“Does the Commonwealth of Independent States (CIS) promote trade among its members? Case study of Georgia”

(1)

“Does the Commonwealth of Independent States (CIS)

promote trade among its members? Case study of Georgia”

Written by Semjons Hazanovs (S2966905)

Supervised by Dr Abdul Erumban

(2)

Declaration

I declare that this dissertation is my own work and that it was composed by myself. Following academic conventions, I have made due acknowledgement of the work of others.

Signed:

(3)

Abstract

This dissertation investigates the effects of Commonwealth of Independent States (CIS) membership on bilateral trade flows between Georgia and its trading partners. Using a panel of 32 country pairs in the period 1996 - 2015, a gravity model of trade is constructed to analyse these effects. The research finds that CIS membership does not affect the volume of bilateral trade when both countries in the pair are members. However, there is a negative effect when only one of the two countries in the pair is a member.

(4)

(5)

Chapter 1. Introduction

This chapter will set the background and the overall context of the research, introduce its aim, as well as outline the structure of the dissertation.

1.1. Rationale of the research

Many economists believe that countries may benefit from being a member of a regional organisation as it creates more trade for its members (Ravenhill, 2014; Wild et al, 2005). Many regional organisations (e.g. NAFTA, CARICOM,

MERCOSUR, ASEAN etc.) have been heavily researched, yet one particular organisation seems to have remained ‘under the radar’ - the CIS. The

Commonwealth of Independent States (CIS) is a regional organisation that was founded in 1991 by Russia, Belarus and Ukraine. There has been a resurgence of interest in the CIS as the year 2016 marks its 25th anniversary. Today, the CIS consists of 11 equal sovereign states with a combined population of over 280 million people. The objective of the organisation is to engage its members in “further development and strengthening of the relations of friendship, good neighbourliness, inter-ethnic accord, trust, mutual understanding and mutually advantageous cooperation among the member states” (Kubicek, 2009).

(7)

recent years. There seems to be a general lack of literature that would provide a coherent view on whether the CIS presents any considerable trade advantages to its members. The most recent attempt to study this was made by Freinkman et al (2004), however significant amount of time has passed and sufficient amount of data has become available to allow us to conduct new research and draw certain conclusions regarding the impact of this regional organisation on member countries. Hence, this dissertation will provide an important

contribution to this area of research.

1.2. The case of Georgia

In particular, this dissertation aims to understand the trade effects of CIS membership on Georgia. Figures 1 and 2 indicate Georgia’s trade with the CIS relative to the rest of the world. From these figures, it is evident that the CIS accounts for a large share of Georgia’s exports and imports. Moreover, it also that Georgia has become more open to trade with the rest of world, and indeed, Georgia has moved up in terms of ease of ‘trading across borders’ in the World Bank’s Doing Business rankings (Hartwell, 2013). Figures 3 and 4 illustrate the share of exports and imports between Georgia and current CIS members. It is peculiar that the import shares remained stable throughout the decades, but the export shares witnessed significant shifts. Figure 5 shows an increasing trend for total trade as a percent of GDP. In recent years this trend has has been above 100 percent which goes to show that Georgia has become heavily dependant on international trade. Figure 5 also illustrates increasing trend in both exports and imports, yet the gap between them shows that the country has been suffering from a trade deficit for a while. Considering the above

(8)

1.2. Structure

The dissertation opens with the review of literature (Chapter 2) where the reader learns the benefits of joining a regional organisation. Moreover, the literature review will explore relevant evidence from academic research on intra-regional trade dynamics within the CIS and raise questions about its perceived economic importance. The methodology chapter (Chapter 3) will discuss in detail the research design, model and methods of data collection. Next, the findings and analysis chapter (Chapter 4) will introduce the reader to the

empirical results of the research, their analysis and the outcomes that are linked to the original hypotheses of the paper and that contributed to answering the research question. The dissertation closes with key results highlighted from the analysis, limitations are presented and future research opportunities are

(9)

Chapter 2. Literature review

This chapter will explore key issues that underpin the topic. In particular, the chapter will discuss and critically assess relevant theoretical literature and empirical evidence on regional trade, and consequently, derive testable hypotheses.

