Economic Integration of the European Union and the EMU

Get Complete Project Material File(s) Now! »

Theoretical Framework

Gravity Model

In congruence with Newton’s law of gravity, Tinbergen (1962) and Pöyhönen (1963) de-veloped the gravity model, which is an economic model that predicts the value of trade between two countries. In its simplest form, the gravity equation shows the volume of trade between two countries, which is dependent on the sizes of the two countries’ GDPs and the distance between them. The general gravity model states that trade between any two countries is proportional to the product of their GDPs and, due to the law of gravity, diminishes with distance, ceteris paribus (Krugman, Obstfeld, and Melitz, 2015).
The gravity model was first empirically tested by Jan Tinbergen in the early 1960’s. With the intention to determine and explain the pattern of international bilateral trade between different countries, Tinbergen proposed the model with an equation where the volume of trade between any two countries can be approximated by the countries’ gross domestic products (GDP) and the distance between them (Tinbergen, 1962). Equation 1 is the equa-tion Tinbergen used as the foundation to analyze trade volumes.
where Tij is the value of trade between countries i and j, C is a constant term, Yi is the GDP of country i, Yj is the GDP of country j and Dij is the distance between the countries.
Although the fundamentals of the gravity model are still based on Tinbergen’s early model, there have constantly been advancements and revisions to make the model more suitable for research on the effects of different variables. Most versions of the gravity model employ exports and bilateral trade flows as the dependent variables, while the number of explanatory variables in the model can vary greatly (Kepaptsoglou, Karlaftis, and Tsamboulas, 2010). These explanatory variables can be divided into factors indicat-ing demand and supply, as well as other factors that affect trade between the two coun-tries. The common factors used in indicating demand and supply are GDP per capita, income level, population, and area size. The other factors are elements that have either negative or positive effects on trade. The main resistance factors are transportation costs, but different gravity model specifications also include factors such as participation in a customs union or a trade agreement, border adjacency, sharing the same currency, and landlocked country (Kepaptsoglou et al., 2010).

Geographic and Cognitive Distances

In the gravity model, the distance variable refers to the geographical distance between country pairs. The standard argument as to why geographic distance is negatively related to trade is that large distances between trading countries increase the transportation costs involved in physically moving the goods (Berthelon and Freund, 2008). Therefore, dis-tance is used as a proxy to describe the transport costs, such as freight charges. A larger distance also increases the time spent in transit, which not only affects the capital costs incurred during the transportation, but also increases the associated risks in aspects such as possible changes in exchange rates (Håkanson, 2014). Advancements in technology during the recent decades have led to lower transporting costs, liberalization of trade, and easier communication, and one could expect that the role of distance would have become less restricting on trade (Carrère and Schiff, 2005; Håkanson, 2014). Nevertheless, re-search suggests that the negative trade impact of distance has remained unchanged or even increased over time4.
Trade is not only affected by the geographic distance, but other aspects such as sharing a common border or having direct access to the sea, are factors commonly considered hav-ing a significant impact on trade between countries. Common borders are positively re-lated to trade, as it appears that adjacent countries trade more with each other than non-adjacent countries (Magerman, Studnicka, and Van Hove, 2015). Research by Fischer and Johansson (1995) indicates that two countries sharing a common border trade about twice as much compared to non-neighboring countries.
A country is considered landlocked when it does not have direct access to the sea (Rabal-land, 2003). The lack of coastal access has a negative impact on the country’s trade, which is mostly due to the fact that it increases the transportation costs for the landlocked coun-try (Limão and Venables, 1999). However, it appears that the effect of landlockedness is not equal across countries as some landlocked countries’ trade is affected more than oth-ers. Coulibaly and Fontagné (2005), find that the non-European landlocked countries are more severely impacted by reduced trade, compared to those located in Europe. A possi-ble reason is that most of the landlocked European countries are closely integrated by being part of the European Union market (Raballand, 2003).
The first study to introduce the concept of cognitive, or psychic, distance was Beckerman (1956) with the goal of broadening the definition of distance so that it could better be used to explain international business patterns. Evans, Mavondo, and Treadgold (2000), propose that cognitive distance results from the cultural and business differences between the trading countries. These distance-creating cultural and business factors that disturb the information flow are, for instance, cultural, educational, structural and political, and language differences, as well as the varying level of economic development and differ-ences in industry structures (Evans et al., 2008; Johanson and Wiedersheim-Paul, 1975; Nordstrom and Vahlne, 1992).
Rose (2000) and Melitz (2008) state that sharing a language increases bilateral trade be-tween countries by economically and statistically significant amounts. Melitz (2008) states that the finding has been backed up by gravity models. He notes that the typical way of including a common language in a gravity model is through a binary measure, but that this approach does not clarify through which channel the effect acts, that is, indirect versus direct communication. In his study about the channels through which common language mainly promotes bilateral trade, Melitz (2008) finds that compared to indirect communication, direct communication is approximately three times more effective in promoting trade.
Regarding the effects of different languages in promoting trade, Melitz (2008) finds that English, despite its dominant position as a language, does not promote trade more effec-tively than other major European languages. However, as a group the major European languages, including English, promote trade more efficiently than other languages. In addition, Melitz (2008) finds that language diversity in a country has a positive effect on its foreign trade. As sharing a common language has a positive impact on trade between two countries, the more official languages a country has, the more it can directly com-municate with other countries that share these same official languages, thereby increasing bilateral trade.
According to Rose (2000) as well as Mitchener and Weidenmier (2008), a country’s prior colonial status has a statistically significant and economically large effect on current bi-lateral trade. The trade effect of a country’s colonial status per se remains ambiguous as trade literature has not studied the effect of a country’s colonial status besides the impact of mutual colonial ties (De Sousa and Lochard, 2012). The observed trade increase asso-ciated with colonial ties is partly due to the use of common language, established currency unions, and the establishment of preferential trade agreements (Mitchener and Wei-denmier, 2008).

