THE RELATIONSHIP BETWEEN EXTERNAL DEBT, INTERNAL DEBT AND ECONOMIC GROWTH: AN EMPIRICAL ANALYSIS ON BRICS-TM COUNTRIES - DIŞ BORÇ, İÇ BORÇ VE EKONOMİK BÜYÜME ARASINDAKİ İLİŞKİ: BRICS-TM ÜLKELERİ ÜZERİNE AMPİRİK BİR ANALİZ

Ozel kesim acigi, butce acigi ve cari acik, ic ve dis borclanmanin temel nedenleri arasinda yer almaktadir. Ulkelerin gelismislik farklarinin azaltilmasinda devlet piyasaya mudahale ederek, borclanma yoluna gidebilmektedir. Bu nedenle, BRICS-TM ulkelerinin yuksek buyume hizlarina sahip olmasinda dis ve ic borclanmanin varliginin arastirilmasi onem arz etmektedir. Bu calismanin amaci, BRICS-TM ulkelerinde 2000-2016 donemine ait verilerle dis ve ic borc ile ekonomik buyume arasindaki iliskinin arastirilmasidir. Arastirmanin ampirik analizi icin panel veri analizi yontemi kullanilmistir. Analizden elde edilen bulgulara gore, dis ve ic borc ile ekonomik buyume arasinda esbutunlesme ve nedensellik iliskisi bulunamamistir. Bu sonuclar, BRICS-TM ulkelerinin yuksek buyume oranlarina sahip olmasinin ic ve dis borclanma ile desteklenmedigini gostermektedir.


INTRODUCTION
Today, when the global competition is increasing continuously, accelerating economic activities cause emerging developmental differences among countries. Governments have a significant role in eliminating these developmental differences. States endeavour to activate the economic activities by intervening in the economy via the public finance policies. Lack of savings and fiscal deficit out of the current deficit are the most significant reasons for borrowing. Public expenditures cannot be met by the public revenues through expansionary policies. Thus, borrowing becomes an important fiscal policy tool in public deficit financing. Borrowing which is an ordinary intervention tool of governments to the economy via Keynesian policies affects the economy negatively or positively as an internal or external debt. The money borrowed should be used in proper areas to enable a positive impact on the economy.
Otherwise, the principal and interest payments put the national economy in a difficult situation as well as the removal of debt sustainability. Therefore, the success of debt management becomes more of an issue.
It is thought that Brazil, Russia, India, China, South Africa, Turkey and Mexico will have a remarkable place in the world economy in the long run due to their high rates of growth and young population. 'BRICS-TM' expression in the literature is composed of the English initials of the countries mentioned. The effect of internal and external debts of the public in the development process of these countries is the subject of current study. This study analyzed the relationship between internal debt, external debt and economic growth by econometrical methods through using the data in annual frequency regarding the period of 2000-2016 in BRICS-TM countries. Panel data analysis was utilized in the analysis of the data. Testing the dependency between the units in panel data analyses is significant to provide more consistent results.
The relationship between internal debt, external debt and economic growth in BRICS-TM has not been researched so far, which brings originality to this study. Moreover, the analysis in this study was conducted via new generation tests by testing the cross-sectional dependence. Our survey differs from other studies due to these reasons. It is seen in previous studies that the hypothesis of "there is a relationship between the internal debt, external debt and the economic growth" is valid. In the study, we first present a theoretical and conceptual framework relating to the issue; later, a literature review is provided. Findings obtained from the analyses are discussed in the last section.