2.1. Motives for joining a regional organisation

There are a lot of reasons why countries may choose to pursue their foreign economic policy objectives by joining a regional organisation. According to Krugman (1993), these reasons are not solely economic as international trade, in essence, is a tool of political economy. Ravenhill (2014) summarises the motivations for joining a regional organisation in two broad categories:

economic and political. Economic reasons include access to a common market, higher potential for attracting FDI, as well as the possibility to protect

uncompetitive/politically sensitive industries that would otherwise be driven out by global competition. From a political perspective, it is easier to negotiate a single agreement with a number of states than to carry out a series of individual negotiations with each state (Ravenhill, 2014). At the same time, it is harder to reach a worldwide agreement in the case of multilateral trade liberalisation arrangements, hence regional arrangements are often preferred (Black et al, 2009). Common tools that are used in regional organisations are customs unions and free trade agreements that have been shown as successfully being able to boost trade among members.

2.2. The Commonwealth effect

By no surprise, many researchers often compare the CIS to the British Commonwealth of Nations (henceforth, the Commonwealth) as both share many traits (Voitovich, 1993; Baslar, 1998). Lundan and Jones (2001) define the Commonwealth as “a de facto trading bloc, which in contrast to existing de jure trading blocs, bridges a wide range of countries not only in terms of

(10)

find that there is “a high propensity on behalf of member nations to trade and invest with each other, with the expectation that this tendency should be clearly higher within the Commonwealth than for comparable non-member nations”. In other words, despite other factors (e.g. free trade zone, customs union etc) more trade is conducted within a country-pair where both parties are members of the Commonwealth in comparison to country-pairs where only one party is a member ceteris paribus . Lundan and Jones (2001) coined the term as the 1

‘Commonwealth effect’. Subsequently, a question arises - does this effect hold for the CIS?

Voitovich (1993) and Baslar (1998) agree that the CIS, to a certain extent, is similar to the Commonwealth. Indeed, both had a common purpose, which was to create a minimally institutionalised intergovernmental organisation that would manage the consequences of decolonisation of the British Empire in one case, and the collapse of the Soviet Union in case of the other. In theory, it is logical to assume that this effect would hold true for the CIS given its resemblance to the Commonwealth. Freinkman et al (2004) contribute to this idea as they find that there is, indeed, a strong trade creating effect within the bloc. However, contrary to the CIS, Krugman (1991) describes the Commonwealth as an ‘unnatural trading bloc’ whereby not all member countries are part of the same region. Because of this, member countries that are far apart will carry the burden of higher transportation and communication costs. Therefore, one could possibly expect a higher magnitude of intra-regional trade in the CIS than in the

Commonwealth due to closer proximity, and as a result, one could observe a stronger (and perhaps, more sustainable) ‘Commonwealth effect’ in the case of CIS.

2.3. Understanding CIS trade dynamics

Kurmanalieva and Vinukurov (2011) identify three key drivers of intra-regional trade within the CIS. The first one, the ‘home bias’ effect, is defined as the tendency to spend more on domestic products than on foreign products (Black et al, 2009). This effect can be explained by either international differences in tastes, or by high transaction costs (e.g. tariffs and other trade barriers). Given

A similar conclusion was made by Bennett and Sriskandarajah (2011).

(11)

the fact that the majority of post-Soviet countries inherited similar economic structure, it is assumed that they should trade more with one another as transaction costs should be lower than with the rest of the world (Kurmanalieva and Vinukurov, 2011). The second one, the ‘holdup effect’, is a situation that occurs in a mutual project when one party gains advantage over the other party that has spent resources first and now has more to lose if no agreement is reached (Wes, 2000). In this context, trade patterns of landlocked Central Asian countries rely heavily on their relationships with Russia being their transit coastal neighbour (Kurmanalieva and Vinukurov, 2011). Finally, the authors identify the ’holding together’ effect which postulates that a higher degree of intergovernmental policy coordination is more likely between countries that are part of a single political entity (Libman and Vinokurov, 2010). However, it seems difficult for former Soviet states to change the direction of their trade strategies in presence of past social and economic ties between them (Freund and Djankov, 2000).

