optimal set. Particular, recent studies on corporate credit spreads such as Elton et al. (2001) and Collin-Dufresne et al. (2000) found that a large part of credit spreads of corporate bonds cannot be explained by changes in the expected default risk of the corporation. A study was conducted on Russian bond market. Duffie D. et al. (2001) that constructs a model of the term structure of credit spreads of the Russian Ministry of Finance (MinFin) over a sample period encompassing the default on domestic Russian GKO bonds in August, 1998 (Duffie D. [2001]). The model is parameterized with the composite credit and liquidity spread of one of the Russian MinFins, serving as a benchmark. The selected data does not include the 1998 Russian default. According to their model, the risks of Russian default should be influenced by government balance-sheet variables and related variables that influence internal and external balances. One such variable is the price of Brent oil. Another relevant credit-related variable is the total level of foreign currency reserves, minus gold, held by the Russian central bank. Duffie D. et al. (2001) estimate the joint distribution of the term structure of LIBOR swap and Russian bond yields. Promised future cash flows are discounted using a default-adjusted rate equal to the risk free rate plus a credit spread which reflects the probability and magnitude of a credit event. They allow for different discount factors since sovereign bonds usually do not have crossdefault clauses. Theoretical determinants of Sovereign spreads in Russia It is important to note that yield spreads of sovereign bonds in emerging markets are important indicators of financial stability of a country. Primarily these indicators are used as a measure of the default risk that a country might bear and assess emerging markets external financing conditions. Those spreads are influenced by a large number of determinants – credit risks, liquidity risks and market risk. Emerging markets countries are seem to be more likely to default on their debt; consequently investors require additional compensation to hold those risky bonds. So this compensation is usually measured by the yield spread. Duffie D. (2001) suggests that credit spread on sovereign bonds accommodates to: (i) default or repudiation; (ii) restructuring or renegotiation – the sovereign announces that it will stop making payments on its debt; and (iii) a “regime switch” such as change of government or the default of another sovereign bond that changes the perceived risk of future defaults. According to Kamin S. et al. (1999) credit spread is defined as the promised annualized yield on the emerging market debt instrument less the benchmark yield, the annualized yield on an industrial country government bond of the same currency denomination and maturity as the emerging market instrument. More formally, Spread = I - Ibm, (2) I is the annualized yield on emerging market debt instrument. Ibm is the annualized yield on an industrial country government bond of the same currency and maturity. It is obvious that when an investor purchases a bond, he or she exposed to a number of risks. For example credit risk perceived as a possibility that borrowers will default on their obligations. Certainly this risk depends on the specific features of the issue, for example for sovereign bonds this includes the ability and also willingness to repay the debt. Another type of risk which investors face, when they are holding fixed income securities is a market risk. As it is quite common for emerging market bond prices to be fluctuated both risks are related to each other. Moreover, the changes in preferences of investors to hold a particular bond may also give rise to market risk. In addition, there is a chance that investors may face a liquidity risk, i.e. the possibility to liquidate their portfolios at a rapid pace. From the exhibit 2 we can observe that spot yield curve at which a particular obligation’s cash flow is discounted is modeled as a benchmark curve plus a spread for credit risk, plus a spread for liquidity risk, plus a spread for imbedded options, etc.: Exhibit 2. An example of credit spread using as a benchmark libor/swap curve Exhibit 2 is an example of credit spread using as a benchmark libor/swap curve A spot yield curve is a simply a curve that shows spot yield values as a function of time- to-maturity and constructed based on actively traded securities (Questa G. [2004]). A benchmark curve might be the securities issued by the US department of the Treasury, which backed by the US government. Hence, investors throughout the world perceive them as having no credit risk. The minimum interest rate that investor wants is called the base interest rate or benchmark interest rate, which investors will require for investing in non-Treasury securities. This rate is the yield to maturity offered on a comparable maturity Treasury security that recently issued or on the run (Fabozzi F. [2000]). Interest rates on non-Treasury securities are traded at a spread to a particular on-the-run Treasury security. This spread is called as a risk premium, which represents the additional risks the investor faces when he or she purchases a security that is not issued by US government. We can depict those interest rates, which offered on a non-Treasury security as: Base interest rate + spread, or Base interest rate + risk premium The chart below presents a spread between US 2030 Treasury bond, Russia 2030 and Turkey 2030 sovereign bonds. Visual inspection of the graph suggests that throughout 2000-2004 spread between emerging market sovereign bonds and US treasury has dramatically narrowed. From this graph we can observe the trend of both economies. In Russian case due to consequences of Russian default in 1998 the spread is quite wide in 2000. However, this magnitude in the spread started to narrow at a slow pace and accelerated to decline from March 2001. The Turkish spread reflects a financial crisis, which deteriorates Turkish economy in 2001 and widened the spread at that time. As a result of favorable macroeconomic performance and steadier policy implementation, the spread saw a decline during 2002 and 2004. Economic outlook of Russia It is crucial to maintain and control financial stability as it plays an important role in a sense of the increased globalization of financial markets. The corporate scandals in the United States and Italy in 2003 showed how destructive their consequences could be and how important it is for the regulators to monitor financial stability. Moreover, world financial markets have become increasingly intertwined and susceptible to contagion effects, which highlight the importance to look at the economy performance for both countries. The Russian economy is mostly affected by following factors (Alfa bank [2001]): 1. Macroeconomic situation. Dynamic economic growth, moderate foreign and domestic debt, low inflation, a balanced budget, an effective balance of payments structure and a stable national currency are the macroeconomic fundamentals which are used as a main tool of assessing general economic situation on the market. 2. Oil price dynamic. A sharp drop in oil prices might reduce the investment attractiveness of Russian oil stocks and undermine the country’s economic stability. 3. World economy or external shocks.

