The effect of the Euro on trade between Israel and the Eurozone countries

(1)

1

The Effect of the Euro on Trade between Israel

and the Eurozone Countries

Student name: Asaf Koren Student number: 10604197 Email: asafkoren52@gmail.com Number of words: 10,005 Date: 15.08.2014

Supervisor: F.J.G.M. Klaassen Second reader: R.M. Teulings

Study: Master of Science Economics – University of Amsterdam, Faculty of Economics and Business Specialization: International Economics & Globalization

(2)

2

1. Introduction

It was expected that the creation of the single currency union, with the introduction of the euro, would affect trade flows. Existing research on the effects of the euro on trade in the Eurozone has shown that there is a positive effect of 3-40 percent (Baldwin et al, 2008). The effect of the creation of the European monetary union on trade with countries outside the Eurozone has been largely, but not fully, neglected. The existing research mainly focuses on estimating the effect of the euro on trade between countries in the European Union. The studies that do focus on trade with outside countries find a low positive effect of the euro on trade.

Economic theory on international trade suggests that trade increases between the Eurozone countries because exporters have lower costs due to lower exchange rate volatility (hedging costs). This effect might lead to trade diversion that decreases exports from outside countries to the Eurozone countries.That happens because, compare with exporters from within the Eurozone, exporters from outside become less competitive. Hence, According to economic theory we expect the euro the cause trade diversion, but the studies mentioned earlier find low positive effect of the euro on trade. This contradiction makes it interesting to further research the euro effect on outside countries and to, perhaps, gain more insight by focusing on the effect of the euro on trade with a specific outside country.

Therefore, this thesis will contribute to the knowledge on this field by estimating the effect of the creation of the Eurozone on exports from Israel to the Eurozone countries. Israel is an interesting study case because it is a small economy that relies on exports and the European Union is its largest trade partner. According to a report published by the European commission in 2012, the European Union (EU) is the top trade partner of 80 countries. This stresses the importance of researching the trade effects of the creation of the euro on outside countries.

Exports are an important source of economic growth for Israel. It is an open economy where the share of exports plus imports divided by gross domestic product (GDP) is above 70 percent. Israel is a small nation scarce in natural resources, hence a high volume of exports is crucial in financing imports of such goods. In 2013 exports from Israel to the European Union accounted for more than 27 percent of the total exports of Israel. If the euro causes trade

(4)

4

diversion, the effect on Israel’s economy would be profound. Lastly, Israel makes for a good case study because of the high quality export data between Israel and most of the Eurozone countries.

The central question in this research is: what is the effect of the euro on exports from Israel to the Eurozone countries between the years 1970-2011? I will research this by using annual bilateral export data from Israel and 17 other countries. In order to estimate the effect of the adoption of the euro on exports from Israel to the Eurozone countries I will use an Ordinary least squares (OLS) regression analysis using panel date and implementing the gravity model. The gravity model is a model commonly used in international economics to predict trade flows based on economic size and distance between countries. The dependent variable will be the value of exports from country i to country j in year t. To estimate the effect of the euro I will use a dummy variable. This dummy variable will have the value of 1 if the importing country is a member of the Eurozone and the exporting country is not. The coefficient of this variable is the main interest of this paper. Because it will show what was the effect of the euro on trade with outside countries.

The structure of the thesis is the following: chapter two contains a literature review. In the third chapter the empirical model and its theoretical foundation is presented. The fourth chapter presents the estimation results of my model and a sensitivity analysis. Finally, conclusion are drawn in chapter five.

2. Literature review

2.1 The effect of currency unions on trade within the member countries

This part of the literature review focusses on studies that estimated the effect of the establishment of a currency union on trade. Unlike the early papers in this field, in my research I only focus on exports and not on trade. The difference between the two is that trade is the sum of both imports and exports. The developments in the currency union effect on trade literature are influenced by papers that focus on trade as well as by papers that focus on exports. For this reason I decided to also include the papers that estimate the euro effect on trade.