2.4. Relevance of the modern CIS

(12)

inability to make strategic decisions at a CIS-wide level . While regional 2

integration offers many benefits, they depend a lot on the degree of political commitment of member states (Mwasha, 2011). Sakwa and Webber (2010) stress that policy coordination is essential for any regional organisation to facilitate effective cooperation and development. One thing to consider here as well is the role of Russia in the CIS. Freinkman et al (2004) assert that overall CIS trade dynamics are determined under the Russian hegemony which hardly comes to any surprise. Alexianu (2015) argues that Russian hegemony has a negative influence on institutional development in post-Soviet states by promoting non-transparent business practices , and by isolating from external 3

(i.e. Western) influences. All of this factors in sum may have an impact on intra-regional trade as well as on trade with countries outside the bloc.

Finally, it is important to discuss whether countries should embrace regionalism or multilateralism (Bhagwati, 1993). Multilateralism implies “collaborating with all other major countries when taking major decisions, rather than a country acting on its own - unilaterally - or partnering with just one other country (or set of countries) - bilaterally” (Conway, 2009). It is given that countries may maximise their welfare provided they trade on a non-discriminatory basis (i.e. multilaterally) as advocated by the WTO. Regionalism, however, is often being criticised for distorting the allocation of resources that can lead to reducing global welfare (Ravenhill, 2014). Sakwa and Webber (2010) find that the prevalent form of intergovernmental interaction in the CIS has been bilateralism which allowed many members to adapt a more discriminating policy towards trade. Should a regional organisation experience an increase in the volume of bilateral trade between its members this would occur through two mechanisms: trade creation and trade diversion (Hoekman et al, 2002). Trade creation is 4

Currently, CISFTA includes Armenia, Belarus, Kazakhstan, Kyrgyzstan, Moldova, Russia,

2

Ukraine and Uzbekistan; EAUC includes Armenia, Belarus, Kazakhstan, Kyrgyzstan and Russia.

Kudina and Jakubiak (2012) find that corruption and bureaucracy are common problems

3

among CIS members that they have inherited from the former USSR which have been shown to have a negative impact on international trade.

Assumptions are based on the the classic argument made by Viner (1950) on welfare

4

(13)

described as an effect whereby domestic goods are displaced by cheaper intra-regional imports. Trade diversion, on the other hand, describes a situation whereby extra-regional imports are displaced by intra-regional imports (Black 5

et al, 2009). Should trade diversion outweigh trade creation, then the net effect of the regional organisation on its members’ welfare will be negative (Ravenhill, 2014; Todaro and Smith, 2006).

2.5. Hypotheses

Considering the review of literature above, the following hypotheses will be put to a test:

• Hypothesis 1: CIS membership positively affects the volume of bilateral trade

when both countries in the pair are members.

• Hypothesis 2: CIS membership negatively affects the volume of bilateral

trade when only one country in the pair is a member. 

Should the importer always chose the lowest-cost import, intra-regional goods will be

5

(14)

Chapter 3. Methodology

This chapter discusses the research design, as well as explains in detail the economic model used for testing the hypotheses put forward. Also, the chapter outlines the methods of data collection and analysis.

3.1. Research design

Quantitative research is mainly associated with an experimental research design as it studies “relationships between variables, which are measured numerically and analysed using a range of statistical techniques” (Saunders et al, 2009). However, as the research will focus on Georgia, elements of a case study strategy are going to be present. Yin (2009) claims that using case studies is one of the most challenging of all the social science research activities, yet a case study is preferred when the focus of investigation is on a phenomenon in a real-life context. Moreover, case studies can also allow for the identification of new ideas, challenging the existing theory, explaining the nature of a problem, providing new research hypotheses, and, finally, putting it all into practice.

3.2. Model specification 3.2.1. Gravity equation

(15)

(2015) show that the gravity model is highly applicable for a single country case study. For the purposes of this research, the following gravity equation is used :6

log(TradeVolijt) = β0 + β1log(ProdGBPijt) + β2log(ProdGBPpcijt) +

+ β3log(Distanceij) + β4(Border) + β5(Soviet) +

+ β6(RusLang) + β7(CIS1ijt) + β8(CIS2ijt) + ε

3.2.2. Variables

Summary statistics of dependent and independent variables is presented in Table 1 below.

Table 1. Summary statistics.

The dependent variable in the above model is TradeVolijt which is the volume of

bilateral trade between country i (Georgia) and country j (trading partner) at time

t . According to Table 1, TradeVol7 ijt has a mean of 2.21 whilst ranging from -6.50

to 6.89 which clearly suggests the intensity of trade between Georgia and its trading partners. In generalised gravity models, trade volume is a function of the countries’ economic mass (GDP), population, geographical distance and a set

All continuous variables are expressed in a logarithm form.