On the whole, 2003 was a year of financial stability in Russia (The central bank of Russia [2003]). The country successfully over-passed the peak of its foreign debt payments, significantly increased its international reserves and accelerated GDP growth. Diminished political uncertainty has combined with expectations that structural reforms will be accelerated to boost asset prices sharply in Russia. With government finances and the current account set to remain in surplus, external debt spreads look likely to narrow further over the medium term as government debt continues to decline and foreign exchange reserves rise (MNB [2003]). Macroeconomic situation in Russia In the past 4 years, the Russian economy has developed under the effect of the factors created during the 1998 crisis. After the crisis, Russia saw a steady recover. Low exchange rate of the ruble and prices of the services determined low production costs in the key industries and stimulated the large scale replacement of imports by domestically manufactured goods. A positive influence on the stabilization of the economy was affected by the favorable situation in foreign markets, the improvement of the social and political situation in the country. Thus, GDP growth rate was maintained by further growth in exports and investments from 2000 to 2002. High oil prices and strong credit expansion helped accelerate output growth in 2003. Real GDP rose a reported 6.3 percent in 2004 (see table 1). The principal sources of GDP growth in 2003 were greater consumption in the household sector and growth in net exports and fixed capital investment. From this table we can observe that GDP growth experienced a downward trend trough the years 2000 and 2002; however, it saw an increase from 4.7% to 7.3% and then a slight fall to 6.3% in 2003 and 2004 respectively. The forecast of GDP accounts for 7.5% in 2006, which was calculated by Goldman Sachs (GS 2004). ANNUAL FIGURES 2000 2001 2002 2003 2004 Real GDP growth (% yoy) 10 5.1 4.7 7.3 6.3 CPI (% yoy) 20.2 18.6 15.1 12 11 Table 1. Russian Real GDP growth and Consumer Price Index. Source: Bloomberg The consumer price index stood at 10% in 2004 since ruble appreciation against the dollar helped slow 12-month consumer price inflation from 12 percent a year before. From table 1 we can observe that CPI has been reduced dramatically from 20.2% to 12% between 2000 and 2003. It accounted for 15.1% in 2002, exceeding the target range of 12-14% for that year. Terms-of-trade gains and a surge in export volumes helped increase the current account surplus to about $35.8 billion in 2003 from $29 billion in 2002 (see table 2 and Appendix A). Russian current account dropped considerably from $46.8 billion to $29billion throughout 2000 and 2002 however, it saw a recovery in the next year. Current account US$ million 2000 2001 2002 2003 2004 Q1 11598 11678 6421 11537 13006 Q2 11812 8959 7674 8192 9500 Q3 10551 7096 7183 7396 8400 Q4 12878 6202 7838 8720 9100 Total 46839 33935 29116 35845 40006 Table 2. Russian Current Account. Source: Bloomberg The terms of Russia’s trade with foreign countries improved significantly in 2000- 2004 owing to rapid growth in the prices of Russian exports. Exports of goods and services remained the main source of foreign currency for Russia. In 2003, Russian exports expanded 24% year on year to an estimated $136 billion (see table 1 Appendix A). Growth in exports in 2003 resulted not only from a rise in contract (export) prices of goods, but also the expansion of export volumes (www.cbr.ru). Growth in the prices of major Russian exports had a favorable effect on the country’s balance of payments (IIF [2004]). The outsized current account surplus was accompanied by a sharp increase in capital inflows after Moody’s upgraded Russia to an investment grade rating in early October in 2003. The result was a surge in foreign exchange reserves of $13 billion during the final quarter of 2003. Reserves rose another $10 billion through February to a record $82 billion, or 11 months’ imports, before declining again to $80 billion by early March, partly because of external debt repayments. Oil price dynamic: the Russian economy is heavily dependent on raw materials and, consequently, on the world energy prices. As Russia continues to ease its foreign currency controls, a sharp fall in oil prices may provoke a crisis in the field of capital flows and foreign debt servicing by the private sector. The overall situation on the market is maintained to be relatively constant as high oil prices (see exhibit 3) helped the government to reduce debt and increase its fiscal reserves. The price of Brent crude rose 15.4% to $28.9 per barrel due to restrictive policy pursued by the OPEC member countries and the military operation in Iraq, which were the decisive factors of growth in oil prices in 2003. The analysis shows further growth in the oil price in 2004 which positively affect on country’s stability. Exhibit 4 Brent oil monthly data in dollars per barrel from 2000-2004. Source: Bloomberg World stability or external shocks: the analysis of the Russian financial sector shows that the Russian economy is growing at rapid rates, the principal source of the money supply is Bank of Russia purchases of foreign exchange, while the main threat to financial stability (concerning banking sector and securities market) comes from a sharp reduction in the flow of foreign currency to the domestic foreign exchange market and changes in the movement of foreign capital. Thus, vulnerability to external economic shocks remains the most formidable threat to financial stability in Russia. Regression Analysis We provide data on Russia and Turkey to offer an additional point of reference for the emerging markets and to create a more comprehensible picture of the most appropriate strategy to be used in Russia. Firstly, we assemble monthly yields of sovereign bonds for Russia 2030 and Turkey 2030. Russia 2030 is a 30 year sovereign 5.00% coupon bond issued in March 2030 (US dollars). Type of maturity is sinking fund with its ratings given by Moody’s and S&P of Baa3 and BB+ respectively. Turkey 2030 is a 30 year sovereign 11.875% coupon bond issued in