Rose (2000) was one of the first prominent researchers that applied the gravity model on currency unions in order to estimate their impact on trade. Estimating the impact of currency

(5)

5

unions on trade is similar to estimating their impact on exports because export is a part of trade. Since his paper was published in 2000 and the euro was only established in 1999, Rose did not use data on the Eurozone. Instead, Rose tried to estimate the effect of currency union on trade by using data from existing currency unions, most of them were formed by small and poor countries. The data used for this work is a panel data set of 186 countries between the years 1970-1990. The result of this research was that trade flows were 300 percent larger between countries that were members of the same currency union. The regression in this paper suffered from problems such as the omitted variable bias and the reverse causality problem according to Baldwin (2006). The reverse causality problem arises because it is not clear from this article if currency union causes more trade or that more trade leads to a formation of currency unions. Nevertheless, Rose laid out the fundamentals of conducting a research about the trade effects of a currency union that were used by most of the researchers in order to estimate the trade effect of the euro.

Micco, Stein and Ordonez (MSO) were the first to publish an article in which they estimated the trade effect of the euro (2003). They used panel data of bilateral trade flows with the gravity model for 22 countries, among them 12 European countries that adopted the euro in 1999. They had three years of data from 1999-2002 that could be used to provide an early estimate of the euro effect on trade. They included both countries that were members of the European Union and countries outside the European Union. In their construction of the gravity model, the authors added other variables to the simple form of the model. Variables that became important according to the empirical trade literature such as GDP, GDP per capita, a dummy variable for a membership in a free trade area and so on. Their paper expanded the model that was used by Rose 2000 in hope to reduce the omitted variable bias. The findings of this research were that the euro had an 8-16 percent positive effect on trade, which meant that the euro increased trade between the countries in the Eurozone compared to trade between countries that do not use the euro.

It is easy to see that the trade effect of a currency union that was found by Micco, Stein and Ordonez is significantly lower than the estimation of Rose. The difference is probably because Rose uses mainly poor and/or small countries and because MSO (2003) use more control

(6)

6

variables. It seems that adding more control variables reduces the coefficient of the monetary union, which can be an indication for omitted variable bias in Rose’s research.

Bun and Klaassen (2002) also estimated the trade effect of the euro, using a data set of annual bilateral exports between 1965 and 2001. The authors use a similar model to the one of Rose and MSO. Instead of using variables such as common language and distance the writers use country pair specific effects that contain these elements and others. In the model presented in this paper, time fixed effects are used to control for the state of the global economy, for example. The conclusion of this article is an estimated 37.8 percent increase in trade between Eurozone countries due to the trade effect of the euro.

In a later article, Bun and Klaassen (2007) point on an omitted variable problem with the way the gravity model was used so far to estimate the effect of the euro on trade. They show that in previous papers a strange pattern arises: the longer the time period is, the higher the coefficient of the euro is. This pattern occurs because the euro dummy variable has picked up trends in trade, over the years, which are caused by omitted variable. To correct for this bias the writers include, in the regression, country pair specific time trends. The paper uses bilateral trade data of 19 countries including the European Union countries (before the 2004 enlargement), Norway, Switzerland, United stated, Japan and Canada. The time span is 1967-2002. According to this paper the trade effect of the euro is 3 percent, a smaller estimation than previous researchers made.

Flam and Nordstrom (2003) also wrote an important article about the euro effect on trade. In their research they focus on aggregated sectorial data and make a comparison between trade flows in the years 1989-1997 to the years 1998-2002. They find that trade within the Eurozone increased by 15 percent while trade between the Eurozone and outside countries had increased in 8 percent. Another significant paper in this field was written by Berger and Nitsch (2008). They used a long time horizon for their data from 1948 until 2003 on 22 industrial countries. They found that, when controlling for trends in trade integration, the effect of the euro on trade disappears. The last paper presented in this section is of De Nardis and Vicarelli (2003). The researchers estimated the trade effect of the EMU on trade between its members and found

(7)

7

that that the adoption of the euro has increase trade by 9-10 percent. At the time in which their article was written this result was lower than results of their peers. They explain this difference by pointing out that the euro was adopted by countries that have gone through many stages of economic integration in the past and that the euro coefficient was influenced by that.

This part of the literature review focuses on developments in the estimation of the currency union effect on trade. All paper presented use panel data with the gravity model for this estimation. It is easy to see that later studies estimated a lower effect of the euro on trade because of their use of more sophisticated estimation methods. One of the most important addition to this estimation is the country pair specific time trends of Bun and Klaassen (2007).