6

Volume of bilateral trade is calculated as the average of imports and exports.

7

Variable Obs Mean Std. Dev. Min Max

(16)

of dummies (Martinez-Zarzoso and Nowak-Lehmann, 2003). The main explanatory variables used in the gravity equation are the following:

• ProdGBPijt - product of real GDP of countries i and j at time t. According to

Baldwin and Taglioni (2006), is the most common choice of measure for economic mass, and it indicates the multiplicative effect of production level in exporting country and income level in importing country. In other words, on one hand, large economies tend to have higher levels of production due to economies of scale. On the other hand, large economies also tend to import more thanks to higher income levels (Martinez-Zarzoso and Nowak-Lehmann, 2003; Sohn, 2005). Therefore, the product of GDP should have a positive impact on bilateral trade.

• ProdGBPpcijt - product of real GDP per capita of countries i and j at time t.

Apart from GDP, another common measure of economic mass that is used in gravity equations is population. Similarly to Doumbe and Belinga (2015), this paper uses the product of GDP per capita which measures the multiplicative effect of per capita income of the two countries in the pair. According to them, the coefficient should be positive.

• Distanceij - distance between the capitals of countries i and j . The coefficient 8

is expected to be negative as more trade is likely to occur between countries within close proximity due to lower transportation costs (Das, 2004).

It is essential to expand the standard model by adding a number of control variables that may potentially affect trade. Spector et al (2011) suggest that researchers should explicitly explain the role of control variables and match hypotheses precisely to both the choice of variables and the choice of analysis. This is because in most cases inclusion of such variables “implicitly assumes that the control variables are somehow either contaminating the measurement of the variables of interest or affecting the underlying constructs, thus distorting observed relationships among them” (Spector, 2011). Therefore, apart from justifying the inclusion of controls it is important to check whether the interpretation of the main coefficients are biased if controls are not included, and check whether the results are similar without the controls. Anderson and

Distance is measured in kilometres.

(17)

van Wincoop (2004) argue that one needs to control for the relative trade costs that are faced by countries. Bussiere et al (2005) discuss which control variables are most frequently used in the gravity model that have some degree of influence on trade costs. Hence, the following control variables were included:

• Border - binary dummy variable which is 1 if countries i and j share a land border. The costs of doing business abroad for countries having a common border are assumed be lower and the coefficient for this dummy is deemed to be positive.

• Soviet - binary dummy variable which is 1 if country j is a former Soviet

republic. It is argued that countries which were once a colony in the past tend to trade more with one another due to historically established trade ties. Following the insights of Libman and Vinokurov (2010), the same argument holds for former Soviet republics, hence it is expected that the coefficient will be positive.

• RusLang - binary dummy variable which is 1 if the official language in country

j is Russian. It is said that common language has a positive impact on trade

(Frankel, 1997), and although Russian is the official language of the CIS it does not have the same status in many former Soviet states (including

Georgia). Yet Russian is still widely spoken among their citizens, therefore it is assumed that the coefficient will be positive.

(18)

• CIS1ijt - binary dummy variable which is 1 if either country i or j is a member of

the CIS at time t .9

• CIS2ijt - binary dummy variable which is 1 if both country i and j are members

of the CIS at time t.

Should either of the hypotheses hold true, the sign of CIS1ijt will be negative and

the sign of CIS2ijt will be positive.

3.3. Data collection

The panel consists of 640 observations for 32 country pairs. The sample 10

includes 11 countries that are current CIS members, and 21 countries are used as a control group . The data is collected for the period 1996 - 2015 with an 11 12

annual interval. Since the research is based on a case study design, data triangulation is used (i.e. collection of variables from multiple sources).

Triangulation is a useful tactic that is used within one study in order to check if the findings provide solid evidence for the research. It has also been mentioned by Perry and Coote (1994) that the complexity of the external reality and the possible limitations of a researcher’s mental capacity make triangulation vital for any research. Case study methodology researcher Yin (2009) shares the point and suggests that using multiple sources of data collection will help maximising the reliability of the analysis. Saunders et al (2009) also agree that a case study design is seen applicable when it uses different data that complement each other and verify that the collected information is really telling you what you think it is telling you. The data for GDP and GDP per capita was obtained from the IMF World Economic Outlook database, whereas export and import data used

CIS1ijt takes the value of 1 when (a) country i is a member of the CIS and country j is not

9

in the period 1996 - 2008, and (b) country i is not a member of the CIS and country j is in the period 2009 - 2015.