Economic Integration of the European Union and the EMU

When looking at the world as a whole, the EU countries are geographically close to each other, which increases trade within the Union according to the gravity model. Free trade within the single market further advances the amount of trade that takes place between the member countries. With the elimination of customs duties, goods are allowed to cir-culate freely within the single market. Besides that, the EU countries are also close to each other in terms of cognitive distance as they are a relatively homogenous group of countries when compared to the rest of the world. The three main languages in the EU are English, French, and German, which are also the languages of the biggest EU coun-tries in terms of GDP (World Bank, 2017). This provides reasoning as to why the major European languages as a group promote trade more efficiently than other languages.
Despite the benefits of free trade within the Union and the status of the largest exporter in the world, there are also downsides to EU membership. The membership is rather costly, and some countries end up being net contributors to the Union’s budget, meaning that they contribute more money to European Union’s budget than they receive from it (Williams-Grut, 2016). The net beneficiaries from the Union’s budget are typically the member states in Eastern and Southern Europe, while the more affluent countries of West-ern Europe are typically the net contributors (Centraal Bureau voor de Statistiek, 2016).
Studying the trade effect of EMU, one can identify various reasons why being a member of the Eurozone is beneficial for a country. The most significant advantages of sharing a common currency are the elimination of the exchange rate uncertainty and the reduction of transaction costs within the Eurozone (Cîndea and Cîndea, 2012). Currency exchange fees as well as the risk of sudden exchange rate fluctuations have in the past been consid-ered to be trade barriers, as they impede trade between countries. A common currency removes such costs between the members of a monetary union. Although the exchange rate uncertainty is eliminated between the member states, the currency risk still exists to a certain degree when trading with countries outside the Eurozone (Gottfries, 2013).
A common currency also contributes to increased price transparency. Quoting prices in the same currency facilitates the price comparison of the goods and services offered in different countries (Gottfries, 2013). More transparent prices are expected to have a pos-itive impact on trade as they promote competition and simplify trade in general. There-fore, one could expect the EMU to have a positive impact on the member states’ trade, as a common currency and shared monetary policy lead to lower transaction costs, reduced exchange rate uncertainty, and price transparency. Previous studies regarding the euro effect on trade have found that adopting the euro has increased bilateral trade between the euro countries, but it has also promoted trade between country pairs when only one country uses the euro as a currency5. Research suggests, however, that the euro effect is not as large as a general currency union trade effect due to the high level of EU trade that already existed before the implementation of the common currency (Rose and van Win-coop, 2001).
When adopting the common currency, the member countries have to give up their inde-pendent monetary policies. This means that the European Central Bank conducts common monetary policy in the Union and individual central banks have no control over the inter-est rates nor the possibility to devalue their currency in a situation of high inflation (Cîndea and Cîndea, 2012). Hence, there are also downsides to the common currency, as the member countries’ economies are not homogenous, and the common currency poses different effects depending on the economic situation in the country. When a country has a higher level of inflation than other Eurozone countries with no possibility of devaluing its currency, the country might become uncompetitive (Cîndea and Cîndea, 2012), which in turn lowers the country’s exports and economic growth, as has been the case for Greece, Spain, and Portugal.