February 2030 (US dollars), with normal type of maturity. The ratings given by Moody’s and S&P are B1 and Bb respectively (sours: Bloomberg). None of these bonds is callable. We chose these bonds as they are pretended to be one of the most liquid and actively traded on the emerging markets. Moreover, those markets are experiencing economic growth and therefore are presented to be attractive markets in terms of liquidity. Thus majority of investment banks such as Goldman Sachs, Lehman Brothers, Barclays Capital and Morgan Stanley are focusing on Russian and Turkish markets and those sovereign bonds. Secondly, we employed US 2030 Treasury bond as the benchmark, starting from January 2000 with 5.375 % fixed coupon payment and normal type of maturity (source: Bloomberg). The yield difference between those bonds and US bond is defined as “spread”. Before starting to analyze the output of the regressions, we shall examine data series of Russian sovereign bond spreads depicted below. During the examined period, the mean of Russian 2030 spread is about 612 basis points and standard deviation is 291.23 basis points (see Appendix B table 1.1.). The graph depicted below illustrates relatively stable downward trend trough the years 2000 and 2004. Exhibit 5. Time series behaviour of Russian 2030 sovereign bond spread over US 2030 treasury bond during the time interval 2000 to 2004. Source: Datastream The Russian 2030 sovereign bond spread was heading to the top position of emerging market league table, since it reached maximum point at 1081.72 basis points in 2000. We can observe that the magnitude of the spread has been maintaining during first two years. “Erik Nielsen, director of new European markets economic research at Goldman Sachs, commented it: A rally of this magnitude cannot continue for ever, but this does not mean spreads will widen from present levels" (Ostrovsky A. 2002). Indeed in this period the Russian economy continued to grow at about 4 per cent and inflation was approaching single digits After 2002 the spread has narrowed considerably, highlighting the fact of the Russian stability overall. Specification of macroeconomic variables In our model described below we employed a set of variables, which we expect to affect Russian spreads: %uDBC0%uDCBE Real exchange rate. Currency overvaluation tends to be associated with currency and balance of payments crises, which often leads to debt servicing difficulties. Moreover, an increase in real exchange rate will cause some lose in an investor’s gain while he or she is converting money to USD from the emerging market’s own currency (Copland S. [2000]). We expect this variable to carry positive sign. %uDBC0%uDCBE Industrial production growth index. It is also important to include this parameter as a determinant. We suggest that investors are more interested in recent information, which can reflect the economic situation dynamically. In this sense we employ IP index, which is reported monthly and echoes the financial situation in dynamic overall rather then GDP. This variable is expected to have a negative sign. %uDBC0%uDCBE Global liquidity (LIBOR). An increase in this variable also should increase spreads (indicating a positive sign in the model) in emerging market countries since it would limit the attractiveness of investments in emerging economies (relative to developed), and reduce capital flows to those markets. %uDBC0%uDCBE GDP. A relatively high rate of GDP suggests that a country’s existing debt burden will become easier to service over time. First a higher, permanent GDP implies higher domestic savings and investment, hence a lower critical capital stock ratio to meet a given borrowing program. This should be reflected through lower default probabilities and spreads, indicating a negative sign in the model. %uDBC0%uDCBE Inflation. A high rate of inflation points to structural problems in the government’s finances. When a government unable or unwilling to pay for current budgetary expenses through taxes or debt issuance, it must resort to inflationary money finance. Hence, inflation is expected to have a positive sign in the model. %uDBC0%uDCBE Budget balance: A large budget deficit shows that the government has a relatively weak capacity to raise additional revenue and to service debt costs. This causes an increase in the spread as it strengths the possibility of default of the emerging economy. Thus, Budget balance should be associated with a negative sign. %uDBC0%uDCBE Balance of trade: A large trade deficit indicates that the public and private sectors together rely heavily on funds from abroad. Trade deficits that result in growth over time soon become unsustainable. Trade deficits are balanced with higher, permanent gross capital inflows. Those inflows reduce the degree of exposure by bringing down the critical capital stock ratio and create a strong demand for the sovereign bonds, which in turn leads to lower spreads. %uDBC0%uDCBE International reserves: The level of international reserves acts as a guarantee in the payment of debts for emerging economies. So that the higher the reserve level, the lower the probability of default. We expect to have a negative sign in the model for this variable. %uDBC0%uDCBE The price of oil: We suggest both countries to be depended on the level of world energy prices, especially for Russian economy and have a positive sign. %uDBC0%uDCBE US inflation rate: The effect of US inflation should be considered via US interest rate as inflation has an impact while Federal Reserve is defining the interest rates. When interest rate is low, consumption level in US increases which in turn causes prices and inflation increase. According to this opposite relation between inflation and US rates, a positive sign is expected for US inflation in the model. %uDBC0%uDCBE Real USTB 30 years bond yield. In fact we know from the international finance theory that an increase in the 30y-bond interest rate will attract more investors, therefore the supply of loanable funds for emerging countries would diminish and, therefore, the country risk would increase. On the other hand, in periods of extreme crisis the flight-toquality effect takes place.