2.2 The effect of the euro on trade between the Eurozone and outside countries

This section discusses existing literature on the trade effect of the Eurozone on outside countries. One of the papers that did so was the MSO paper from 2003. This article is relevant because it attempts to estimate the impact of the euro on trade with outside nations that are not in the European Union and it is considered an important article in this field. In order to estimate the euro effect on outside countries, the writers construct a dummy variable that is one when only one country has the euro and zero otherwise. The coefficient of this dummy variable gives information about whether trade increased between the Eurozone countries and outside countries. According to this paper, trade between Eurozone and outside countries has increased by 9 percent. This estimation is lower than the estimation of the increase in trade between Eurozone countries, nonetheless it is a sign that trade with outside countries had increased due to the euro. This means that there is no sign of a trade diversion effect and it even suggests that there is negative trade diversion effect of the euro with outside countries. That is, an increase in trade between outsiders and insiders, due to the euro effect, that theoretically was expected to decrease.

Baldwin and Di Nino (2006) notice that in previous research there is no evidence of trade diversion due to the euro, but rather an effect of negative trade diversion. Even though Bun and Klaassen (2007) show that this surprising result also disappears when using country-pair time trends, it is nevertheless interesting to discuss the explanation that the authors present for this

(8)

8

findings. In response to these finding the authors constructed the new goods hypothesis. According to this hypothesis the euro decreases the costs of exporting a new good, that was until now consumed only at home, to the Eurozone. The new range of goods are now exported with lower costs because the Eurozone is perceived by the exporter as one market. For example, the costs of establishing distribution centers decreases due to economies of scale. The new goods increase the volume of trade, thus causing negative trade diversion. In this paper the authors also test the new good hypothesis by using sectorial data. Their results suggest support for the hypothesis but they fail to deliver conclusive evidence.

Both papers presented here found a small positive effect of the euro on trade. In the first paper the euro effect on trade with outsiders is only a small part of the research and is only given a short analysis. While the second paper focuses on this issue to a greater extent and the authors also develop a theoretical explanation to the finding. In addition, all the papers presented above did not focus on one specific country but choose to focus on the estimation of the euro trade effect on roughly the same countries (The EU ones and other advance economies such as the United Stated, Switzerland, Norway, Canada, Japan, Australia). In my research I focus on the effect of the euro on trade with a specific outside country, which is Israel. For that reason, I turn to discuss papers and reports that were written about Israel’s trade pattern and the Israeli economy.

2.3 The Israeli economy

In order to understand the importance of high levels of exports from Israel to the European Union, and in that to the Eurozone countries, this section presents the specific characteristics of the Israeli economy. Israel is a small country both in area and in population and it has a very open and developed economy that relies on high levels of exports. According to the Israeli Central Bureau of Statistics in 2013 exports accounted for 34.2 percent of GDP while imports accounted for 37.2 percent of GDP. Israel is a developed economy, as can be seen from the high GDP per capita (USD 35,658). The GDP per capita of Israel is higher than the average of the European Union countries making Israel an advanced economy, similar in many ways, to the countries that researchers focus on in the currency union effect on trade literature.

(9)

9

According to the World Trade Organization (WTO) trade policy review on Israel from 2013, in the two recent decades the country has developed a very advanced high tech sector. Since 2008, high tech goods accounted for more than 90 percent of the total merchandized exports. In respect to the direction of trade, the World Trade Organization report stresses that due to geopolitical reasons Israel’s trade with its neighbors is very low. As a result of Israel’s relation with its neighbors, the European Union and the United States (US) remains Israel’s main trade partners. The trade between Israel, the United States and the EU is done mainly under free trade agreements that were signed between Israel and the European Community (EC) in 1975 and between Israel and the US in 1985. In fact, Israel was the first country in the world to have free trade agreements with both the United States and the European Union.

Exports to the EU are a major part of the total exports of Israel. In 2013 exports from Israel to the EU accounted for 27.1 percent of total exports, while exports to the Eurozone accounted for 19.1 percent of total exports. The largest export destination countries in the Eurozone are Belgium, the Netherlands, Germany and France while outside the Eurozone the United States, United Kingdom, Hong Kong, India and China are the largest export destinations. In that regard it is important to mention that exports to Asian markets are becoming more and more important to the Israeli economy. The WTO report emphasis that foreign demand will remain a critical determinant of overall growth in Israel. Especially Israeli exporters are heavily dependent on the macroeconomic performance of its largest trade partners; the EU and the US.

