The Estimation of Growth Effects of Regional Economic Integration Between Turkey and New Turkic Republics by Using Fixed Effect Technique (19911998)
30 March 2014 Bayram GÜNGÖR
1. Introduction
as known, Kazakstan, Kyrgyzstan, Turkmenistan, Uzbekistan and Azerbaijan are the Turkic States which are situated in Central Asia and Caucasia. The collapse of Soviet regime has led to a competition between Muslim, national and panTurkish groups. Although Islam functions are as system of norms rather than a unified religion, it is important in the emancipation from communist ideology and cultural influence of Russia. The people of Central Asia and Azerbaijan are Muslims. Even though Central Asia and Azerbaijan regard themselves as belonging to the Muslim World, a secular society is the goal for all republics. Language is an essential agent in nationbuilding, but it also has discirminatory effect. National languages are replacing Russian in all these countries. The Turkic States generally have backward and underdeveloped economies based on export of raw materials in exchange for assistance from Moscow. The collapse of the Soviet system, economy has made very difficult for the Turkic States to export their products (Graeger, www.prio.no/html/oscecentralasia.htm, 03.01.2001). Turkic States have very rich natural sources such as petroleum, natural gas, coal, uranium, copper, lead, zinc, chrome, gold (Frishcenschlager, 1995; Gouliev, 1997; Van der Leeuw, 1997).
An open market is a precondition for entering the modern world economy. Therefore, Turkic States have initiated a gradual transition process, called “from plan to market economy” through economic reform and privatisation to conform to the World Market (Melo, 1996; Aslund, 2001, Gros, 2000). According to the European Bank for Reconstruction and Development (EBRD) data, the most succesful country in the Turkic States is Kyrgyzstan (Gürgen, 2000). The performance of transition process to market economy in Turkic States has been affected by some noneconomic factors which could be called for geographic (landlocked), factor endowment, environmental deterioration, historical and cultural reasons (De Melo, 1997; Stern, 1998; MacKellar, 2000, Jukes, 1998).
Turkic States seek to a new role in the World economy on three different levels. The first is their domestic role, with in Central Asia and Caucasia. The second is their role in relation to the Commonwealth of Independent States (CIS) and Russia, with Russia as the main partner. The third is their role within the World Economy. These three levels comprise the principal framework needed for Turkic States to succeed in the World Economy: to corporate with each other, to cooperate with the Soviet Union Countries and to cooperate with partners outside the CIS (Akimov, 1995).
The possibility of regional economic integration between Turkey and New Turkic States has been also accepted by various scientists throughout the world. Drucker (1993), Thurow (1997), Kubicek (1996) expressed that there is a “Turk Common Market” in which majority culture is Turk, will be formed in the near future among Europe, Middle East and the Central Asia. Wimbush (1994) pointed out that the most important obstacle of the economic integration of the region will be created by Russia, China and Iran. Because, these countries conceive the region as a lebensraum for economic, politics and security respects.
It is possible that an economic union between Central Asia and the outside world can be accomplished by relatively little outside assistance and bye minimal foreign trade turnover. At least in the agricultural and energy sectors, the integrated balance of this union can be useful even without importing food and energy. Ferrous and nonferrous metal, heavy industry of Kazakstan, Uzbekistan and Azerbaijan, as well as light industry of these states add to the integration potential of this kind of economic union. However, this union can only be created on a noneconomic basis by the decision of the Turkic leaders, as there are no close businesses ties among these states. Nevertheless, usual personal contacts are rather extensive, and primitive trade and minor business contacts are active as well. No public opposition to integration is expected (Akimov, 1995)
There are several initiatives concerning regional economic integration in Turkic World, such as Economic Cooperation Organization (ECO), Commonwealth of Independent States (CIS), and Kazakstan, Kyrgyzstan and Uzbekistan Free Trade Area. However, such organizations do not provide the economic autarky to these countries. Can the regional economic integration between Turkey and Turkic States be a option to provide the autarky as a whole?
 Methodology
In the study, the estimation of growth effects of regional economic integration between Turkey and newly independent Turkic States which consist of Azerbaijan, Kazakstan, Kyrgyzstan, Uzbekistan and Turkmenistan under the current conditions was examined. For the estimation, the Ordinary Least Squares (OLS) and Fixed Effect Technique for panel data were used.
In the econometric applications, the most effective way to avoid the problems rooted from using insufficent observations and to enhance the effectiveness of the cross section technique is to use panel data. This study contains both cross section and time series observations. So, in the second approach, under the panel data, it was applied to the OLS technique for the Log Y_{jt}=a+blogTH_{jt}+e_{jt} equation. However, it is impossible to remove the problems completely by running OLS on panel data.
