The Effect of Global Risk Indicators on Developing Country Stock Exchanges: The Case of BRICS-T

Countries, as well as, financial markets are affected by global risk factors and leave deep marks on their stock markets and economies. It is common knowledge that risk factors increase, especially in times of crisis. Therefore, it is important to anticipate risk factors and minimize the risk at all times and even more in crises periods. Risk is a factor that can occur at any time and can do great damage. Risk factors have created many risks, especially before and after the 2008 global crisis. At the same time, it can be seen that its effects are great for individual and institutional investors and observed that the effects of some global crises continue for years. As a result, global crises have left much credit, financial, operational, and strategic damages on developing and developed countries. Even if the crisis starts in a country, it can spread all over the world or most of the financial markets. Therefore, risk management must be managed correctly in order to minimize the impact of crises. During times of crises, from the past to the present, great traces were left on the economies when the crises could not be detected in advance and correctly. Risk factors must be analysed well in order to minimize and predict outcomes of the risks of these crises. First, the risk should be at the lowest or reasonable level for investments. This is a pre-requisite for investors to make the correct investment. The lower the risk, the more attractive it is for investors to get involved in the financial markets.

There are many global risk indicators for financial markets. Some of them are common and popular but others are not. This paper targets to determine the effects of global risk factors on the developing BRICS- T (Brazil, Russia, India, China, South Africa, and Turkey) countries' stock markets. To this end, CDS (Credit Default Swaps), VIX and Credit Ratings are used as global risk indicators in this current research.

CDSs create an important place for investors. Investors do not predict a situation such as non-payment when the maturity of their financial instruments is up. Among the risks investors take, this is the most preferred method because it contains the least amount of risk. It can also be said of replacing a country’s own credit risk with another country. Credit rating notes are considered as the numerical evaluation of the credit ratings of countries. A decision can be made by looking at the credit rating notes in order to understand a country's investment-note situation. The VIX index is not calculated separately for countries and is evaluated on a grade. The VIX index can be regarded as the most striking issue recently. The VIX index is also known as the fear index. Therefore, the VIX index not only shows short-term changes but also fluctuations in risk perception. Besides, the VIX index is used to detect volatility or how fast prices change during time. In addition, the VIX index is seen as a way of measuring market sentiment, and especially the degree of fear among market participants.

The rest of the paper is organised as follow. In the following section, variables and the theoretical framework of the study are explained one by one. For the second part of the study, a literature review is prepared based on previous studies. After the explanations of methodology and data set, findings and results are summarized. Study ends with conclusion and recommendations.

The source of the word risk comes from French. The word originating from the French language, is actually "hazard". When the dictionary meaning of the word “hazard” is examined, it is defined as the danger of injury (Ozbilgin: 2012: 88). Risk can be explained as a threat that prevents events that may or may occur in the future. Also, risk can prevent situations where goals can be achieved. The risk is seen as the possibility of situations that may occur instantly or in the future. Risk can lead to opportunities as well as threats. In addition to being considered as neutralising the negative impact of unexpected events, risk can also mean creating opportunities from negativities with a new perspective over time (Ankara: 2013: 1). Apart from the definition of risk, that of risk management is also important for practitioners. Risk management is the controlling of effects of negative developments on institutions, the management of uncertainties by developing strategies through risk assessment. There are various strategies, namely, transferring the risk to another segment can be a form of risk aversion, mitigating the negative effects of the risk, and accepting some of the consequences of risks. Financial risk management can be described as managing risks through the exchange of financial instruments (Durak: 2009: 4). In this study, risk factors are examined from a global perspective and VIX, CDS and credit ratings are determined as global risk indicators. L. Tappan first initiated credit rating agencies between 1837 and 1841 in relation to companies' failure to fulfil their commitments after the collapse of the US economy. In 1909, Moody's companies started to provide ratings such as 'A', 'B' and 'C' and the first credit ratings were created. Subsequently, the Fitch Company was established. In 1941, Standard and Poor's (S&P) company was established and these were and have remained the three major companies for credit ratings (Kargi: 2014: 356). In the globalized world, credit rating agencies have increased in importance with the emergence of financial crises. There have been improvements in credit ratings from past to present. It is seen that the importance of credit ratings, which were of little importance at first, has increased in importance recently. These institutions are the institutions that guide investors in the right way, even if they have differences in terms of grading.