Previous studies

Most of the literature written about the trade effects of the European Union and the Mon-etary Union employ an augmented gravity model and have found evidence of positive trade effects. Numerous studies have taken interest in the trade effect of a common cur-rency by studying the overall trade effect of currency unions and the EMU in particular. Rose (2000) conducted pioneering investigation on the impact of a common currency area on trade using a panel data set including bilateral observations for 186 countries between the years from 1970 to 1990. He found that two countries sharing a common currency trade approximately three times more than countries with different currencies. Several studies have built on Rose’s initial research when studying the trade effects of the euro and revised Rose’s high estimates of the euro’s trade effect. De Nardis and Vicarelli (2003) and Micco et al. (2003) present a euro effect of trade increases between 9 percent and 19 percent. Results by Micco et al. (2003) suggested an increase in trade between non-EMU countries as well. Bun and Klaassen (2002) estimate that the trade increase of 4 percent in the first year of EMU will cumulate to nearly 40 percent in the long run. Later, Bun and Klaassen (2007) reform the model by removing an upward trend and obtain a more realistic euro impact of 3 percent increment on goods trade.
Despite the large variations in the magnitudes of the positive trade effects of a currency union, all studies6 come to the same conclusion that currency unions positively affect trade in member countries and that the euro has substantially increased trade since its implementation. Furthermore, Micco et al. (2003) believe that the impact of the EMU on trade can be compared to the impact that EU membership has had on trade.
The effects of the financial crisis of 2008 have been widely documented, and it is clear that nearly all countries were affected through international declines in trade. Some phe-nomena of financial crises seem to be distinctive and to appear repeatedly in different crises through several decades, as indicated by Abiad, Mishra, and Topalova (2014). They studied how the effects of financial crises during the period from 1970 to 2009 affected different countries’ trade. By employing an augmented gravity model, they found that the declines in imports are often more severe and persistent compared to the falls in exports during financial crises.
Combining the positive EU effects and the negative crisis effects, literature presents al-ternative views of the state of the EU trade during the crisis. Fojtíková (2010) studied the data provided by Eurostat and the World Trade Organisation, and found that EU imports and exports were negatively impacted by the crisis both on an intra-EU and extra-EU level, but that the countries were affected differently. Although the European Union kept its leading position regarding the share of world trade, in terms of import and export val-ues, EU trade did on average decline more than the rest of the world between 2007 and 2008. Kren et al. (2015) conducted a study regarding the export performances of the 27 EU countries before, during, and after the financial crisis by using bilateral export data from the time period from 2003 to 2013. They found that in the long-run, the export in-tensity differed between the countries. The study also found considerable cross-country heterogeneity in the trade effects between the EU member countries before, during, and after the financial crisis. Furthermore, Kren et al. (2015) conclude that Eurozone mem-bership seemed to protect exporters during and after the financial crisis.
In her study, Fojtíková (2010) did not perform an inferential analysis, but only studied the export and import values between the countries during the period of 2007 and 2008. Thus, it cannot be analyzed whether or not the changes in EU trade can be considered to be statistically different in comparison to the rest of the world. In contrast to Fojtíková, we perform statistical testing and econometric analysis regarding the crisis’ impact on trade between countries. Kren et al. (2015) conducted the study using solely EU countries as the sample and focused on how trade of the members of the European Union was af-fected before, during, and after the crisis. They later compare how Eurozone countries did in comparison to the EU as a whole. In order to analyze the protective aspects of the customs and the monetary unions, our research employs a larger sample of 40 industrial countries to conduct a comparative study on the trade effect of EU and EMU member-ships. By including the larger country sample, we are able to compare how the EU and Eurozone countries’ trade was impacted compared to other industrialized countries dur-ing the financial crisis.