INTERNAL AND EXTERNAL BORROWING IN FINANCING OF ECONOMIC GROWTH
Internal debt is defined as "obtained finance from domestic sources by the government through issuing domestic government bonds". In other words, a part of the national income is transferred from individuals and public enterprises to the governments. Governments prefer internal borrowing because of reasons such as paying the maturing liability, removing the imbalances in the economy, public  Temmuz 2020 July 280 deficits, economic and social difficulties arising from new taxes and increasing tax rates (Bayraktar, 2011:2). Keynesian view mentions that internal debt will stimulate investments, and productivity will increase when governments reserve the sources that have come from internal borrowing for the investments that support private sector investments. The positive effect of internal debt on the private sector investments via public expenditures is known as the "crowding-in effect" in the literature (Çevik and Cural, 2013:117). Using the sources obtained through internal borrowing so as to contribute to the economic growth is important in paying the debts.
There is not an addition to the available sources of a country via the internal borrowing; however, a transfer is actualized in the economy. Especially the internal borrowing which aims to enable bringing nonexpendable funds in the economy contributes to economic growth. Internal borrowing may cause negative outputs in economy in an environment in which productivity and general economic structure are not considered (Çiçek et al., 2010:143). The size of an internal debt stock is effective on inflation and income distribution. Since an increasing internal debt brings along high interests which will be paid for the savings, it creates a destructive effect on price stability. High inflation will affect the purchasing power of the society, notably the fixed-income employees negatively. Moreover, since the high-interest rates will decrease the private sector investments, the "crowding out effect" will emerge.
It is also known that this high internal borrowing rate will increase interest rates and rollover risk by shaking the confidence in the economy (Çoban et al., 2008:249;Berkay and Ağcakaya, 2017:3). The level of internal borrowing is significant in the financing of economic growth. While internal borrowing contributes to economic growth up to a certain point, it may damage economic growth after exceeding that certain point.
External debt is defined as "the debts that are borrowed by a country from external resources under the condition of repaying them with the capital and interests at the end of a given period" (Egeli, 2003:124). External borrowing is preferred due to reasons such as closing the current deficit, lack of domestic savings, budget deficits, global interest rates, lack of financial sources for development thrusts, paying off the external debts by the external debts, lack of enough foreign currency reserve (Peker and Bölükbaş, 2013:289-290). Since external borrowing provides an input entrance into the borrowing country, the investments will be stimulated, and this will contribute to economic growth. Marginal proceeds of the investments which are realized by the external debts need to be higher than the cost of the external debts. An increase in national income can be observed after increasing the production capacity if this condition is guaranteed (Çiçek et al., 2010:143).
The credibility score of the countries in external borrowing is highly important. Problems such as borrowing at a higher interest and unavailability of new sources may occur when confidence to that country is lost. External debts are discharged by the foreign currency, which differentiates external debt from internal debt. Any changes that occur on the exchange rate will cause more effect on external debt in comparison to internal debt. In addition to this, while a country can pay off its internal debts by  (Esener, 2013:14).
The most important advantage of external borrowing is that there is a lower "crowding out effect" risk in comparison with internal borrowing. Moreover, governments decrease their real debt burden by external borrowing without having inflation problem stemming from internal borrowing. In the inflationary environment, external borrowing enables long term debt due to its lower interest rates in comparison with the higher interest rates of internal borrowing. The disadvantage of external debt is that it creates a "crowding out effect" on private investments by the nominal exchange rate effect.
Meanwhile, external borrowing increases the sensitivity to the developments in other countries (Berkay and Ağcakaya, 2017: 4).
Expansionary policies that are applied in underdeveloped and developing countries to increase the rate of capacity utilization provide a positive contribution to the economic output in the short term.
However, economic growth will occur in the long term via the policies that can increase production capacity. The factors that affect economic growth are as follows; technology, human and real capital, and political and economic stability. An increase in the level of these factors cannot affect the long-term economic growth at the level desired. This is because borrowing may appeal to countries to stimulate investments and economic growth. South Africa, which started its efforts to participate in BRIC group in 2010, became a member of the group on 24 December 2010. The group was called as BRICS. With reference to the World Bank Statistics, BRICS countries have more than 40% of the world population and also occupy the one-fourth of the world area. In conclusion, Brazil, Russia, India, China, and South Africa are important economic forces (Nistor, 2015:982). Long term estimations show that South Africa will grow with an average of 3.5% in the next 50 years. As the population growth rate decreases, there will be a faster increase in per capita income. It is estimated in the light of these projections that South African economy will be smaller than the economy of BRIC (US $1.2 billion for Russia in comparison with the US $5.9); however, the per capita income will be higher (Wilson and Purushothaman, 2003:11).