In the study, exclusion of unobservable determinants of individual countries from the estimation causes very important problem in economic respect. Thus, even though the investigator has managed to include a number of true explanatory variables into the model, the coefficents will be biased, and the bias is seen to be some linear combination of the true coefficents of the excluded variables. This of course destroys the conventional b.l.u.e. property of OLS estimators (Johnston, 1984). One way to account for the omitted country specific or time specific effect is the possible inclusion of detailed control variables in all models. When panel data is available, fixed effect technique (the observed variables are transformed by subtracting out the appropriate timeseries means or crosscountry means, and then applying the OLS to transformed data) or least squares dummy variable technique (dummies are used to capture the time specific effects or country specific effects) is another way to eliminate omitted or unobserved country or time specific effect. To increase the likelihood of obtaining unbiased estimates, especially in the fixed effect technique, the equation is specificed such as that there may be some omitted or unobserved variables whose value remains constant across individual countries at a given time, but exhibits variation through time (time effect) or whose value remains constant through time for a given individual countries (individual effects). Obviously, here the main objective in the application of the fixed effect technique will be to capture the possible effects of omitted or mismeasured country specific or integrating country variables which are correlated with the regional economic integration (Karahasan, s.1993). Here, the length of time period (19911998) is not enough to lead the variables of the model to change significantly. Therefore, we necessarily assume that the slope of coefficents of all explanatory variables are constant over the period 19911998. Thus, under this assumption by employing the OLS technique to panel data, following specification is estimated to allow a varying intercept over time or across individual countries and a common vector of slope coefficents over time and across individual countries.
The computation method of the average of the variables under both specifications is performed in the following manner. In the first specification with time effects, each variable is averaged across the countries for given year while in the second spesification with individual country effects, each variable is averaged across eight years for a given country. The difference between two specifications which comes from the construction of the means of the variables which represents time specific effects or individual country effects.
As mentioned above, in the study, it was applied to the observations on the panel behaviours of a panel of decision units. This application was generally realised some part of the succesive time periods.
i=1……….p
t=1………m
n= p.m
Here,
n= observed sample number
t= time (19911998)
p=crosssection
i= number of country
The variables which is used in the model will be represented as follows;
Y_{it}= dependent variable value for unit i in period t
i= 1……….p
t=1………m
X_{jit}= value of jth explanatory variable for unit i in period t
Thus, linear hypothesis would then be
Y_{it}= a+b_{2}X_{2it}+b_{3}X_{3it}+…………………….b_{kit}+U_{it}
Here, it will be assumed that a common set of parameters for all units in all time periods exit. To illustrate some of many possible applications of the model consider some examples.
1. The panel consists of 6 different countries whose Gross Domestic Product (GDP) and explanatory variables are represented Y_{it} and X_{jit.} The values of variables are export, import and volume of foreign trade.
2. The panel consists of a set of countries. The main goal of study is the country and timing of its Y_{it} as a function of the group of explanatory variables to influence welfare.
3. The panel consists of Turkey, Azerbaijan, Kazakstan, Kyrgyzstan, Uzbekistan and Turkmenistan. Y_{it } indicates the GDP in country i in year t. The relevant question is whether the main economic variables such as export, import and volume of foreign trade can adequately explain the variation in Y_{it.}
 Model and Data Set
In the study, the models was formed in 8 versions (Güngör, 1999);
Model 1. Log Y_{jt} = a+blogTH_{jt}+e_{jt}
Model 2. Log Y_{jt} = a+blogTH_{jt}+g_{1}D1+g_{2}D2+g_{3}D3+g_{4}D4+e_{jt}
Model 3. Log Y_{jt} = a+blogTH_{jt}+g_{5}D5+g_{6}D6+g_{7}D7+g_{8}D8+g_{9}D9+g_{10}D10+e_{jt}
Model 4. Log Y_{jt} = a+blogTH_{jt}+g_{1}D1+g_{2}D2+g_{3}D3+g_{4}D4+g_{5}D5+g_{6}D6+
g_{7}D7+g_{8}D8+g_{9}D9+g_{10}D10+e_{jt}
Model 5. Log Y_{jt} = a+blogM_{jt}+llogX_{jt}+e_{jt}
Model 6. Log Y_{jt} = a+blogM_{jt}+llogX_{jt}+g_{1}D1+g_{2}D2+g_{3}D3+g_{4}D4+e_{jt}
Model 7. LogY_{jt}=a+blogM_{jt}+llogX_{jtt}+g_{5}D5+g_{6}D6+g_{7}D7+g_{8}D8+g_{9}D9+g_{10}D10+e_{jt}
Model 8. Log Y_{jt} = a+blogM_{jt}+llogX_{jt}+g_{1}D1+g_{2}D2+g_{3}D3+g_{4}D4+g_{5}D5+g_{6}D6+
g_{7}D7+g_{8}D8+g_{9}D9+g_{10}D10+e_{jt}
_{ }
Y_{jt }: GDP of country j in year t (USD)
X_{jt }: export from country j to Turkey in t year (USD)
M_{jt :}: import from Turkey to country j in year t (USD)
TH_{jt }: volume of foreign trade between country j and Turkey in year t (USD)
D1, D2, D3………D10 : dummy variables.
D1 : for Azerbaijan 1, for others 0
D2 : for Kazakstan 1, for others 0
_{ }D3 : for Kyrgyzstan 1, for others 0
D4 : for Turkmenistan 1, for others 0
D5 : for 1991 1, for others 0
D6 : for 1992 1, for others 0
D7 : for 1993 1, for others 0
D8 : for 1994 1, for others 0
D9 : for 1995 1, for others 0
D10 : for 1996 1, for others 0
The models that are carried out consist of the period of 19911998. Here, the values of export and import of 1991, 1997 and 1998 were estimated by aritmetic means to increase econometric efficiency of the models. Therefore, observation number for each variable were increased to 48.
 Empirical Findings
Dependent
Independent 
Model 1  Model 2  Model 3  Model 4  Model 5  Model 6  Model 7  Model 8 
Log Y  log Y  Log Y  log Y  Log Y  Log Y  log Y  log Y  
Constant  4.890c
(2.522) 
13.128 a
(0.837) 
0.478
(3.107) 
10.346 a
(1.286) 
4.6414 c
(2.393) 
12.222 a
(0.953) 
0.703
(2.977) 
8.846 a
(1.185) 
Log X  0.1987
(0.172) 
0.0090
(0.0735) 
0.1563
(0.1678) 
0.0935 c
(0.0511) 