A more risk-averse behaviour can push US bond rates down, consequently this will lead to an excess demand of USTB bonds and increase country risk of emerging markets. %uDBC0%uDCBE Current Account This factor includes transactions that either contribute to national income or involve in spending. A deficit implies that a country is a net borrower so that the increase of the current account deficit leads to an increase of the probability of default immediately affecting on the spread. %uDBC0%uDCBE External Debt A higher debt burden should correspond to a higher risk of default. The weight of the burden increases as a country’s foreign currency debt rises. This variable has a positive influence on the spread between the yields of the sovereign and US Treasury bond. We also think that it is appropriate to include a number of ratios such as Trade balance to GDP, International reserves to GDP, Budget balance to GDP, Current account to GDP and Debt to in order to capture significant variables that explain the credit spreads. Econometric methodology It is worth to point out that we took seasonal adjusted data for each macroeconomic, liquidity and solvency variables in order to avoid seasonal effects in the regressions. Moreover, as some variables are reported quarterly such as GDP and External debt we used a linear interpolation method. During interpolation stage we derive monthly data for GDP and External Debt. The reason of doing that is because raw monthly data can involve issues of seasonality and volatility that may seriously bias long-run estimates. The OLS Method is used for investigating the interaction between endogenous (spread) and exogenous (macroeconomic) variables. Monthly data are used in the period from February 2000 to May 2004. As we have seen from previous studies (Cantor and Packer [1996], Min H. [1998], Edwards [1985], Duffie at el [2001]) bond yields are related to macroeconomic fundamentals and business condition, and hence time series which may be able to capture those relations are used in our investigation. The common problems mentioned in the previous researches while constructing such a model are: • The number of variables that are considered to affect the spread is huge. • Some variables are highly correlated with each other. • It is difficult to extract monthly data for macroeconomic variables as they are issued quarterly or even annually and making interpolations on the data makes the model biased. Considering these problems, we searched for every related variable for explaining spreads listed in chapter 3.2. Moreover, we also derived new factors by getting a proportion of those factors to GDP. Regression analysis of Russia 2030 sovereign bond In getting the final result, we tried all variables described earlier in the paper and also checked if there was a presence of multi-collenearity. Thus we exclude those variables, which tend to be highly correlated and arrived to the final model presented below: SPREAD = f (INTERNATIONAL RESERVES, OIL, LIBOR, US INFLAT) This model is the starting point of all regressions. The result of the regression is presented in table 19 (see Appendix B) and our regression model can be expressed as: SPREAD=0.0054*Internreserves-15.20*Oil+83.59*Libor+12567.63*USinflat According to the results the p-value of the variables turns out to be significant at 5% confidence level. Moreover the explanatory power is about 93%. However, considering the fact that variables might be non-stationary, spurious regression is highly possible in the model. Therefore, before making any inference we should test our variables for presence of stationarity. Broadly speaking, a time series is stationary if there is no systematic change in mean (no trend), if there is no systematic change in variance, and if periodic variations have been removed (Chatfield C. [2000]). The use of non-stationary data can lead to spurious regressions i.e. if standard regressions techniques are employed to this data, this would lead to misleading results. In fact, many time series, particularly macroeconomic time series, are non-stationary (Hill C. (2001:335-338)). Thus, stationarity of the variables should be searched. Considering above statement, the first step in any time series analysis is to plot the observations against time. Those graphs of each macroeconomic variable for each country are presented in Appendix B. If we look at the Russian macroeconomic time series of our model plotted on the graphs we can see different trends that confirm non-stationary process. The International reserve time series plotted on the graph 7.2 (see Appendix B) shows peaks at several frequencies and rarely crosses its mean. Chart 9.2 (see Appendix B) plots the sample of Libor index and shows a significant time trend. The US inflation index illustrated in table 10.2 (see Appendix B) shows visible spikes, these spikes however, are minor in terms of crossing its mean value. The time series of oil price depicted in chart 6.2 (see Appendix B) shows a time trend in a long run. i) Stationarity test The best way to test for staionarity is to conduct a test for a unit root. The Augmented Dickey-Fuller (ADF) tests are carried out. The objective of the test is to examine the null hypothesis that f = 1 i.e. series contain a unit root. diagnostic test is conducted for each variable in order to check if series are stationary I(0). The results are reported in the tables 20-24. (see Appendix B). After testing these entire variables, ADF test statistic is found to be greater than 5% critical value, meaning that all these variables are nonsationary. While tests are conducted, a problem arises in determining the appropriate number of lags of each variable. Often, financial theory has a little say on what is an appropriate lag to choose. It is important to use an optimal number of lags of the dependent variable in the test regression, as including too few will not remove all of the autocorrelation and using too many will increase the coefficient standard errors. Therefore the absolute values of the