Data

For the purpose of this study, the chosen data method is to use pooled data for a total sample of 40 countries that includes all of the 35 OECD countries, 22 of which are EU countries, as well as 5 EU countries that are not members of the OECD7. The time period of the study was chosen to be from 2000 to 2016, which is the latest year for which data is available. By employing this time period, the study aims to obtain a comprehensive view of the state of the sample countries’ economic situation before, during, and after the financial and trade crises that started in 2007.
The study divides the sample countries into three sample groups that are studied to ana-lyze the effect of EU and EMU membership. To separate the three groups, dummy vari-ables are created for EU and EMU membership. The base group that represents the gen-eral economic situation in the industrialized countries includes all 40 countries and can be regarded as the most heterogeneous of the three samples, as it includes countries from all around the world. By world standards, however, this group of countries can still be regarded as rather homogenous as the countries are all industrialized, democratic, and share similar trade policies. The second sample group includes 27 EU countries, which are considered to be more homogenous as a group due to common legislation and policies under the Union. The third group includes exclusively EMU countries that have adopted the euro and is used to study the trade effect of the common currency.
Data for some countries, such as Malta, Belgium, and Luxembourg were missing before the year 2000, and in order to build a balanced panel, the time period has been restricted to begin from the year 2000. Furthermore, Croatia has been left out of the entire sample as it joined EU in 2013, which is relatively late in the period this study examines, as the most severe part of the crisis had already been overcome in 2013. This study also excludes four countries that joined the euro after 2008: Estonia, Latvia, Lithuania, and Slovakia, since these countries adopted the common currency after the crisis had begun. In order to get a reliable view of the trade effect of the euro, the study focuses on the 15 EMU coun-tries that already used the common currency when the crisis began. A list of the countries included in each sample can be found in the Appendix.
The data for the Gross Domestic Product (GDP) of the sample countries was retrieved from the OECD database, while the total merchandise export of the countries was ob-tained from the UN Comtrade virtual database. The GDP is measured in Purchasing Power Parity (PPP) adjusted current US dollars, which makes the data internationally comparable across the countries (OECD, 2018). The data for the geographical and cog-nitive variables included in the model, such as bilateral distance, colonial ties, and com-mon language, were retrieved from the CEPII8 virtual database. The EU and the EMU dummy variables were constructed by using information from the official European Un-ion website.