THE SOCIO-ECONOMIC STATUS OF BRICS-TM COUNTRIES
Turkey and Mexico have become the locomotive of the world economy due to their high economic growth rates and young population structures. Therefore, Turkey and Mexico have been added to BRICS countries and they are now called "BRICS-TM" countries. While Turkey got involved in this group by means of 9.2% and 8.8% developmental performance obtained in the years of 2010 and 2011 As seen in the data of per capita income, the highest values based on the averages of the years of 2010-2017 belong to Turkey, and Brazil is on the second rank. The country that has the lowest income is India with 1446 dollars, and China is on the second rank with 5268 dollars. According to 1990-2017 data, China has the highest increase by 8.3% as the income level, and India has the second highest increase by 4.7% as the income level. Regressing the population growth in China and India is one of the most important reasons for the high growth rate in spite of their low-income level. The population growth takes positive value in all the countries except for Russia. In other words, the population has decreased in Russia though it continues to increase in other countries. The debt structures of BRICS-TM countries reveal that the share of the internal and external debt from the gross domestic product has increased in China, Mexico and South Africa while the same share has decreased in Brazil, Russia and Turkey. Even though India has reduced its internal debt, it is still in the position with the highest debt. The country which has the lowest debt is Russia. While the country which has the highest ratio in terms of the external debt is Turkey, the country that has the lowest ratio about the same issue is China. The internal debt is more than the external debt in all the countries except for Russia and Turkey. The reason of high external debt in comparison with the internal debt in Russia and Turkey is that the cost of the external borrowing is lower than the cost of the internal borrowing in these countries.

LITERATURE REVIEW
The relationship between economic growth and public debt, which is the sum of internal and external debts, is the leading issue that has been researched for many years. Debt crises that emerged in Latin American countries in the 1980s affected the economic performances of those countries. It is pointed out that meeting the financing demands of countries by internal and external borrowing will negatively affect economic growth. The relationship between internal and external debt was analyzed in the literature by considering underdeveloped and developing countries or country groups. Besides, the literature includes studies that survey the relationship of public debt with respect to internal and external debt.
Schclarek and Ballester (2005) and Favour et al. (2017) researched the relationship between these variables by analysing only a single country, and they found out that there was a negative relationship between internal debt, external debt and economic growth. According to Çevik and Cural (2013), there is not a causality relation between internal debt and national income when there is a unidirectional causality from external debt to national income. Aminu et al. (2013) argue that there is not a causality between internal debt and economic growth when there is a bidirectional causality between external debt and economic growth.
In addition, while external debt negatively affects growth, internal debt affects the growth in a positive direction. Ntshakala (2015)  and Daud (2016) surveyed the relationship between these variables for only a single country and determined that public debt positively affected economic growth. According to Balassone et al. (2011), the effect of public debt on economic growth is negative.
The other studies which found a causality relation between external debt, internal debt or public debt and economic growth include Onakoya and Ogunade (2017) (2017). On the other hand, there are also studies that could not determine a causality relationship between these variables (Ezeabasili et al., 2011;Ademola et al., 2018;Daud, 2016).
Much as there are different results about the direction of the relationship between internal debt, external debt and economic growth; it is pointed out that internal and external debt negatively affect economic growth. We see that the general run of the studies was employed time series methods, OLS, and the first-generation tests by considering the data of a single country. Only one of the studies (Erataş It can be thought that our research study differs from the other studies in literature due to using data from more countries and considering the cross-sectional dependence.

Method
The panel data analysis was used to investigate the relationships between the internal-external debts and the economic growth in BRICS-TM countries. The existence of the dependence between the cross sections of the variables was scrutinized by the LMadj test of . The stationarity of the series was analyzed by the MADF unit root test of Taylor and Sarno (1998). The Delta test of Pesaran and Yagamata (2008) was utilized to determine whether the slope coefficient changes between the units. The Durbin-Hausman (2008) cointegration test specified the long-term relationship between the variables. The Dumitrescu-Hurlin (2012) panel causality test was used to find the causality between the variables.