test statistics will be reduced, and hence will reduce the power of the test. Brooks C. in his book “Introduction in econometrics for finance” describes two methods of choosing an optimal lag length. First, we should take into account the frequency of the data. Second an information criterion can be used. In the first method, Brooks C. suggests to use 12 lags for monthly data. According to the second method, we checked each lag of those variables in terms of Schwartz criteria and chose the lag, which performs the minimum. Although the appropriate number of lags is changing from variable to variable, in general it varies between 10 and 12. It is important to note that ADF test cannot yield very powerful results, as it cannot distinguish between a unit root process and a near unit root process. This suggests looking at possible cointegration. ii) Testing for cointegration The main concept of cointegration is that two (or more) series might be individually be nonstationary, but a linear combination of them might be stationary, therefore the Engle-Granger 2 step method should be conducted. First we estimated cointegrating regression by OLS and tested whether the first difference of each variable is stationary. The results seen in tables 25-30 (see Appendix) found to be stationary, therefore second step can be taken. The estimation results (see Appendix B table 31) show three significant outputs: Libor, Oil prices and US inflation at 5% significance level in explaining credit spreads on Russia 2030 sovereign bond. This consistent with Alfa Bank findings (Alfa Bank [2000]) that the most common factors affecting on the Russian economy is the level of oil prices, conditions in US economy and external shocks. iii) ARCH test In order to observe the residuals adequacy we perform a couple of tests shown in table 52 (see Appendix B). Firstly, we checked whether there is an ARCH process or not in the residuals and examined if residuals are influenced by the previous lags of residuals. We performed an ARCH test. As seen the probability of F statistic is greater than 5% confidence level indicating no ARCH process in the model. iv) Heteroscedasticity test In order to detect heteroscedasticity we run the White’s general heteroscedasticity test. We run an auxiliary regression of the squared residuals, from our original regression, against a constant, all the original regressors, their squared values and their cross-products. In our case (see Appendix B table 32) the p-value for the Ftest is 0.19, i.e. there is no heteroscedasticity. v) Autocorrelation test According to DW statistics we conclude that there is no first-order autocorrelation as DW statistic is around 2 (2.11). In order to investigate higher-order autocorrelation we run a BG test (see Appendix B table 33). We performed the test with different lag lengths. All of them give the same result of no serial correlation in the model, as F statistic probability is about 0.13 which is obviously higher than critical value of 5%. vi) Test for presence of non-normality The histogram (see Appendix B chart 34) shows more or less a normal distribution, a part from the presence of certain outliers on the left tail of the distribution. A deeper investigation is needed. We run a JB test, an asymptotic test of normality based on skewness and kurtosis. In our case skewness is -0.02 and kurtosis is 2.37 pretty close to the values of normality. In order to be more rigorous we check the probability of the JB value. The p-value associated with the JB value is 66%, asymptotically we cannot reject the null hypothesis of normality. vii) Checking for multi-colleniarity of final model It seems reasonable to look at the presence of multi-colleniarity. One of implicit assumptions when using OLS estimation method is that explanatory variables are not correlated with one another, it is vital to identify if there is a presence of multi-colliniarity The correlation matrix for explanatory variables is presented below: DLIBOR DOIL DUSINF DINTRESERVES DLIBOR 1.000 0.166 -0.125 0.121 DOIL 0.166 1.000 0.446 0.093 DUSINF -0.125 0.446 1.000 0.076 DINTRESERVES 0.121 0.093 0.076 1.000 Table 8. Correlation matrix of macroeconomic variables during the time interval 2000-2004 (Russia) The results show that all selected variables do not have high correlation. viii) Testing for functional form A further implicit assumption of the classical linear regression model is that the appropriate ‘functional form’ is linear. This means that the appropriate model is assumed to be linear in the parameters, and the relationship is presented by a straight line (Brooks C. [2002]). To check the functional form of the model, we use Ramsey RESET test. The result is shown in table 53 (see Appendix B). According to the critical value, the model is correctly specified. ix) Results Table 31 (see Appendix B) illustrates the regression output after tests conducted for autocorrelation, and heteroscedasticity. Our research work identifies several important factors, which affect on Russia 2030 sovereign bond credit spread. First of all DOIL (the change in the variable ‘oil price’) appeared to be significant with its p-value of 0.02. Moreover, DLIBOR (changes in Libor) and DUSINF (changes in US inflation rate) also affect on the spread, with their p-values of 0.003 and 0.0001 respectively. However, DINTRESERVES (changes in international reserves) turns out to be insignificant as the p-value is 0.5. Thus, model explains spread changes with Libor, oil price and US inflation changes. In the model, US inflation and LIBOR are external shocks, which have significant impacts on Russian economy. Meanwhile although Oil price is an external factor, it can also be considered as a liquidity variable since it has a huge effect on Russian balance of payments and current account. Those significant variables explain 35% of the changes in the spread. Considering the regression as a financial model, explanatory level of the regression is satisfactory. Intercept (B) is the error term of the previous model run in the 1st step of Engle Granger test. To clarify the effects of each dependent variable on change in credit spread, the table below is used: Variable Coefficient t-statistic Impact on spread from 1% increase Intercept 0.36 3.29 Change in LIBOR 111.00 3.17 111 Change in Oil price -8.04 -3.35 -8 Change in US inflation 11.564.52 4.41 115 Table 9. The final result of estimated model for Russia 2030 sovereign bond spread According to the table 9, 1% increase in LIBOR will cause an increase in spread by 111 basis points and 1% increase in US inflation will cause an increase in spread by 115 basis points. Additionally a 1% change in Oil price will have a negative impact on the spread by 8 basis points. Since the changes in spreads are measured in absolute terms, care has to be taken when interpreting the coefficients.