Variables

Dependent Variable
In this study, trade will be treated as the dependent variable. It is measured as the total value of bilateral merchandise exports between the country pair i and j. The theory of the classic gravity model does not specify any particular measure for trade values between countries, but most versions of the gravity model employ exports or bilateral trade flows as the dependent variables (Kepaptsoglou et al., 2010). Employing exports as the depend-ent variable follows the lines of previous studies by Bun and Klaassen (2002) as well as De Nardis and Vicarelli (2003). Refer to tables 5.1, 5.2, and 5.3 for variable descriptions.
Independent Variables
As previously mentioned, GDP and distance are the fundamental variables in the classic gravity model. To modify the gravity model to fit this particular study, we must also include variables that are believed to have a significant effect on trade between the sample countries. These additional variables will be introduced to the model using binary varia-bles.9
GDP functions as a measure of the economic size of a country, hence the GDP values for each sample country are included. Following the lines of a study by Glick and Rose (2002), this study employs GDP instead of GDP per capita. Since all the sample countries can be considered well-developed, GDP per capita, which measures the level of economic development rather than absolute economic size of the country, is not the variable of interest. According to the early studies of Tinbergen (1962), GDP has a positive effect on trade, and therefore, we expect the coefficient for GDP to have a positive sign. Distance is the other fundamental independent variable. It is measured in kilometers between the capital cities of the sample countries. As previous studies have found a negative relation-ship of geographic distance and trade between countries (Barr et al., 2003; Micco et al., 2003; Persson, 2001; Rose 2000), we expect this variable to have a negative effect on trade.
To analyze the effect that a common language has on bilateral trade between the sample countries, a dummy variable is included. The variable takes on the value of 1 if the coun-try pair shares an official language and 0 otherwise. As stated previously, sharing a lan-guage increases bilateral trade between a country pair by statistically significant propor-tions (Melitz, 2008; Rose, 2000). Consequently, we expect the coefficient for the com-mon language variable to be positive.
Common border, or contingency, is another independent variable that has previously been found to have a positive effect on bilateral trade values (Fischer and Johansson, 1995; Magerman et al., 2005). Common border is introduced to the model through a dummy variable, which obtains a value of 1 if the country pair shares a border and 0 otherwise. Due to the findings in the previous studies, we expect the variable for common border to have a positive effect on trade in our estimated model. A dummy for landlockedness is also included in the research model as the lack of coastal access has been shown to have a negative impact on a country’s trade due to increased transportation costs (Coulibaly and Fontagné, 2005; Limão and Venables, 1999; Raballand, 2013). The dummy variable takes on the value of 1 when at least one country in the country pair is landlocked and 0 otherwise. In line with the results of earlier studies, we expect the variable to have a neg-ative effect on trade. Another dummy variable is the effect of colonial ties between the country pair. The dummy variable takes on the value of 1 if the countries have ever had colonial ties and 0 otherwise. According to Rose (2000) as well as Mitchener and Wei-denmier (2008), past and current colonial ties reinforce bilateral trade between countries. In line with that, we expect colonial ties to have a positive effect on trade in our research model.
Dummy variables for EU and EMU memberships are included in the model in order to scrutinize the overall effects they have on trade (Table 5.2). These dummy variables are introduced with the purpose of separating the groups of countries that are members of the custom and currency unions. Additionally, for EU, there are two dummy variables: One indicating a country pair when both countries are members of the EU, and the second for a country pair where one country is a member of the EU and the other one is not. Two additional dummy variables for EMU membership are constructed the same way. The dummy variable for membership takes on a value of 1 the year the country became a member and onwards. Since the European Union is a free trade area where all countries are under the same legislation and relatively close in both geographic and cognitive terms, we expect the EU variables to have a positive effect on trade. Previous studies introduced earlier state that being a member of the Eurozone reinforces trade (Rose and van Win-coop, 2001; Bun and Klaassen, 2002; Glick and Rose, 2002; de Nardis and Vicarelli, 2003). Therefore, we expect the coefficients for the EMU variables to have a positive effect on trade as well.
When analyzing the regression results, one must consider that different EU and EMU sample groups have different control groups, and hence, somewhat different interpreta-tions. Therefore, we cannot directly make comparisons between the impact of the crisis on different sample groups. The EU dummy takes into account the overall trade effect of being a member of the EU, with the control group consisting of non-EU OECD countries. The Both in EU dummy measures the intra-EU trade effect with the control group of country pairs where either one is an EU-member or when both countries are non-EU OECD countries. Lastly, the One in EU dummy variable considers the extra-EU trade effect with the control group consisting of the country pairs when both are EU-members or both are non-EU OECD countries. The same logic applies to the EMU dummy varia-bles.
The time period of the crisis is also included as a dummy variable. The dummy takes on a value of 0 for all years in the sample before the year 2007 and 1 for the years after the beginning of the crisis. The reason, why the crisis dummy takes on the value of 1 all the years after 2007, is that there is no concrete answer as to when the financial crisis offi-cially ended. The economic recession that followed the crisis may have ended in 2012, but there are signs of how the global economic crisis still is impacting the world (Tathyer, 2017). The years after 2012 display the post-crisis economies of the sample countries. Lastly, the model also controls for years through dummy variables, representing the 17 time periods being analyzed from 2000 to 2016. There is assumed to be some variation in the trade patterns over the years, which can result in unobserved heterogeneity in the regressions. Such effects can incorrectly impact the effect of other variables, and there-fore, by including year dummies, the unobserved factors affecting trade can be captured and controlled for. In each regression, the year of 2000 is considered the base year to avoid a case of perfect multicollinearity.
Through interaction terms of the three EU dummy variables and the Crisis variable, and later the interaction term between the three EMU dummy variables and the crisis variable, we are able to study the combined effects of EU and EMU memberships and the crisis on the export values of the sample countries. The different interaction terms enable the re-search of the within-effects of the EU and EMU member countries, as well as the be-tween-effects from the EU and EMU countries to non-member countries during the crisis. If the effect is positive for the EU member countries, it implies that they were less affected by the crisis, in export terms, than the overall sample of countries and a negative effect displays the opposite conclusion. The effect of EMU membership is studied in a corre-sponding way. In accordance with the results from the research conducted by Kren et al. (2015), we expect the interaction terms to produce positive outcomes for both EU and EMU member countries, as it would imply that there is a protective aspect to EU and EMU, and that member countries’ exports were less severely affected by the crisis.

Table of Contents
1. Introduction
2. Background
2.1 The growth of world trade
2.2 The Financial Crisis of 2008
3. Theoretical Framework
3.1 Gravity Model
3.2 Geographic and Cognitive Distances
3.3 Economic Integration of the European Union and the EMU
4. Previous studies
5. Data
5.1 Data
5.2 Variables
5.3 Descriptive Statistics
5.4 Correlation Matrix
6. Empirical Model and Analysis
6.1 Model
6.2 Robustness Test
6.3 Results
6.4 Analysis
7. Conclusion
References
GET THE COMPLETE PROJECT