The Cross-Sectional Dependence
In the cross-sectional dependence (CD), it is determined whether the panel units are affected by the shock at the same degree when a specific shock comes to the variables in the panel data analysis.
The cross-sectional dependence should be tested in studies because it is a determinant in the reliability of estimations or selection of estimation methods (Ün, 2018: 88 While the main hypothesis shows that there is no cross-sectional dependence, the alternative hypothesis proves the presence of the cross-sectional dependence. If the test statistics computed are smaller than 0.1 at 10% significance level; 0.05 at 5% significance level; and 0.01 at 1% significance level, the main hypothesis is rejected. Accordingly, the alternative hypothesis that accepts the presence of the cross-sectional dependence is accepted.

The MADF Unit Root Test
Spurious regression problems may occur when the analyses are applied in nonstationary panel data models. Therefore, it is important to test the stationarity of variables before the estimation and analyzing the presence of the relationships between the variables when the time dimension increases in the panel data (Tatoğlu, 2017: 4 (Şak, 2018: 262).
One of the second-generation unit root tests used in case of CD is the broadened ADF (MADF) unit root test, developed by Taylor and Sarno (1998). It estimates the system by an approach that recalls the ADF equations and establishes test statistics for the whole panel. It can also be used when the time dimension is bigger than the cross-section.
While the main hypothesis shows that the variable has a unit root, which means it is nonstationary, the alternative hypothesis assumes that the variable does not have a unit root, which means it is stationary. If the t statistics are bigger than the critical value at 5% significance level, the main hypothesis is rejected, and the alternative hypothesis is accepted assuming that the variable is stationary. If the t statistics are smaller than the critical value, the main hypothesis cannot be rejected because the variable is not stationary.

The Homogeneity Test
Before researching the long-term relationship between the variables, it should be determined whether the slope coefficients are homogeneous or heterogeneous to provide consistent results. While the homogeneity shows that the slope coefficients are the same for all units, heterogeneity proves that slope coefficients of at least one of the units are different. To this end, the homogeneity test which was developed by  was applied. Two results were obtained from this test.
Delta_tilde statistics are offered for big samples, and Delta_tilde_adj statistics are offered for small samples (Küçükaksoy and Akalın, 2017: 27).
While the main hypothesis shows that the slope coefficients are homogeneous, the alternative hypothesis assumes that the slope coefficients are heterogeneous. The main hypothesis is rejected if t statistics computed are smaller than 0.1 at 10% significance level; 0.05 at 5% significance level; 0.01 at 1% significance level. Accordingly, the alternative hypothesis is accepted, which means that the slope coefficients are heterogeneous.

The Durbin-Hausman Cointegration Test
The Durbin-Hausman cointegration test, which was developed by Westerlund (2008), was utilized to carry out the analysis of cointegration relation between the variables. This method enables the cointegration analysis to be realized in cases when the explanatory variables are stationary at the level value or the first degree on condition that the dependent variable is stationary at the first degree  (Topal, 2017: 195-196).
While the main hypothesis indicates that there is no cointegration relation between the variables, the alternative hypothesis accepts the presence of the cointegration relation between the variables. The main hypothesis is rejected if t statistics computed are smaller than 0.1 at 10% significance level; 0.05 at 5% significance level; 0.01 at 1% significance level. So, the alternative hypothesis is accepted, which means that there is a cointegration relation between the variables.

The Dumitrescu-Hurlin Causality Test
The Dumitrescu-Hurlin causality test is one of the causality tests that gives effective results if there is a cross-sectional dependence or cross-sectional independence. Moreover, it can also be used in circumstances in which there is a cointegration relation or in which there is not such a relation. The main hypothesis shows that there is not a causality relationship from the first variable to the second variable in this causality test in which the Z-bar and Z-bar tilde statistics are calculated (Dumitrescu and Hurlin, 2012). The Z-bar tilde statistics are more powerful when the Z-bar and Z-bar tilde statistics give different results. Therefore, it is important to interpret the results by considering the Z-bar tilde statistics. In this method, the stationarity of variables is essential. If they are not stationary, they must be turned into stationary variables. The stationary variables should be used when researching the causality relationship.
While the main hypothesis shows that there is no causality relation between the variables, the alternative hypothesis accepts the causality relationship between them. The main hypothesis is rejected if t statistics computed are smaller than 0.1 at 10% significance level; 0.05 at 5% significance level; 0.01 at 1% significance level. So, the alternative hypothesis is accepted, which means that there is a causality relationship between the variables.