Countries included in the sample - Argentina, Armenia, Australia, Azerbaijan, Belarus,

10

Brazil, Canada, China, Estonia, France, Germany, India, Indonesia, Italy, Japan,

Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Mexico, Moldova, Russia, Saudi Arabia, South Africa, South Korea, Tajikistan, Turkey, Turkmenistan, Ukraine, United Kingdom, United States, Uzbekistan.

Control group consists of 3 countries that are not part of the CIS but are former Soviet

11

republics, and 18 countries that are part of the G20. 01/01/1996 - 31/12/2015

(19)

(20)

Chapter 4. Findings and analysis

In this chapter the data collected and its analysis will be presented and discussed in detail, supported by relevant tables and figures where appropriate. All tests and estimations are done in Stata.

4.1. Robustness checks13

4.1.1. Normality

According to Hill et al (2012), normally distributed residuals help achieving more accurate results, however if the residuals are not normally distributed then it may lead to biases. A Skewness/Kurtosis formal test is performed (see Appendix 1) which shows that most of the variable residuals are not normally distributed. The histogram of residuals (see Appendix 2) confirms that residuals are not normally distributed as the graph is not perfectly bell-shaped - the residuals are slightly skewed to the right with a kurtosis close to that of normal distribution. Although the residuals do not show full normality there is no substantial deviation from it, therefore the results are not affected a lot by it.

4.1.2. Multicollinearity

Multicollinearity is a problem that occurs when two or more independent variables are highly correlated (Studenmund, 2001). When such a problem occurs it is more difficult to obtain credible results in the regression analysis (Hill et al., 2012). A correlation matrix of coefficients is constructed to check for this (see Appendix 3), and it shows that there are a couple of independent variables that are indeed particularly correlated. Normally, it would be advised to omit variables that are suspected to cause the problem of multicollinearity, however Studenmund (2001) mentions that the issue may be disregarded as long as standard errors remain small. In order to check if multicollinearity is going to be an issue for the model, a variance inflation factor (VIF) test is done (see Appendix 4). If the test shows that the mean VIF value is below 10, then it

There are some econometric issues that need to be addressed to help assessing the

13

(21)

means that there is not evidence to support the presence of multicollinearity (Chatterjee and Hadi, 2012). Test results indicate that there are no multicollinearity issues for the model not to work well. Therefore, there is no need to omit variables that were previously suspected in causing econometric issues.

4.1.3. Serial correlation and heteroskedasticity

The model should be tested for serial correlation and heteroskedasticity which can lead to overestimation in the sense that some estimates could be seen as significant when in reality they are not (Hill et al, 2012). The Wooldridge test for autocorrelation in panel data is carried out with the Breusch-Pagan test for heteroskedasticity (see Appendix 5 and 6); both tests indicate the presence of autocorrelation and heteroskedasticity in the data. In order to adjust the regression results to them, option “robust” is added in the end of the Stata command.

4.1.4. Estimation methods

One of the challenges in panel data modelling arises when working with data that exhibits group-level variation. In these cases, using a pooled OLS model can lead to poorness of fit and misleading results as it does not account for heterogeneity in the sample (Clark and Linzer, 2015). Because it is assumed that variation across countries has influence on bilateral trade between them it is preferred to employ individual effects. The choice between pooled OLS and individual effects model is validated by the Breusch and Pagan Lagrangian multiplier test (see Appendix 7).

(22)

effects are more likely to provide accurate regression results. The Hausman test is carried out in order to determine model specification (see Appendix 8); results suggest that a random effects model should be used which is consistent with assumptions made above.

4.2. Empirical results

Table 2 reports empirical results of estimating the gravity equation as specified in Chapter 3 using random effects. Column (1) reports empirical results of the standard gravity equation where trade volume is regressed on the main explanatory variables - products of GDPs, product of GDPs per capita and distance between capitals. Subsequent columns have one control variable added at a time until the final ‘augmented’ gravity equation is reached in Column (5). It is worth to point out at this stage that the R-squared value in Column (5) is 0.689 which indicates good overall performance of the model as it explains 68.9 percent of variation in the dependent variable.