Credit default swaps (CDS) are a type of insurance against a particular company's default risk. The company is referred to as the reference asset and it is called the credit default event. It is a contract between two parties called the protection buyer and the protection seller. Under the contract, the protection buyer is compensated for any damage arising from a credit event on a reference instrument. In return, the protection buyer makes periodic payments to the protection seller (The Economic Times: 2020). While credit rating emerged as a tool to express credit risk: Credit Default Swaps (CDS) emerged as an instrument that transfers credit risk and over time has become a credit risk measuring instrument. In this respect, credit rating grades and CDS premiums are important tools that show credit risk (Senol: 2012: 49). Credit default swaps (CDSs) proved to be one of the most successful financial innovations of the 1990s. They are vehicles that provide insurance against a particular company that does not pay its debts. The company is known as the reference asset and the default of the company is known as the credit event. The buyer of the protection makes periodic payments to the protection seller at a predetermined fixed rate each year. Payments continue until the end of the contract's life or a loan event (Hull and White: 2003: 3).

The VIX volatility index is an index that measures the degree of fear found in the Chicago Board Option Exchange Volatility index. Another name for the VIX index is the fear index. The VIX index was established in 1993. Since the VIX index follows the volatility in the markets, it is referred to as the fear index. The increase and amount of expected dividends in the VIX index are calculated using binomial methods. The VIX index was initially calculated on the S&P 100 index. After 2003, options were followed on the S&P 500 index. In 1993 the Chicago Board Options Exchange (CBOE) introduced a volatility index based on the prices of index options. This was an implied volatility index based on option prices of the S&P100 and it was traced back to 1986. Until about 1995 the index was not a good predictor of realized volatility (Brenner, Shu and Zhang: 1989: 63; Jung: 2015: 189). The VIX is an index, like the Dow Jones Industrial Average (DJIA), that is computed on a real-time basis throughout each trading day. The only meaningful difference between the VIX and DJIA is that the VIX measures volatility and the DJIA measures price (Whaley: 2009: 98).

The VIX Index, CDS premiums and credit ratings are the main variables of this study and the authors examined the impact of these variables on BRICH-T countries based on time series analyse methods. In the following part, literature review of similar studies is summarized.

The literature review part consists of three parts. These are studies related to credit ratings and its effect on stock markets, VIX Index, and CDS premiums and their impact on financial markets, especially on emerging markets. Firstly, we perused the impression of credit rating scores on stock markets in developing countries. Although credit rating agencies are not of great importance for governments and stock exchanges, credit ratings are accepted as an essential tool for decision-making. It has been determined that credit ratings, particularly in emerging markets, are one of the most crucial parts for institutional investors. Assessment of credit ratings on emerging markets are those mostly subjected to studies (Mihaelajeno, 2015; Afonso et.al. 2012; Fatnassi et.al., 2014; Hull et.al., 2004; Brooks et.al., 2004; Hilscher and Wilson, 2016; Cantor and Packer, 1996; Reisen and Maltzan, 1999; White, 2010; Hooper et.al., 2008; Reinhart, 2002; Kim and Wu, 2011; Aizenman, et.al., 2013; Benmelech and Dlugosz, 2010; Kaminsky and Schmukler, 2002). The findings of these studies address the positive correlation between credit rating and stock market returns. This means that increases in the credit rating create an increase in the stock markets while the opposite is also valid, whereby a decrease in the credit rating create a decrease in stock market returns. The effects of credit ratings on Turkey's stock market is also one hot topic in the literature review, especially for researchers in Turkey. There are a considerable number of studies (Pirgaip, 2017; Tutar et al., 2011; Ovali et al., 2020; Çağlak et al., 2018; Tekin, 2018; Iskenderoglu and Balat, 2018; Yıldırım et al., 2017; Kargi, 2014; Toraman and Yuruk, 2014) that found that negative movement of credit ratings led to a decrease in stock market returns, especially in Borsa Istanbul Stock Indices. The second risk indicator for the global markets is the VIX index. There are many studies in the literature review to test the effect of the VIX Index on financial markets. Some of these studies focus on the relationship between BIST Indices and VIX Index (Başarır, 2008; Kaya, 2015; Kula and Baykut, 2017; Sarıtas and Nazlioglu, 2019; Bayrakdaroglu and Celik, 2015; Gunay, 2019; Erdogdu and Baykut, 2016; Akcalı et.al., 2019), while some of them dissected the efficacy of VIX ındex on emerging markets (Korkmaz and Çevik, 2009; Adjasi et.al., 2008; Mikhaylov, 2018; Oner et.al., 2018; Jayasuriya, 2005; Aizenman and Marion, 1999; Iskenderoglu and Akdag, 2020; Gursoy, 2020; Silva, 2002; Ozdemir, 2020). A significant amount of these studies concluded that the VIX index harms stock markets and economies, especially in emerging markets. Studies about credit default swaps and their effect on financial markets are also popular in the literature review. Kırca et al. (2018), Yenice et al. (2019), Conkar and Vergili (2017), Reyhan and Gazel (2019), Bektur and Malcioglu (2017), Celik and Boztosun (2010), Ozkan and Cakar (2020) examined the CDS and BIST Index from a different perspective. An intensive amount of these studies have found a negative impact of CDS values on BIST Indices. So as with the VIX Index, the CDS rating also has adverse effects on Turkey’s stock markets. In addition to studies about CDS premium in Turkey, there is a sufficient amount of studies (Arestis et al., 2001; Levina and Zervos, 1998; Demirer et al., 2010; Bologna and Cavallo, 2002; Buberkoku, 1997, Park and Bae, 2004; Kassimatis, 2002; Puliga et al., 2014; Aksoylu and Gormus, 2018; Ericsson et al., 2009; Sahin and Sumer, 2014; Brigo et al., 2012; Abdellahi et al., 2017, Dullmann and Sosinska, 2007; Vurur and Ozen, 2020) that focus on the effect of CDS premiums, especially on emerging markets. Their findings are parallel to the results of studies on BIST Indices.