Log M  0.0951
(0.1855) 
0.1890 a
(0.0563) 
0.3981 c
(0.2092) 
0.0936 c
(0.0477) 

Log TH  0.2639 c
(0.1365) 
0.2205 a
(0.0448) 
0.5244 a
(0.1608) 
0.0771 a
(0.0663) 

D1 

0.4446 a
(0.0967) 
0.3936 a
(0.0688) 
0.5550 a
(0.1431) 
0.5293 a
(0.083) 

D2  1.7740 a
(0.0955) 
1.7596 b
(0.0650) 
1.8511 a
(0.1123) 
1.8350 a
(0.0689) 

D3  0.5356 a
(0.1223) 
0.2908 a
(0.1304) 
0.3428 c
(0.1970) 
0.0009
(0.1565) 

D4  1.7951 a
(0.0974) 
1.7323 a
(0.0709) 
1.8762 a
(0.1219) 
1.8150 a
(0.0780) 

D5  1.1168 b
(0.5418) 
0.4272 a
(0.1079) 
1.1501 b
(0.5375) 
0.5003 a
(0.1000) 

D6  1.3885 b
(0.5973) 
0.2221
(0.1497) 
1.4816 b
(0.6104) 
0.3106 b
(0.1837) 

D7  0.6692
(0.5250) 
0.1957 b
(0.0927) 
0.6957
(0.5202) 
0.2494 a
(0.0865) 