As mentioned in Chapter 3, product of GDPs is expected to positively affect the volume of bilateral trade, and it is statistically significant at 1-percent level. Surprisingly, the product of GDPs per capita is negative contrary to the paper by Doumbe and Belinga (2015). This would suggest that an increase in GDP per capita of Georgia or any of its trading partner causes a decrease in the volume of bilateral trade , however the coefficient is very small and is 14

statistically insignificant in order to make this argument. In addition, Russian language and common land border dummies do not show any statistical significance which seems to be interesting. Distance, has a negative effect which is significant at a 10-percent level which proves that, indeed, ‘proximity promotes trade’ (Das, 2004). Soviet background dummy coefficient is positive and is also statistically significant at a 10-percent level. The coefficient is line with findings of Libman and Vinokurov (2010) which means that Georgia is more likely to have higher volumes of trade with former Soviet republics. This is well predicted as one to expect to capitalise on past trade linkages.

Recall Figures 3 and 4 where it is shown that since 1996 Georgian trade has increased

14

(23)

Table 2. Regression results  15

Robust standard errors in parentheses

(24)

The results that this paper is most interested in are variables CIS1 and CIS2. As expected, CIS1 shows a negative sign and is statistically significant at a 5-percent level. The interpretation of the result indicates that Hypothesis 2 holds true in the case of Georgia. Also, CIS2 shows a positive sign as expected, but surprisingly it does not demonstrate any statistical significance and the value of the coefficient is rather small. Therefore, Hypothesis 1 is not supported. The above mentioned analysis of results suggests that, in the case of Georgia, CIS membership does not increase the volume of bilateral trade when only one country in the pair is a member, but it does not have any significant effect when both countries are members.

(25)

Chapter 5. Conclusion

This chapter provides a summary of the findings that were obtained in the research and final conclusions are drawn from the discussion.

5.1. General remarks

The aim of this research was to find out how CIS membership affects bilateral trade flows between Georgia and its trading partners in the period 1996 - 2015. The results reveal that when Georgia was a member of the CIS, its volume of trade with countries outside CIS was significantly lower than when it is outside the CIS. However, when Georgia was in CIS, its volume of trade with other CIS countries is not significantly different from when it is outside the CIS. This shows that despite not having implemented such regional integration tools as bilateral free trade agreements and customs union, there has been no

Commonwealth effect observed for the CIS in the case of Georgia during the period in the sample. Instead, Georgia has been capitalising on past trade linkages as all current and past CIS members used to belong in the former Soviet Union. These findings mimic points drawn from the literature review in Chapter 2. So, overall joining the CIS only had a negative impact on Georgia’s trade, as it limited Georgia’s trade possibilities with non-CIS countries, while did not improve its trade with CIS countries.

5.2. Limitations

The main constraint that has presented a challenge in executing the analysis is poor quality of trade statistics in CIS countries (mostly concerning the

(26)

quality of trade data through flawed imports evaluation that introduces a bias toward the under-valuation of imports” (Freinkman et al, 2004).

On a minor note, some authors may argue that a single case study will provide little evidence for scientific generalisation. However, Yin (2009) argues that the case study research can include both single and multiple case studies, where the single one is appropriate when the case study is a representative case which aims at extending theoretical views as it can give valuable insight into the observed phenomenon. Although this study sheds some light into the impact of CIS membership on trade, further research expanding into more countries would indeed help understand this subject better.

5.3. Future research opportunities

As mentioned above, it is evident that there is a need for future research to be conducted. The main topics for future should be researching the effects of CIS free trade agreements and customs union on trade once enough data has been accumulated in time.

Results of this dissertation tell us little (if anything) about the welfare effects of CIS membership. It is unclear, in general, whether trading blocs create trade diversion or creation (Krugman, 1993). One possible extension to this study in future would be researching these effects.

Another suggestion is linked to Alexianu (2015) where he establishes that Russia has a negative impact on institutional development in post-Soviet states. It would be interesting to study what impact does it have on the institutional framework of the CIS and how it translates onto other members in terms of trade.

(27)

References

• Alexianu, M. (2015) ‘Post-Soviet Transition: The Impact of Russian Influence on Institutions’, International Finance Review, 16.