To test the impact of VIX, CDS and credit ratings on emerging markets, the authors had to choose the best-fitted model to detect long-run relationship between variables. In the next part of the study, the methodology is examined and briefly explained.

Based on the unit root test result of the variables, we have a different level of unit root for the emerging market (BRICS-T) stock exchanges and VIX, CDS, and Credit ratings. This is why, in order to test the long-run relationship between dependent and independent variable, ARDL model will be applied. The data set of this study consists of stock market indices of BRICS-T (dependent variables) and VIX, CDS, Credit ratings from 2008 to 2020.

ARDL (Autoregressive Distributed Lag Bound) Test is also known as ARDL Bound Test. The ARDL bound test was developed in 2001 by Muhammad Hashem Pesaran and Sangcheol Shin. This test is used to examine the concept of cointegration with the combination of the series for at least two non-stationary series. Besides, the ARDL bound test method has recently become a more effective and frequently used method compared to the cointegration tests of Johansen (1988), Johansen Juselius (1990) and Engle- Granger (1987) (Turna: 2017: 80). ARDL Test has many advantages compared to other cointegration methods. In contrast to other cointegration techniques, ARDL test does not impose a restrictive assumption that all variables under study should be combined in the same order (Şimşek: 2016: 71). ARDL approach is based on the least squares method and unlike classical cointegration analysis, it is not necessary to apply a unit root test beforehand in ARDL analysis (Esen et.al. 2012: 256). ARDL does not have the conditions to be stationary like other cointegration tests. In addition, it aims to express a stationary combination for two non-stationary series. In addition, ARDL has more advantages than other cointegration tests. The ARDL model provides an advantage over other cointegration tests since the variables are not taken into account in the integration degree. It brings about spurious regression in non- stationary time series depending on time series. At the same time, the difference is made in series in order to ensure stability.

When applying ARDL, this is performed in two parts. First, for the ARDL test, the variables of the subject and model are examined, and categorised as long term or short term. Flexibility will be decided by using the ARDL test in the test results. At the same time, under the condition that there is a cointegration relationship, long and short runs are tested among variables. In order to determine lag lengths, long and short term relationships between dependent and independent variables are determined. To determine lag lengths, AIC and SHC (Schwartz) selection criteria of dependent and independent variables are used. On the other hand, the smallest criterion among the lag lengths constitutes the lag length of the model (Baykut: 2020: 101). The first unconstrained error correction model defined for the ARDL test by Pesaran and Shin is given below:

$\Delta \gamma_{t=a_0}+\sum_{i=1}^m a_{1 i} \Delta \gamma_{t-1}+\sum_{i=0}^m a_{2 i} \Delta M_{t-i}+\sum_{i=0}^m a_{3 i} \Delta E_{t-i}+a_4 \gamma_{t-1}+a_5 M_{t-1}+a_6 E_{t-1}+\mu_t$

In order to determine that there is an ARDL test cointegration relationship, the coefficients of the first period lags of dependent and independent variables are tested collectively by applying the F-statistics test (Wald test) to determine significance. H0 hypotheses expressing the absence of cointegration between variables of the ARDL test are formed. These H0 hypotheses are as follows:

$H _0: a _1= a _2=\ldots= a _{ k }=0 \longrightarrow$ No cointegration hypothesis

ARDL test is the alternative H0 hypothesis expressing the presence of cointegration in variables is as follows:

$H _1: a _1 \neq a _2 \neq \ldots \ldots \ldots \ldots \ldots \ldots a _{ k } \neq 0 \longrightarrow$ Cointegration exists alternative hypothesis

The ARDL technique is preferred when results are I(0) and I(1), or when there were both combinations dealing with variables integrated in different order. However, if the situation is I(2) for the ARDL model, this technique becomes disabled. One also notes that when using the ARDL test technique, there is only one long-term relationship between basic variables of small size. If the basic forms have a long-term relationship, this is determined by the F-statistics (Wold Test). The distribution of F-statistics is not standard regardless of whether the variables in the system are I (0) and I(1) (Nkoro and Uko: 2016: 81). Although there are no requirements for ARDL testing, unit root testing is recommended. In order to avoid misapplication, prediction and interpretation based on forecasting and policy stance, it is necessary to investigate the necessary conditions that reveal the ARDL co-integration technique. If the conditions are complied with, it may lead to inconsistent and unrealistic estimates with the erroneous determination of the model and its effect on the estimation and policy (Nkoro and Uko: 2016: 63). ARDL bound test can be formulated as follow (Baykut: 2020: 101):

$\Delta Y_t-\alpha_0+\sum_{i=1}^m \alpha_{1 i} \Delta Y_{t-i}+\sum_{i=0}^m \alpha_{2 i} \Delta X_{t-i}+\alpha_2 X_{t-i}+\alpha_4 X_{t-1}+e_t$

In the ARDL test, after the long-term is determined, the optimum length structure should be selected using standard criteria such as Swartz Bayesian (SBC) or Akaike Information (AIC). Therefore, the next step is to determine the long-run and short-run coefficients in the models. The long-term formula for the ARDL test is as follows (Özçalık: 2014: 367):

$Y_t-\alpha_0+\sum_{i=1}^m \alpha_{1i} Y_{t-i}+\sum_{i=0}^{m} \alpha_{2i} M_{t-i}+\sum_{i=0}^m \alpha_{3i} E_{t-i}+\mu_t$

Error correction term is used in ARDL short run model. The short run dynamic model can be presented as follows (Özçalık: 2014: 367):

$\Delta Y_t=\alpha_0+\sum_{i=1}^m \alpha_{2 i} \Delta Y_{t-i}+\sum_{i=0}^m \alpha_{3 i} \Delta M_{t-i}+\sum_{i=0}^m \alpha_{4 i} \Delta E_{t-i}+\alpha_1 E C_{t-i}+\mu_t$

In the ARDL error and short-run test, the ARDL long model formulation with Y dependent variables and two independent variables (E and M) can be predicted by the above formulas by econometrics and statistics programs. The EC in the ARDL test formula shows the Error Correction Model. ARDL Error Correction Model aims to show how soon shocks that occur due to independent variables will stabilize in the long term.

In the analyse part of the paper, the first part consists of descriptive statistic and unit root test applied to detect structure of the data set. The paper aims to expose the long-run relationship between global risk indicators and emerging markets stock market. After determining whether the data sets are stable or not, the long-term relations of the countries were examined by applying the time series model. To this end, the first section of the analyses is focussed to generate descriptive statistic. The descriptive statistic of the data set is provided in Table 1.

After obtaining the descriptive statistics of the series, a high volatility was observed in the data set in the relevant period. It should also be taken into account that there were global financial crisis and corona virus effects in this period. Even if the number of data of the countries is different, there is an increasing momentum in the stock market returns of developing countries. This situation arises from the graphs of the stock market data of each country index. In addition, VIX index and CDS premiums generally follow a stable course, except in crisis periods. High volatility is observed in the data for these two variables only in times of crisis. After the descriptive statistics were obtained, the unit root tests of the series were performed and reported in Table 2.

As a result of the unit root test analysis, different levels of unit root for each data set were obtained. VIX index is the same for all countries and calculated globally. Since the unit root result of different orders was obtained for each variable, it was decided to use the ARDL model to determine the long-term relationship between the series. Because of the different unit root level of variables, the ARDL model was deemed the best model for the assessment of relationship between stock exchanges and global risk indicators. For this purpose, the ARDL model was applied for all country data and the best fitted ARDL models are shown in Table 3.