D8  0.4585
(0.5240) 
0.0011
(0.0917) 
0.5105
(0.5205) 
0.0681
(0.0889) 

D9  0.1886
(0.5135) 
0.0522
(0.0810) 
0.2337
(0.5101) 
0.0300
(0.0765) 

D10  0.1553
(0.5112) 
0.0007
(0.0784) 
0.1871
(0.5074) 
0.0375
(0.0749) 

R2  0.101693  0.964108  0.319771  0.986911  0.136038  0.962296  0.357731  0.989161 
F  3.7358 c  155.79 a  1.8132  157.66 a  2.5193 c  119.1033 a  1.8102  167.31 a 
Source: Güngör, 1999
 Conclusion
Model 1 measures the effect of volume of foreign trade between Turkey and Azerbaijan, Kazakstan, Kyrgyzstan, Uzbekistan, Turkmenistan on the their GDPs. The coefficent of regression equation is 0, 2639 and significant statitistically at 10 percent level. It means that when the volume of foreign trade between Turkey and one of the Turkic States increses (or decreases) at 10 percent, the GDP of Azerbaijan, Kazakstan, Kyrgyzstan, Uzbekistan, Turkmenistan increases (or decreases) at 0,26 percent. R2 of the equation is 10 percent level, and little bit low statistically significant. However, when the right side of equation with single independent variable is taken into consideration, it is possible to understand that the reason of this situation is more efficent. F statistics of the equation is 3,73, significant at 10 percent level.
Model 2 indicates that in year t, the economies of all Turkic States which are represented j depend heavily on the foreign trade relations with Turkey. Here, it is assumed that the constant term for each country is different from each other. The coefficent of volume of foreign trade variable is found to be negative. Therefore, the results of Model 2 is in contradiction to Model 1. According to the Model 2, if the volume of foreign trade in year t between the Turkic States that is represented by j, and Turkey decreases as 10 percent, the GDPs of the all countries increases over 0,22 percent. The explanatory variables on the right side of the Model describes the approximately 0,96 of one unit change of dependent variable. F statistics of the model is 155,79, significant at 1 percent level. However, according to the model, on the condition that Turkey and the Turkic States form a regional economic integration,in the short run, economic welfare of the participating countries is affected negatively.
Model 3 indicates that GDP’s of the Turkic States, which is represented by j in year t, are affected by the volume of foreign trade with Turkey. Here, it is assumed that constant term is the same for each country, but different for every year. In the regression equation, the coefficent of the volume of foreign trade variable is 0,5244, statistically significant at 1 percent level. Therefore, when the volume of foreign trade between the Turkic States and Turkey increases at 1 percent, GDP’s of the Turkic States increase at 0,52 percent. F statistics of the model is not significant at any acceptable level, 1,81.
Model 4 indicates the relation between GDP’s of the Turkic States, which is represented by j in year t and their volume of foreign trade with Turkey. Here, it is assumed that the constant term of the regression is different both every country and year. R2 of the equation is 0,99. Therefore, the explanatory variables on the right side of the Model describe approximately 0,99 of one unit change of GDP which is dependent variable. F statistics of the model is 157,66, statistically significant at 1 percent level.
In this model, the effect of volume of foreign trade on the Turkic States is negative as in Model 2. However, if the effects of this variable is examined, Model 4 seems to be more efficent than Model 2. At the same time, the coefficent of the volume of foreign trade is found to be statistically significant at 1 percent level. There is a striking issue in this model. The effects of other variables (out of the volume of foreign trade) on the economy is getting slower. This means that the dependence to the volume of foreign trade is getting higher. That is to say, if the barriers which are imposed on international trade relations in integrated area are removed gradually, economic growth of the participating countries increases as a whole.