For example the coefficients for US inflation and for LIBOR look very different as they have very similar effect on the spread changes. This is due to the unit difference used for measuring LIBOR and US inflation. LIBOR is presented in percentage (such as 6%) while US inflation is presented in decimals (such as 0.003). Moreover our expectations are consistent with significant level of US inflation rate and LIBOR, since economic situation in the world and in US affects on Russian economy’s stability and sovereign credit spread. In addition the model captures the external shock variable of oil price. The significance of oil price makes a strong sense and as it was outlined before in Russian economic overview, oil prices play a vital role in Russian economy. In our model for Russia, macroeconomic indicators have low significant impacts on spread due to: %uDBC0%uDCBE LIBOR is highly correlated with Debt to Export Ratio (97%). In order to avoid multi-collinierity in the model, we eliminated “Debt to Exports” ratio from the model. %uDBC0%uDCBE There is no more incremental information in the publicly available macroeconomic variables that is useful for predicting the yield spread. This is because price movements and other economic variables do not often reflect changes in the country’s ability to pay. In addition, it should be noted the theoretical signs of variables converge with our estimations. Conclusion This paper proposes the most effective strategy for entering Russian market for providing financial advisory services, mainly the services concerning sovereign bond market in this country. The main finding for Russia is that US inflation rate, Libor and Oil price are significant determinants of the local market. All these factors belong to the external fundamentals, highlighting the fact that Russia is heavily dependent on world economic fundamentals, especially in case of oil price, which is consistent with our macroeconomic analysis. The explanatory power of the Russian model is 35%. Another important aspect of our investigation is that solvency variable, which measured by the international reserves, was found to be insignificant in explaining credit spread for Russia. This might be interpreted, as there is no incremental information in this publicly available macroeconomic variable that is useful for predicting the yield spread. Furthermore, after profound analysis of Russian economy, one of the lessons for this economy seeking greater access to the international bond market with lower spreads seems clear – it is the sound management of macroeconomic fundamentals. Let us sum up all of the main points that we have made on the issue by saying that there some considerations that cannot be avoided when starting the business in Russia: the question of needing a partner, the current economic factors, considering alternate locations, and developing a tax strategy as well as all of the economy and risk-related factors that were outlined in this work. When debating whether or not a partner is needed or wanted, you need to know if you’re going to need additional equity as well as sharing the risk of failure. For these reasons, a partnership seems to be a great idea, but there are also many cons that should be recognized. Having too many partners can alter the ease of decision-making, shared liability can cause obvious problems, and sharing profits means less for you. Added to this, getting out of a partnership can be very difficult in Russia. Evaluating the current economic factors in this country simply means to know what you are getting into. Know how to make and sell the product efficiently and in a service industry, be sure to know the current and correct way things are done-sometimes they are not one in the same. Location is key. Location of the target can be a major determinate in both the financing of the deal and probable success in managing the business after closing. There’s no sense spending time, effort, and money on a target located in the wrong place. Along with this, the personal strife of having to travel a great distance to get to work can be very frustrating. So, be sure that the location of your potential business is profitable in every way. One the greatest minds of the 20th century, Albert Einstein, once said that tax is the most difficult thing in the world to understand. Unfortunately, with the ever-changing laws, that problem gets worse every year. This means that you should have knowledge of the current tax laws in Russia. You will have a unique opportunity to make decisions on exactly how much money will change hands, and how it will be allocated on the payment schedule. Maximizing profit for both you and the seller can only be done through proper knowledge of tax law, if you are not comfortable handling this alone, a consultant might not bad a bad idea. After all of the above is settled, the next thing to figure is the amount of initial income is required. Not only the income required to purchase the entity (which will be elaborated upon later), but also the amount of money that you need to survive for the years to come. For instance, if you need $100,000, then don't look at smaller companies which can only yield $30,000. Most of the aspects have