• Anderson, J.E. and Van Wincoop, E. (2004) ‘Trade costs’, Journal of Economic literature, 42(3).

• Baldwin, R. (2006) ‘Multilateralising regionalism: spaghetti bowls as building blocs on the path to global free trade’, The World Economy, 29(11).

• Baldwin, R. and Taglioni, D. (2006) ‘Gravity for dummies and dummies for gravity equations’, National Bureau of Economic Research.

• Baslar, K. (1998) ‘The Concept of the Common Heritage of Mankind in International Law’, Developments in International Law, 30.

• Bennett, J. and Sriskandarajah, D. (2011) ‘The Commonwealth Effect Revisited’, The Round Table, 100(412).

• Bhagwati, J. (1993) ‘Regionalism and multilateralism: an overview’, New dimensions in regional integration, 22.

• Black, J., Hashimzade, N. and Myles, G. (2009) ‘A Dictionary of Economics’, Oxford University Press.

• Bussiere, M., Fidrmuc, J. and Schnatz, B. (2005) ‘Trade Integration of Central and Eastern European Countries: Lessons from a Gravity Model’.

• Chatterjee, S. and Hadi, A. S. (2012) ‘Regression Analysis by Example’, Wiley. • Freund, C.L. and Djankov, S. (2000) ‘Which Firms Do Foreigners Buy?:

Evidence from the Republic of Korea’, World Bank Publications.

• Clark, T.S. and Linzer, D.A. (2015) ‘Should I use fixed or random effects?’, Political Science Research and Methods, 3(02).

• Conway, E. (2009) ’50 Economics Ideas You Really Need to Know’, Quercus Publishing.

• Das, D.K. (2004) ‘Regionalism in global trade’, Edward Elgar Publishing. • Doumbe, E.D. and Belinga, T. (2015) ‘A Gravity Model Analysis for Trade

between Cameroon and Twenty-Eight European Union Countries’, Open Journal of Social Sciences, 3(08).

(28)

• Freinkman, L., Polyakov, E. and Revenco, C. (2004) ‘Trade performance and regional integration of the CIS countries’, World Bank Publications.

• Greene, W.H. (2008) ‘Econometric analysis’, Prentice Hall. • Gujarati, D.N. (2003) ‘Basic Econometrics’, McGraw-Hill.

• Hartwell, C.A. (2013) ‘A Eurasian (or a Soviet) Union? Consequences of further economic integration in the Commonwealth of Independent States’, Business Horizons, 56(4).

• Hill, R.C., Griffiths, W.E. and Lim, G.C. (2012) ‘Principles of Econometrics’, John Wiley and Sons.

• Hoekman, B., Mattoo, A. and English, P. (2002) ‘Development, trade, and the WTO: a handbook’, World Bank Publications.

• Kang, Y. D. (2016) ‘Development of Regionalism: New Criteria and Typology’, Journal of Economic Integration, 31(2).

• Krugman, P. (1991) ‘The move toward free trade zones’, Economic Review, 76(6).

• Krugman, P. (1993) ‘Regionalism versus multilateralism: analytical notes’, New dimensions in regional integration, 58.

• Kubicek, P. (2009) ‘The Commonwealth of Independent States: an example of failed regionalism?’, Review of International Studies, 35(1).

• Kucharcukova, O.B., Babecky, J. and Raiser, M. (2010) ‘Gravity approach for modelling international trade in South-Eastern Europe and the Commonwealth of Independent States: the role of geography, policy and institutions’, Open economies review, 23(2).

• Kudina, A. and Jakubiak, M. (2012) ‘The Motives and Impediments to FDI in the CIS’.

• Kurmanalieva, E. and Vinokurov, E. (2011) ‘Holding together or falling apart: Results of gravity equation of the CIS trade’, University of Munich.

• Libman, A. and Vinokurov, E. (2010) ‘Regional Integration and Economic Convergence in the Post‐Soviet Space: Experience of the Decade of Growth’, Journal of Common Market Studies, 50(1).

• Linder, S.B. (1961) ‘An essay on trade and transformation’, Almqvist and Wiksell.

(29)

• Lundan, S. M. and Jones, G. (2001) ‘The Commonwealth Effect and the process of internationalisation’, The World Economy, 24(1).