Findings
This research study decided which unit root tests would be used in stationarity tests based on whether there were dependencies between the sections. The table below shows the cross-sectional dependence test results. As given in Table 3, the probability values of CD tests are smaller than 0.01. Since T is bigger than N in the research, the CDlm1 and LMadj results can be read. Since the CDlm1 test gives biased results, the LMadj test results are considered. Accordingly, the hypothesis of "there is no dependency between the sections" is denied. There is a cross-sectional dependence between the countries in the panel data, and a one-unit shock affects other countries as well. The interaction between the countries is quite high by the effect of the globalization. Accordingly, the policymakers have to consider the economy policies of other countries at the same time.   The Delta homogeneity test is applied to determine whether the slope coefficients change between the units (Table 5). With reference to these test results, since the probability values of both test statistics are bigger than 0.1, the slope coefficients do not change between the units in the long term; in other words, they are homogeneous. Therefore, the panel statistics should be used instead of the group statistics to examine the relations between the variables.

CONCLUSION
The welfare levels of societies can be increased by borrowing when there are no sufficient opportunities for realizing economic growth and developmental goals. This borrowing can be actualized in two ways which are internal and external borrowings. One of the most important reasons for appealing to external borrowing is the lack of resources. Another important reason is the lack of foreign currency as an instrument of payment to buy the goods and services that public needs. Countries can also appeal to internal borrowing if the public revenues cannot meet the expenses and if the domestic sources are sufficient. Internal and external borrowings positively affect the economic activities of countries. That's why some countries may prefer borrowing from time to time. The money borrowed will create an increasing effect for the national income when it is used in productive areas by increasing the capital accumulation. Interest payments except for the capital need to be made in repayment process. The size of the interest payments may create a reducing effect on the gross domestic product in the long-term.
The relationship between internal debt and external debt in BRICS-TM countries was analyzed by the panel data analysis in this study. The data belong to 2000-2016 period based on the availability of the data. First, the cross-sectional dependence test was applied for more consistent estimations.
According to the cross-sectional dependence test results, there is a cross-sectional dependence between the countries. A one-unit shock that comes to one of the countries affects the other countries as well.
The MADF test was utilized because the cross-sectional dependence was determined. The INT variable is steady at the level value, and the lnGDP and EXT values are stationary when their first differences are computed. The Heterogeneity test was applied to determine whether the slope coefficients change  Ademola et al. (2018) and Daud (2016), the same results conflict with a number of studies including Onakoya and Ogunade (2017), Burhanudin et al. (2017), Nantwi and Erickson (2016), Lau and Kon (2014), Brini et al. (2016), Favour et al. (2017) and Stauskas (2017). The structures of the variables, which include being a different country, country groups and having a different data set, are effective on the conflicts.
To sum up, the results of the current study put forth that the size of internal and external debts does not have a dimension that can negatively or positively affect the economic growth in BRICS-TM countries. In other words, the BRICS-TM countries are not growing through debts. That's why, the cointegration and causality relationships between the variables could not be found.
While there are positive developments in the economic performances of the BRICS-TM countries, the share of the internal and external debts in the gross domestic product increases in China, Mexico and South Africa. These countries need to avoid from the negative effects of internal and external debts to ensure a sustainable economic growth. They should extend their debts to long-term loans and utilise the debts in areas with high added value that positively contribute to the national economy.
It is thought that our study can be a model for future studies in which different variables in actualizing the economic performances achieved by the BRICS-TM countries may be considered and different methods may be applied.