been covered in the above, but there are many more little things that may need to be considered depending on the uniqueness of the business. There are many helpful sources, which will help an individual obtain information on how he/she should start their own business plans. The better organized an individual goes into a business start up the better prepared they will be for the many complicated decisions that will arise in the future. Bibliographical References Alfa Bank Year 2001 outlook (Moscow 2001) Brooks C. (2002) Introductory econometrics for finance (Cambridge University press) Cantor, R. & Packer, F. (1996) Determinants and impact of sovereign credit ratings (FRBNY Economic Policy review) pp. 37-54 Central Bank of the Republic of Turkey Developments in the Turkish Economy www.tcmb.gov.tr Copland S. (2000) Exchange rates and international finance 3d addition Pearson Education Limited, Edinburgh Duffie, D .Pedersen, L. & Singleton, K. (2001) Modeling sovereign yield spreads: A case study of Russian debt pp.1-50 Dufresne, C. Goldstein, P. & Spancer, J. (2001) The determinants of credit spread changes (The journal of finance) pp. 2177-2208 Edwards, S. (1985) The pricing of bonds and bank loans in international markets: Analysis of developing countries’ foreign borrowing (NBER Working paper N1689) pp.1-42 Federal Deposit Insurance Corporation (1997) History of the Eighties. Lessons for the Future (FDIC Public Information Center) pp. 191-210 Elton, E. and others (2001) Explaining the rate spread on corporate bonds (The Journal of Finance) pp. 247-276 Fabozzi, F. 2000 Bond markets, analysis and strategies, 4th edition Prentice Hall Inc, USA Ferrucci, G. Empirical determinants of emerging market economies sovereign bond spreads (BIS Working Paper No 205) pp. 1-38 Goldman Sachs (2004) EMEA Economics Analyst (Fixed income research group 2nd April 2004) Grandes, M. (2001) Convergence and Divergence of Sovereign Bond Spreads: Theory and Facts from Latin America (BIS working paper) pp.1-50 Greene W. (2003) Econometric analysis 5th addition New Jersey Pearson Education, Inc. Hattori, M Koji, K and Tatsuya, Y Analysis of credit spread in Japan’s corporate bond market (BIS Papers N5) pp.113-146 Hill, C. Griffiths, W. and Judge G. (2001) Undergraduate econometrics 2d edition New York John Wiley & Sons, Inc. Kamin, S. & Kleist, K. (1999) The evolution and determinants emerging market credit spreads in the 1990s (BIS working paper) pp.1-34 Morgan Stanley (2004) Emerging Market Debt Prospective (Fixed Income Research Sovereigns second quarter 2004) Min H. (1998) Determinants of Emerging Market Bond Spread: Do Economic Fundamentals Matter? (Development Research Group World Bank) pp.1-34 Moscow Narodny Bank (2003) Economic overview (Moscow 2003) Ostrovsky A. (2002) Russia’s bonds stay top of a league table (FT London November 26) Questa G. (2004) Lecture notes on Fixed-Income Securities, Derivatives, and Risk Management Republic of Turkey (2003) Pre-Accession economic program 2003 (Ankara: August 2003) The Central Bank of Russian Federation (2004) Financial Stability Review of the year 2003 (Moscow 2004) The Institute of International Finance (2004) Summary appraisal Russian Federation (March 29, 2004) The Institute of International Finance (2003) Summary appraisal Turkey (December 30, 2003) Westphalen, M. (2001) The determinants of sovereign bond credit spreads changes (.Ecole des HEC, Universit.e de Lausanne, and Fame) pp.1-27 Appendix A Russian current account Exhibit 1 Source: Bloomberg Current account export and import of Russia Current account export (USD$) 2000 2001 2002 2003 2004 q1 23892 25560 21886 31080 q2 25457 26153 26292 31749 q3 26635 25594 28929 34945 37289 q4 29050 24578 30195 38156 Current account import (USD$) q1 -9980 -11291 -12347 -15830 q2 -10379 -13615 -14768 -18106 q3 -11127 -13238 -15725 -19403 -19195 q4 -13375 -15619 -12125 -22097 Table 1 Source: Bloomberg Appendix B I. VARIABLES AND DESCRIPTIVE STATICTICS FOR RUSSIA 1. SPREAD Chart 1.1. Chart 1.2. 2. BUDGET RATIO Chart 2.1. Chart 2.1 3. CPI index Growth/Inflation Chart 3.1. Chart 3.2. 4. EXCHANGE RATE Chart 4.1. Chart 4.2. 5. INDUSTRIAL PRODUCTION INDEX Chart5.1 Chart 5.2. 6. OIL Chart 6.1. Chart 6.2. 7. INTERNATIONAL RESERVES Chart 7.1. Chart 7.1. 8. TRADE RATIO Chart 8.1. Chart 8.2. 9. LIBOR Chart 9.1. Chart 9.2. 10. US INFLATION Chart 10.1 Chart 10.2 Appendix C Dependent Variable: SPREAD Method: Least Squares Date: 08/02/04 Time: 11:27 Sample: 2000:03 2004:05 Included observations: 51 Variable Coefficient Std. Error t-Statistic Prob. C 987.1108 83.52159 11.81863 0.0000 INTRESERVES -0.005478 0.001305 -4.198472 0.0001 OIL -15.20984 4.342084 -3.502890 0.0010 LIBOR 83.59914 11.13166 7.510032 0.0000 USINFLATION 12567.63 5397.197 2.328547 0.0243 R-squared 0.934569 Mean dependent var 612.6761 Adjusted R-squared 0.928880 S.D. dependent var 291.2359 S.E. of regression 77.66782 Akaike info criterion 11.63565 Sum squared resid 277485.4 Schwarz criterion 11.82505 Log likelihood -291.7092 F-statistic 164.2590 Durbin-Watson stat 0.593349 Prob(F-statistic) 0.000000 Table 19 This table illustrates that p values of all selected variables are significant at a 5% chosen significance level i.e. Variable International reserves is significant, its p-value is 0.0001 Variable Oil is significant, as its p-value is 0.001 Variable Libor has p-value of about 0, hence its strongly significant Variable US inflation is also significant with its p-value of 0.0243 STATIONARY TESTS OF THE MODEL FOR THE VARIABLES BEFORE GETTING THE DIFFERENCE OF THE VARIABLES SPREAD * ADF Test Statistic -1.880694 1% Critical Value* -3.6067 5% Critical Value -2.9378 10% Critical Value -2.6069 MacKinnon critical values for rejection of hypothesis of a unit root. Augmented Dickey-Fuller Test Equation Dependent Variable: D(SPREAD) Method: Least Squares Date: 08/02/04 Time: 16:44 Sample(adjusted): 2001:03 2004:05 Included observations: 39 after adjusting endpoints Variable Coefficient Std.