• Martínez-Zarzoso, I. and Nowak-Lehmann, F. (2003) ‘Augmented gravity model: An empirical application to Mercosur-European Union trade flows’, Journal of applied economics, 6(2).

• Mwasha, O.N. (2011) ‘The Benefits of Regional Economic Integration for Developing Countries in Africa: A Case of East African Community (EAC)’. • Perry, C. and Coote, L. (1994) ‘Processes of case research methodology: tool

for management development’.

• Ravenhill, J. (2014) ‘Global political economy’, Oxford University Press.

• Rose, A. (2005) ‘Which international institutions promote international trade?’, Review of International Economics, 13(4).

• Sakwa, R, and Webber, M. (2010) ’The commonwealth of independent states, 1991-1998: Stagnation and survival’, Europe-Asia Studies, 51(3).

• Saunders, M., Lewis, P. and Thornhill, A. (2009) ‘Research Methods for Business Students’, Prentice Hall.

• Snijders, T.A. (2011) ‘Multilevel analysis’, Springer.

• Sohn, C.H. (2005) ‘Does the Gravity Model Explain South Koreas's Trade Flows?’, Japanese Economic Review, 56(4).

• Spector, P.E. and Brannick, M.T. (2011) ‘Methodological urban legends: The misuse of statistical control variables’, Organisational Research Methods, 14(2).

• Studenmund, A.H. (2001) ‘Using econometrics: A practical approach’, Addison Wesley.

• Todaro, M.P and Smith, S.S. (2006) ‘Economic Development’, Pearson Education.

• Tinbergen, J. (1962) ‘Shaping the world economy: suggestions for an international economic policy’.

• Viner, J. (1950) ‘The customs union issue’, Carnegie.

• Voitovich, S.A. (1993) ‘The Commonwealth of Independent States: An Emerging Institutional Model’, EJIL, 403.

• Wes, M. (2000) ‘International trade, bargaining and efficiency: The holdup problem’, The Scandinavian Journal of Economics, 102(1).

(30)

(31)

List of figures

Figure 1. Value of exports from Georgia 1996 - 2015, Millions US$

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015 CIS World

Source: IMF, Direction of Trade Statistics (2016)

Figure 2. Value of imports to Georgia 1996 - 2015, Millions US$

0 1200 2400 3600 4800 6000 7200 8400 9600 10800 12000 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015 CIS World

(32)

Figure 3. Export share between Georgia and CIS members 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015

Armenia Azerbaijan Belarus Kazakhstan Kyrgyzstan

Moldova Russia Tajikistan Turmenistan Ukraine

Uzbekistan

Source: IMF, Direction of Trade Statistics (2016)

Figure 4. Import share between Georgia and CIS members

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015

Armenia Azerbaijan Belarus Kazakhstan Kyrgyzstan

Moldova Russia Tajikistan Turmenistan Ukraine

Uzbekistan

(33)

Figure 5. Trade performance of Georgia 1996 - 2015, % of GDP 0 10 20 30 40 50 60 70 80 90 100 110 120 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015

Exports Imports Total trade

(34)

Appendices

Appendix 1: Skewness/Kurtosis test for normality

Appendix 2: Histogram of residuals 

(35)

Appendix 3: Correlation matrix of coefficients

log(TradeVol) log(ProdGDP) log(ProdGDPpc) log(Distance) Border Soviet RusLang CIS1 CIS2

(36)

Appendix 4: Variance inflation factor (VIF) test

Appendix 5: Wooldridge test for autocorrelation in panel data

Ho: no first-order autocorrelation F (1, 31) = 29.982

Prob > F = 0.0000

Appendix 6: Breusch-Pagan / Cook-Weisberg test

Ho: Constant variance

Variables: fitted values of log(TradeVol) chi2(1) = 13.85

Prob > chi2 = 0.0002

Appendix 7: Breusch and Pagan Lagrangian multiplier test

Var(u) = 0

chibar2(01) = 1283.06 Prob > chibar2 = 0.0000

Appendix 8: Hausman test

Ho: difference in coefficients not systematic chi2(7) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 9.16 Prob > chi2 = 0.0573

(V_b-V_B is not positive definite)

Variable VIF 1/VIF

“Does the Commonwealth of Independent States (CIS) promote trade among its members? Case study of Georgia”