Levi and Friedman (2005) stress the importance of Israel being an open economy with a high export and import rates. According to their research, the Israeli economy accumulated its capital by creating a high debt to foreign creditors. To repay this debt Israel needs to maintain high levels of exports minus imports. High rates of exports are also important to finance imports and to maximize the advantages of economies of scale. In addition, openness to trade has a positive effect on GDP because it encourages productivity and the adoption of new technological developments. Their paper claims that Israel is an open economy with low protection for the domestic production and without a subsidy system for imports. It finds that the European demand for Israeli exports is essential to Israeli exporters and to the sustainable growth of the Israeli economy.

(10)

10

Because of the importance of high levels of exports to the Israeli economy, and the understanding that Israel is an open and developed economy, a reduction of exports from Israel to the Eurozone, due to trade diversion, could have a large negative effect on the Israeli economy. For that reason, it is important to estimate the euro effect on the Israeli economy through its effect on exports.

3. The empirical model

3.1 The theoretical impact of a common currency on trade

As my thesis estimates the effect of a currency union (Eurozone) on trade with outside countries (Israel), it is important to consider economic theory regarding the trade effects of a common currency. Most economists and economic theories suggest that the use of different currencies in different countries has a negative effect on trade. Hence, a currency union should boost trade. From a theoretical point of view there are several ways in which a common currency can impact trade.

The first one is through the elimination of the nominal exchange rate volatility. Hooper

and Kohlhagen (1978) state that traders do not appreciate risk. Thus, an increase in the volatility of the exchange rate would reduce the volumes of trade of the exporter and the importer. It is believed that risks associated with exchange rate volatility can discourage economic actors from engaging in international trade. It is difficult for Producers, in the exporting country, to estimate what will be the purchasing power of the foreign money they receive, for their goods sold abroad, in their own country. For example, when an Israeli exporter, who receives payments in dollars, faces unexpected appreciation of the shekel vis-à-vis the US dollar, his profit margins will decrease. The exporter can hedge against such appreciation, but hedging has a cost that might make his product less competitive. Kenen (2003) showed that it is very difficult to predict the exchange rate changes and completely hedge against them. Those exchange rate volatility risks can reduce the amounts of international trade by increasing the costs of exporting and importing leading more economic agents to not engage in international trade.

Sometimes traders do not know when they will receive the payments for their goods sold

(11)

11

of the contract makes it more difficult and costly to hedge against exchange rate volatility. The cost of hedging is determine not only according the stability of the two currencies but also according to the length of the hedging period. Since it is more difficult to estimate the exchange rate for longer time. Hence, the additional cost of hedging makes the exporters less comparative. The level of the exchange rate can also decrease demand for imported goods abroad and contribute even more to the uncertainty that the exporter faces. Klaassen (2004) stated that many researchers have tried to estimate the effect of exchange rate volatility on trade but without conclusive results due to the difficulties in finding an effect of exchange rate volatility on trade. According to the author, the exchange rate distribution of one year ahead is the one that has most influence over the levels of exports. Because the volatility of this one year ahead exchange rate remains relatively constant over time, there is little variation in the available date. Those low levels of variation make it very difficult to find the effect of exchange rate volatility on trade.

When countries form a currency union, the nominal exchange rate volatility risk disappears and exporters face lower costs. In my research, the nominal exchange rate volatility between Eurozone countries was abolished but the one between the Israeli currency and the euro remains. Hence, according to the theory above, the Israeli exporters to the Eurozone become less competitive than those within the Eurozone. The theory implies that due to the higher costs faced by exporters from outside of the currency union compared to exporters from inside the Eurozone, outside exports will decrease.

A second theoretical effect of a currency union on trade is through the elimination of transaction costs. These are the costs that the exporters pay to financial intermediates for the exchange of currencies. In addition, there are transaction costs in the administration of the currency exchange. For instance, the costs of a department in an exporting firm that is in charge of the currency exchange. Countries in a currency union do not encounter these costs when trading with one another and thus their products become cheaper. This increases their competitiveness compared to outsiders and, like the first effect, predicts that exports from outsiders to insiders will decrease.