Error t-Statistic Prob. SPREAD(-1) -0.060281 0.032052 -1.880694 0.0713 D(SPREAD(-1)) -0.144549 0.179336 -0.806025 0.4275 D(SPREAD(-2)) -0.075239 0.174859 -0.430282 0.6705 D(SPREAD(-3)) -0.071838 0.170966 -0.420185 0.6778 D(SPREAD(-4)) 0.124166 0.143423 0.865739 0.3946 D(SPREAD(-5)) -0.096082 0.147259 -0.652472 0.5198 D(SPREAD(-6)) -0.139655 0.147576 -0.946324 0.3527 D(SPREAD(-7)) 0.183056 0.151798 1.205917 0.2387 D(SPREAD(-8)) -0.179498 0.148970 -1.204926 0.2391 D(SPREAD(-9)) -0.043895 0.147373 -0.297847 0.7682 D(SPREAD(-10)) 0.152004 0.150140 1.012413 0.3207 D(SPREAD(-11)) 0.278592 0.144654 1.925915 0.0651 C 11.93882 26.75834 0.446172 0.6592 R-squared 0.472270 Mean dependent var -18.58385 Adjusted R-squared 0.228703 S.D. dependent var 49.70003 S.E. of regression 43.64833 Akaike info criterion 10.65141 Sum squared resid 49534.59 Schwarz criterion 11.20593 Log likelihood -194.7025 F-statistic 1.938969 Durbin-Watson stat 1.852635 Prob(F-statistic) 0.076649 Table 20 In the table 20 the ADF test statistic is -1.88 which is greater than 5% critical value of -2.93, hence we fail to reject the null hypothesis i.e. SPREAD is non-stationary. LIBOR ADF Test Statistic -2.226842 1% Critical Value* -3.6019 5% Critical Value -2.9358 10% Critical Value -2.6059 *MacKinnon critical values for rejection of hypothesis of a unit root. Augmented Dickey-Fuller Test Equation Dependent Variable: D(LIBOR) Method: Least Squares Date: 08/02/04 Time: 16:47 Sample(adjusted): 2001:02 2004:05 Included observations: 40 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. LIBOR(-1) -0.076392 0.034305 -2.226842 0.0345 D(LIBOR(-1)) 0.145962 0.149797 0.974395 0.3385 D(LIBOR(-2)) -0.212190 0.143961 -1.473943 0.1521 D(LIBOR(-3)) 0.285527 0.142597 2.002331 0.0554 D(LIBOR(-4)) -0.103831 0.148379 -0.699770 0.4901 D(LIBOR(-5)) 0.120116 0.146748 0.818523 0.4202 D(LIBOR(-6)) -0.163321 0.147160 -1.109819 0.2769 D(LIBOR(-7)) 0.064111 0.148411 0.431982 0.6692 D(LIBOR(-8)) 0.251882 0.146730 1.716641 0.0975 D(LIBOR(-9)) -0.001602 0.129829 -0.012339 0.9902 D(LIBOR(-10)) -0.011992 0.122777 -0.097672 0.9229 D(LIBOR(-11)) -0.149821 0.111791 -1.340180 0.1914 C 0.082853 0.060622 1.366707 0.1830 R-squared 0.651904 Mean dependent var -0.111406 Adjusted R-squared 0.497195 S.D. dependent var 0.198946 S.E. of regression 0.141070 Akaike info criterion -0.822166 Sum squared resid 0.537319 Schwarz criterion -0.273280 Log likelihood 29.44331 F-statistic 4.213738 Durbin-Watson stat 1.991765 Prob(F-statistic) 0.000928 Table 21 From the table above the ADF t-stat is -2.22 which is greater then chosen 5% critical value, hence LIBOR is non-stationary. OIL ADF Test Statistic -0.994247 1% Critical Value* -3.6019 5% Critical Value -2.9358 10% Critical Value -2.6059 *MacKinnon critical values for rejection of hypothesis of a unit root. Augmented Dickey-Fuller Test Equation Dependent Variable: D(OIL) Method: Least Squares Date: 08/02/04 Time: 16:48 Sample(adjusted): 2001:02 2004:05 Included observations: 40 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. OIL(-1) -0.185906 0.186982 -0.994247 0.3289 D(OIL(-1)) 0.160790 0.236074 0.681100 0.5016 D(OIL(-2)) -0.271690 0.211550 -1.284281 0.2100 D(OIL(-3)) 0.353859 0.219430 1.612629 0.1185 D(OIL(-4)) -0.232322 0.201876 -1.150818 0.2599 D(OIL(-5)) 0.384136 0.202246 1.899351 0.0683 D(OIL(-6)) -0.090682 0.190597 -0.475776 0.6381 D(OIL(-7)) 0.340451 0.185742 1.832927 0.0779 D(OIL(-8)) -0.134800 0.174351 -0.773153 0.4461 D(OIL(-9)) 0.271867 0.174205 1.560617 0.1303 D(OIL(-10)) -0.001988 0.148715 -0.013365 0.9894 D(OIL(-11)) 0.365381 0.140637 2.598051 0.0150 C 5.230770 4.984322 1.049445 0.3033 R-squared 0.422543 Mean dependent var 0.332500 Adjusted R-squared 0.165895 S.D. dependent var 2.755309 S.E. of regression 2.516406 Akaike info criterion 4.940498 Sum squared resid 170.9721 Schwarz criterion 5.489384 Log likelihood -85.80996 F-statistic 1.646391 Durbin-Watson stat 1.933688 Prob(F-statistic) 0.137043 Table 22 We can conclude that OIL is nonstationary as -0.99>-2.93, meaning that we reject Ho: series contains a unit root. USINFLATION ADF Test Statistic -1.912283 1% Critical Value* -3.5973 5% Critical Value -2.9339 10% Critical Value -2.6048 *MacKinnon critical values for rejection of hypothesis of a unit root. Augmented Dickey-Fuller Test Equation Dependent Variable: D(USINFLATION) Method: Least Squares Date: 08/02/04 Time: 16:51 Sample(adjusted): 2001:01 2004:05 Included observations: 41 after adjusting endpoints Variable Coefficient Std.Error t-Statistic Prob. USINFLATION(-1) -1.190307 0.622454 -1.912283 0.0658 D(USINFLATION(-1)) 0.485145 0.576866 0.841002 0.4072 D(USINFLATION(-2)) 0.442545 0.531972 0.831896 0.4123 D(USINFLATION(-3)) 0.200881 0.486122 0.413232 0.6825 D(USINFLATION(-4)) 0.409097 0.461410 0.886625 0.3826 D(USINFLATION(-5)) 0.350624 0.423804 0.827325 0.4148 D(USINFLATION(-6)) 0.195245 0.381117 0.512297 0.6123 D(USINFLATION(-7)) 0.327261 0.350060 0.934873 0.3576 D(USINFLATION(-8)) 0.133617 0.307554 0.434450 0.6672 D(USINFLATION(-9)) -0.165140 0.240325 -0.687154 0.4974 D(USINFLATION(-10)) 0.013578 0.186657 0.072743 0.9425 C 0.002229 0.001190 1.872230 0.0713 R-squared 0.524110 Mean dependent var 0.000101 Adjusted R-squared 0.343600 S.D. dependent var 0.002880 S.E. of regression 0.002333 Akaike info criterion -9.044193 Sum squared resid 0.000158 Schwarz criterion -8.542660 Log likelihood 197.4060 F-statistic 2.903496 Durbin-Watson stat 1.883144 Prob(F-statistic) 0.010561 Table 23 US inflation is found to be non-stationary, its ADF t-statistic is -1.91, which is greater 5% critical value of -2.93. INTERNATIONAL RESERVES ADF Test Statistic 1.754434 1% Critical Value* -3.6067 5% Critical Value -2.9378 10% Critical Value -2.6069 *MacKinnon critical values for rejection of hypothesis of a unit root. Augmented Dickey-Fuller Test Equation Dependent Variable: D(INTRESERVES) Method: Least Squares Date: 08/02/04 Time: 16:53 Sample(adjusted): 2001:03 2004:05 Included observations: 39 after adjusting endpoints R-squared 0.630443 Mean dependent var 1469.862 Adjusted R-squared 0.438273 S.D. dependent var 2255.093 S.E. of regression 1690.158 Akaike info c