Model 5 indicates the relation between GDP’s of the Turkic States, which is represented by j in year t and their export to Turkey and import from Turkey. In this model, it assumed that constant term of the regression is the same in respect of years and countries. R2 of the equation is 0,14. F statitistics value of the model is 2,52, significant at 1 percent level. However, in the short run, export and import have no effect on economy separetely. Because, the coefficents of both export and import are not significant at any acceptable level.
Model 6 indicates the relation between GDP’s of the Turkic States, which is represented by j in year t and their export to Turkey and import from Turkey. Here, it is assumed that the constant term is not the same for every country. R2 of the equation is 0,96. Therefore, the explanatory variables on the right side of the Model describe the approximately 0,96 of one unit change of GDP which is dependent variable. F stastistics value which is calculated is 119,1, statitistically significant at 1 percent level. According to the model, Exports among the Turkic States do not affect their economies positively or negatively. However, the import from Turkey to Turkic States affect the economic welfare of their economies negatively. This situation is statitistically significant and appropriate to the theory. Nevertheless, the coefficents of dummy variables, which are added to the model in order to catch the country effect, are calculated statitistically significant at 10 percent level. In this respect, even if the Turkic States are the same economic and political conditions, there are some important differences among themselves. There are also same important differences between Turkey and as a group of Turkic States.
Model 7 indicates the relation between GDP’s of the Turkic States, which is represented by j in year t and their export to Turkey and import from Turkey. Here, that the constant term is the same for each country, but different for every year were assumed. The coefficent of import variable, which demonstrates the import from Turkey to the Turkic States, is calculated as 0,398, statitistically significant at 10 percent level. It means that if one of the Turkic States increases its import from Turkey at 1 percent level, The GDP’s of the country increase at 0,40 percent level. Explanatory power of the equation is calculated at 36 percent. The most interesting finding of the model is that export among the Turkic States has no impulse on their own economies, but import from Turkey has some important effect on their economies. This Model is more detailed type of the Model 3. Therefore, the results of this are similar to the Model 3.
Model 6 indicates the relation between GDP’s of the Turkic States, which is represented by j in year t and their export to Turkey and import from Turkey. Here, it is assumed that the constant term is not the same for every country and year. The coefficent of export variable from Turkic States to Turkey is 0,0935, statistically significant at 10 percent level. It means that if one of the Turkic States increases its export to Turkey at 1 percent, the GDP’s of the countries decrease at 0,09 percent level. Likewise, if one of the Turkic States increases the their import from Turkey at 1 percent level, the GDP’s of the countries decrease at appraximately 0,936 percent level. R2 of the equation is 0,99. The explanatory variables on the right side of the Model describe the approximately 0,99 of one unit change of dependent variable. F statitistics of the model is 167,31, significant at 1 percent level.
The explanatory power of this model is the best in all models. If it is examined to the theoric respect, import causes to foreign exchange leakage to the outside and export causes to foreign exchange leakage to inside. Thus, export affects the domestic economy positively. But, import affects the domestic economy negatively. Here, what is the more important is that, when it is examined for export and import, the sizes of the eport and import are close to each other. As a result, export and import have neutral effect on economy. As in Model 4, here, The economic growth which is come from the out of the export and import is getting smaller year by year. Therefore, in the longrun, regional economic integration which will be realised between Turkey and New Turkic States will be more crucial in order to develop the economies of the participating countries as a whole than that of ever before.