(12)

12

Some researchers empirically estimated these costs as a percentage of the total GDP of countries. Emerson et al (1992), found that these cost account for 0.5 percent of the EU’s GDP. The estimate changes to one percent when focusing on small and open economies in the EU. A paper written by Soffer (2007) tried to estimate the savings to the Israeli economy due to the elimination of transaction costs. Soffer focuses on two cases: one wherein Israel adopts the euro and the other in which it adopts the US dollar. The results are surprisingly low. The dollar adoption will save 0.18% of GDP while the adoption of the euro will save only 0.09% of GDP. The reason for the difference between the Emerson et al (1992) estimation and the Soffer (2002) estimation is that conversion costs of currencies have become lower over the years due to better financial technologies, deepening of financial markets and more competition in financial markets. It is important to mention that according to Soffer (2002) those saving estimation are a lower boundary. This is the case because his research does not estimate the following scenario. In this scenario once a foreign currency is adopted the use of other foreign currencies is moved towards the adopted currency causing the costs associated with the exchange of currencies to decrease.

With respect to trade between countries inside a currency union and countries outside, the above theoretical impacts suggest that outsiders’ exports to the currency union will decrease. This happens because consumers in the currency union face lower prices for goods originating within the union. This means that consumers in the Eurozone will consume the goods originating within the currency union because of their lower price. This change in consumption is sometimes referred to as trade diversion. The term trade diversion comes from the famous theory of Viner (1950) about how preferential trade agreements can cause the member countries to change their imports from efficient non-members to a less efficient member suppliers. Viner called this effect trade diversion. The switch in suppliers caused by a currency union can also be called trade diversion because of the fact the trade divert from outside the union to within the union. Regarding the difference in efficiency between the outsiders and the insiders, unlike Viner’s theory, in case of a currency union, the insiders are more efficient. They are more efficient because, they have less costs related to exchange rate volatility and other currency exchange costs. Even though the insiders are, in case of a currency union, become more efficient I still refer to the change of imports from outsiders to insiders as trade diversion. As mentioned above, I

(13)

13

will not try to estimate the above theoretical effects in my thesis, I only describe them to show that according to economic theory I expect to see a decrease in the exports from Israel to the Eurozone due to the adoption of the euro.

3.2 Export data

This thesis uses annual bilateral export data from 18 countries between the years 1970 to 2011. Every observation is the value of total exports in one year from one country to the other. For example, the total value of Israel’s exports to Germany in 1985 is one observation. I obtained all my export data from the International Monetary Fund Direction of Trade Statistics (IMF DOTS) database. The countries I choose to focus on are: Israel, Austria, Denmark, Finland, Germany, France, Italy, the Netherlands, Portugal, Spain, Greece, United Kingdom, Sweden, Switzerland, Canada, Japan, United States and Ireland. I choose almost all countries that formed the Eurozone when it was established and Greece that joined only two years later.

I did not include countries that joined the Eurozone later, because of the short time they had the euro. Belgium and Luxembourg are excluded due to problems of data availability, since for many years trade data for these countries was collected jointly. I also included all other EU countries that joined before 1995 and other advanced economies in Europe and other continents that have available trade data and that are significant trade partners of Israel. I do not include

0 5000 10000 15000 20000 25000 30000 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 valu e o f exp o rts in U SD m ill ion Year

Figure 1: Exports from Israel

(14)

14

Israel’s neighboring countries in my data set because, due to political reasons, trade between Israel and its neighbors is very low and there are problems with the availability of the data from these countries. I choose to have 1970 as the earliest year in my data set because this is the first year that trade data from Israel seemed reliable and consistent. I did not use data after 2011 due to data availability issues. Having data of bilateral export flows between 18 countries, in a time span of 41 years, results in 153 country pairs and a total of 12,853 observations.

Before doing a regression analysis with control variables, to estimate the euro effect on the Israeli exports, it is important to first look at the data and check for correlations and trends. Figure 1 presents Israel’s exports patterns. The horizontal axis represents years, from 1990 to 2011. The vertical axis shows the nominal value in millions USD of exports from Israel to countries in the Eurozone and other major trade partners of Israel. The upper line represents exports to the countries in the data set that did not join the Eurozone and the bottom line represent countries in the data set that joined the Eurozone.

Two things, which are not directly related to this research, appear in the graph. The first one is the decrease in exports due to the dot com crisis that took place around the year 2000 with the collapse of the internet bubble. And the second one is the decrease in exports after the financial crisis of 2007-2008. We are also able to see another trend from the graph. Exports from Israel to “other countries” have increased more sharply than exports to Eurozone countries in the years after the establishment of the Eurozone. Did export from Israel to Eurozone countries decrease due to the establishment of the euro? This research will try to provide an answer to this question.