In order to determine the fitted ARDL model, firstly, the suitable lag length must be detected. To detect appropriate lag length, the VAR model was generated, applied, and summarized in Table 4. In the last column of Table 4, the most suitable ARDL model lag length for each country is given. After determining the most suitable lag length, the ARDL Bound test application phase was started and the results are shown in Table 4. In the ARDL Bound test approach, in order to have a long-term relationship between variables, the detected f statistic must be higher than the upper bound value. If the f statistic is lower than the lower bound, there is no long-term relationship, and if the f statistic is between the lower and upper bound, no definite relationship can be mentioned. The f statistic between the lower and upper bounds is expressed as the grey area and is generally not reported. While reporting, calculations should be made according to the significance levels of 1%, 2.5%, 5% and 10%. Based on these rules, results are summarized in the following table.

According to the results obtained after the application of the ARDL model, there is a long-term relationship between the variables in all countries except China. 5% significance level is considered sufficient to determine the relationship between variables in social sciences. From this perspective, statistically significant relationships were found in most of the established ARDL models starting from the 2.5% significance level. F statistic values obtained in the model established for Turkey is significant at the 1% significance level. This f-statistic value was determined to be higher than the upper limit value at all levels. The same is true for the model established for South Africa and India. In the ARDL model established for the data set of South Africa and India, the long-term relationship between variables is valid at 1% significance level. On the other hand, in the analyses for Brazil and Russia, no statistically significant relationship was found at the 1% significance level. Just after 1% significance level, f statistic is significant for 2.5%, 5% and 10% level. Whereas, there is no statistically significant relationship detected for China based on ARDL Bound test. It means that, VIX, CDS and stock exchange variables are not cointegrated in the long-run. So, CDS and VIX ındex values are not a meaningful decision criteria for the China stock market. For the rest (Turkey, Brazil, Russia, South Africa and India) CDS and VIX index numbers must be taken into consideration for the investment decisions relating to the stock markets.

Through globalization, the borders surrounding various countries have disappeared. The crisis that occurs in one country, thanks to globalization, is spreading quickly among other countries. Hence, any crisis can be said to go beyond any geographical barriers. Risks increase due to developments in markets and economies between countries. Accordingly, given the problems that occur in the markets in this globalized world, risk management and risk protection methods have had to be addressed in all countries. Although a crisis might start in one country, this could cause havoc across the whole world. The 2008 crisis can be cited as the most important example. Although the 2008 crisis emerged in the developed country, it affected the economies of the whole world from one day. This crisis caused great damage to the economies of the whole world. Countries that want to avoid certain crisis and risks have tried to implement measures to protect countries’ economies. They aimed to keep the economies of the country under control through indices such as CDS, Credit rating and VIX. Despite the fact that the credit rating method is the most commonly used method, one has to consider that it carries a little risk. However, CDSs also carry little risk and protect the investors from said risks. For this reason, CDS are the most traded products among loan derivatives and are very popular in the liquid market. The VIX index is seen as the subject that has been followed and researched recently. Countries try to avoid risks by following the VIX index, that is, the fear index.

In this study, the effects of global risk indicators on developing country stock markets were examined. To this end, BRICS-T (Brazil, Russia, India, China, South Africa and Turkey) countries are determined as an emerging market for the current research. The data set of the study covers a period between 2008 and 2020. This period contains both global financial crises and the coronavirus pandemic. Both of them affected the stock markets very negatively. To determine the relationship of global risk indicators and stock markets, ARDL model was determined as the best fitted model for each stock market just after applying unit root test. Based on the ARDL model, it was determined that 5 out of 6 countries showed that there was a significant relationship between the variables. Except for China, there is a statistically significant relationship between CDS premiums, VIX index and stock market returns of emerging markets. It means that, VIX and CDS premiums significantly affect the returns of Turkey, Brazil, Russia, South Africa and India stock market indices. Therefore, investors should take into account VIX and CDS premiums when investing in these countries and especially in their securities. These outputs are consistent with the previous studies by (Buberkoku, 1997; Levina and Zervos, 1998; Arestis et al., 2001; Bologna and Cavallo, 2002; Park and Bae, 2004; Dullmann and Sosinska, 2007; Ericsson et al., 2009; Demirer et al., 2010; Brigo et al., 2012; Sahin and Sumer, 2014; Puliga et al., 2014; Kaya, 2015; Erdogdu and Baykut, 2016; Kula and Baykut, 2017; Abdellahi et al., 2017; Aksoylu and Gormus, 2018; Sarıtas and Nazlioglu, 2019; Gunay, 2019; Akcalı et.al., 2019; Vurur and Ozen, 2020) in the literature. For further studies, is the authors recommend to carry out research on MSCI Emerging Markets Index countries by applying not only CDS and VIX Index but also MOVE Index and JP Morgan Volatility Index to detect the effects of risk indicators on emerging markets stock indices.