In conclusion, current and realised volume of foreign trade between Turkey and New Turkic States has no expected impacts on their own economic growth wholly. However, that the data which are used in the models consist of the shortrun and dummy variables are getting smaller indicates that the regional economic integration provides the positive growth effects on the participating economies in the longrun.
References
——————————————– 
TİKA Avrasya Dosyası, 1998/2, Ankara 
——————————————– 
Orta Asya Türk Cumhuriyetlerinde Yatırım İmkanları ve Mali Sektör İnceleme Raporu (Basılmamış), T.C. Türk Cumhuriyetlerinden Sorumlu Devlet Bakanlığı Hizmetler Sektörü Komitesi, 1999, Ankara 
——————————————– 
DİE ,1997 
——————————————– 
DİE, TSU 00197, Yayın No.2, 9 Kasım 1998 
——————————————– 
Azerbaycan, İGEME, 1998, İstanbul 
——————————————– 
Azerbaycan Ülke Profili, Mevzuat ve Türk Girişimcileri , İTO, Yayın No: 199734, 1997, İstanbul 
Akimov, A., 
CENTRAL ASIA: Conflict, Resolution, and Change, Edited by Ronald Z. Sagdeu and Susan Eishenower, The Center for Political and Strategic Studies, 1995 
Aslund, A., 
“Transforming Post Communist Political Economies: Possible Future Directions for Economies in Transition”,www.nap.edu.readingroom/books/transfom/ch18htm., 2001 
Bozkurt, V., 
Avrupa Birliği ve Türkiye, Alfa Basım Yayım Dağıtım, 1997, İstanbul 
Frischenschlager, A., 
“Siyasal ve Ekonomik Bağımsızlık Yolundaki Türkmenistan”, Avrasya Etütleri, Cilt.2, Sayı.3, 1995, ss.3850 
Gouliev, R., 
Petrol ve Politika, Ar Matbaacılık, 1997, İstanbul 
Graeger, N and Kolsto P., 
“Central Asia”, www.prio.no/html/oscecentralasia.htm, 03.01.2001 
Gros, D., and Suhrcke, M., 
“Ten Years After: what is special about transition countries”, EBRD WP:56, August 2000 
Gumpel, W., 
“Orta Asya Türk Cumhuriyetlerinde Ekonomik ve Politik Gelişme”, Avrasya Etütleri, Cilt.1, Sayı.2, 1994, ss.1546 
Güngör, B., 
Türkiye ile Yeni Türk Cumhuriyetlerinin Ekonomik Entegrasyonu Üzerine Bir Deneme, Basılmamış Doktora Tezi, KTÜ, SBE, Trabzon, 1999 
Gürgen, E., 
“Central Asia Achievements and Prospects”, Finance and Development, International Monetary Fund, Volume.37, Number.3, 2000 
Gürgen, E; Snoek, H; Craig, J; McHugh, J; Izvorski, I., and van Rooden, R., 
“Economic reforms in Kazakstan, Kyrgyz Republic, Tajikistan, Turkmenistan and Uzbekistan”, IMF Occasional Paper.183, 1999 
Johnston, J., 
Econometric Methods, Mc Graw–Hill Book Company, 1984, New York 
Jukes, G.J.; Naurzhanov, K., and Alexandrov, M., 
“Race, Religion, Ethnicity and Economics in Central Asia”, The Slavic Research Center, 1998 
Karahasan, N., 
Effect of Unionization on the Firm Profitability: The U.S. Manufacturing Firms, 19861988, Phd. Thesis, Texas Tech University, 1993, USA 
Kubicek, P., 
“Orta Asya’da Topluluklararası Etnik İlişkilerin İdaresi: Teori ve Uygulama”, Avrasya Etütleri, Cilt.3, Sayı.3, 1996, ss.85101 
MacKellar, L; Wörgötter,A; Wörz, J., 
“Economic Development Problems of Landlocked Countries”, Transition Economies Series, Institute for Advanced Studies, Wienna, January, 2000 
Melo,M.D;Denizer,C, and Gelb A., 
“Patterns of From Plan to Morket”, The World Bank Economic Rewiev, Vol.10, Number.3, 1996 
Melo,M.D;Denizer,C; Gelb A. and Tenev, S., 
“The Role of Initial Conditions and Policies in Transition Economies”, WB, IFC,WP,1997 
Stern, N., 
“The Future of Economic Transition”, EBRD WP:30, 1998 
Thurow, L.C., 
Kapitalizmin Geleceği (Redaksiyon:Demirtaş, S., ve İlseven, N.), Sabah Kitapları, Çağdaş Bakışlar Dizisi: 14, 1997, İstanbul 
Van der Leeuw, C., 
“Oil in AzerbaijanBlack Gold and Vile Poison”, CAUCASUS and unholy alliance (Edited by Antero Leitzinger), 1997, pp.307314 
Wimbush, S.E., 
“Rusya’dan Sonra Rusya”, Avrasya Etütleri, Cilt.1, Sayı.2, 1994, ss.5361 
* This article has been published in Geopolitics of Central Asia in the Post Cold War Era: A Systematic Analysis (Edited by Ertan Efegil), Research Centre for Azerbaijan and Turkestan, SOTA, pp.507518 Haarlem, Netherlands, 2002.
[1] Associate Prof. Dr., Member of Faculty, Department of Economics, Karadeniz Technical University, 61080 Trabzon/TURKEY
POST YOUR COMMENTS
You must be logged in to post a comment.