3.3 The model

To estimate the effect of the euro on exports from Israel to Eurozone countries I use the gravity model. The basic gravity model, or the basic gravity equation, is a way to predict trade flows between two countries based on certain characteristic of these countries. The two most important characteristics in this equation are the countries’ GDP and the distance between the two countries. Evenett and Keller (1998) showed us that there are several economic theories that explain the gravity model. The increasing returns to scale from the Helpman and Krugman (1985)

(15)

15

model, are one of the gravity equation foundation. The two countries Heckscher Ohlin model with the differences in factors of endowment as well as other papers such as Anderson (1979) and Anderson and Wincoop (2001) provide theoretical foundation for the gravity equation.

According to the Helpman and Krugman (1985) increasing returns of scale model, the countries’ GDP is an indicator of the producer’s economics of scale. Countries with a relatively high GDP can produce cheaper goods due to lower cost that arise from economies of scale. As a result of the lower costs they are more competitive in the international market and can export more. Thus, according to the gravity model, a higher GDP has a positive effect on exports. Naturally, the GDP of the importing country is an indication of the demand for the goods offered by the exporting country.

The second component of the of the gravity equation, distance, referring to trade related costs such as the resources spent on transporting the exported good from one country to another. In addition, other factors were added to the basic form of the gravity equation. Factors like common border, common language, and levels of tariffs represent trade related costs like distance, from the basic form, which increase the price of the exported goods and hence reduce competiveness of the exporter.

In my research I will use a common form of the gravity equation, which is often used in the literature on the currency union effect on trade. The dependent variable is 𝐸𝑋𝑃𝑖𝑗𝑡. It is the logarithm of the nominal value of export, expressed in US dollars minus the logarithm of the US producers’ price index in year t. I use the US producer price index to correct for inflation. 𝐸𝑈𝑅𝑂1_𝑖𝑗𝑡 is a dummy variable that takes the value of one when the importing country is a part of the Eurozone and the exporting country is not. The coefficient of this variable will show how the euro affected exports from outside countries to Eurozone countries. The coefficient estimation of this variable is the purpose of this research.

𝐺𝐷𝑃_𝑖𝑡 is the logarithm of the exporting country’s GDP that represent economies of scale. 𝐺𝐷𝑃𝑗𝑡 is the logarithm of the product of the importing country in year t. 𝑅𝐸𝑅𝑖𝑗𝑡 is the logarithm of the nominal exchange rate of the exporting country vis-à-vis the importing country deflated by the consumer price index (CPI) of both countries. Put in other words 𝑅𝐸𝑅𝑖𝑗𝑡 is defined as

(16)

16

currency of the exporting country per currency of the importing country. So an increase means real depreciation of the exporting currency. The real exchange rate is important because it provides information about the exporters’ competitiveness. I expect a positive sign for this variable coefficient because a devaluation (the exchange rate increases – more exporters currency for importers currency) should increase the competitiveness of the exporter and hence increase exports. Another control variable is 𝐺𝐷𝑃𝐶𝑖𝑡 which is the logarithm of the GDP of the exporting country minus the logarithm of the population size of the same country in time t. 𝐺𝐷𝑃𝐶𝑗𝑡 Is the logarithm of the GDP of the importing country minus the logarithm of its population size in time t. Including GDP per capita as a control variable is done in almost all papers written in this field.GDP per capita also gives indication for the purchase power of the importing and exporting country, with the assumption that rich inhabitants of a country consume more exported goods.

In addition, in order to take into account barriers to trade such as tariffs, which increase the costs of exports, I include the dummy variable 𝐹𝑇𝐴𝑖𝑗𝑡. This variable takes the value of one in the case that both countries are in a free trade agreement in year t and zero otherwise. Since there are many degrees of trade integration agreements, I decided to make a clear cut and follow the lists provided by the world trade organization (WTO) regarding free trade areas agreements. Now I will present the two other different euro dummy variables. The first dummy variable is 𝐸𝑈𝑅𝑂_𝑖𝑗𝑡. This variable takes the value of one when both countries are members in the Eurozone in time t and zero otherwise. The second dummy variable is 𝐸𝑈𝑅𝑂2𝑖𝑗𝑡. This variables takes the value of one when the exporting country is a part of the Eurozone and the importing country is not. I include this variable to have a complete model where all the euro dummy variables are included.

I also include time fixed effect 𝜆𝑡, by using yearly dummies. In this way I control for changes that affect all countries, like the state of the world economy or a worldwide trade agreement that was signed after a negotiation round of the WTO. Furthermore, the 𝜂𝑖𝑗 parameter represents country pair fixed effects. By adding country pair dummies, I control for all the specific characteristics of country pairs that can influence trade like distance, political ties,

(17)

17

cultural ties, common language, shared border and so on. The country pair fixed effects is especially important when using the gravity model, because it controls for characteristics that might influence trade related costs. Distance is the most obvious one, since it is part of the basic form of the gravity equation, but also other characteristics can affect trade related costs like a common language which saves the money spent on translation and cultural ties which reduce the negative effect of cultural differences.

Following Bun and Klaassen (2007) I also include the term 𝜑𝑖𝑗 × 𝑡 which is country pair specific time trends. I do so because some country pair characteristics tend to change over time. Therefore, including time trends and country pair constants is not enough. An example for such a characteristic is transportation costs. Transportation costs are a country pair specific characteristic that depends on the distance between the two countries and on transportation technology. Because transportation technology changes over time, transportation costs are a country pair specific characteristic that changes over time and should be accounted for.

The model is the following:

𝐸𝑋𝑃𝑖𝑗𝑡 = 𝛽1𝐺𝐷𝑃𝑖𝑡+ 𝛽2𝐺𝐷𝑃𝐶𝑖𝑡+ 𝛽3𝑅𝐸𝑅𝑖𝑗𝑡 + 𝛽4𝐺𝐷𝑃𝑗𝑡+ 𝛽5𝐺𝐷𝑃𝐶𝑗𝑡+ 𝛾1𝐸𝑈𝑅𝑂𝑖𝑗𝑡+ 𝛾2𝐸𝑈𝑅𝑂1𝑖𝑗𝑡+ 𝛾3𝐸𝑈𝑅𝑂2𝑖𝑗𝑡+ 𝛾4 𝐹𝑇𝐴𝑖𝑗𝑡+ 𝜂𝑖𝑗 + 𝜆𝑡+ 𝜑𝑖𝑗 × 𝑡 + 𝜀𝑖𝑗𝑡 (1) Where 𝜀𝑖𝑗𝑡 is an error term. My assumption is that the error term is not correlated with the regressors. This assumption might be challenging especially because high levels of exports might encourage countries to join a monetary union and not the other way around. This is one of the reasons that I include fixed effects in my model. By adding time fixed effects I allow for a large part of this reaction between high levels of exports and joining a monetary union. Additionally, in my opinion, this is not the case because countries join a monetary union mainly due to political reasons. The adoption of the euro can be regarded as the next step of integration between European Union countries. That is why I assume that time developments in trade were not an important cause for this decision to adopt the euro. Furthermore, I allow the error term to have heteroskedasticity and serial correlation characteristics.

(18)

18

The source of the data: I obtained all the data through the DataStream program. Like I mentioned the export values data is from the IMF’s Direction of trade statistics. The US producer’s price index is obtained from the IMF’s international financial statistics. The information regarding free trade areas was received from Bun and Klaassen 2002, the WTO website, EFTA website, European Commission website and the Canadian Foreign Affairs, Trade and Development website. I got the GDP data, and the population data from the World Bank database. The CPI and Nominal Exchange rate were downloaded from the IMF International Financial Statistics.

4. Results

4.1 Results of the model

Table 1 presents the results of the model as described in the previous chapter. Because country pair specific time trends are a relatively new addition to the gravity model literature and are only estimated in the Bun and Klaassen (2007) study, I present two models. The first one is without country pair specific time trends (restricted model). This model represent the specific case where 𝜑𝑖𝑗 = 0. The second one is with country pair specific time trends (unrestricted model).

The 𝐸𝑈𝑅𝑂1𝑗𝑡 coefficient is not significant in both models. This means that according to my estimation, the euro effect on exports from outside countries to the Eurozone is statistically not different from zero. If we assume that exports from Israel are not different from other outside countries we see that the euro did not affect exports from Israel to the Eurozone.

Before discussing the coefficient of the other euro dummy, it is important to first focus on the control variables and see if they are in line with the theory behind the gravity model. The coefficient for 𝐹𝑇𝐴𝑖𝑗𝑡 is positive and significant in both models, these findings are the same as those we expect form theory, that is, free trade areas that reduce barriers to trade, increase exports. The coefficient of 𝑅𝐸𝑅𝑖𝑗𝑡 also follows the gravity model theory as it is positive and significant. It is positive because we anticipate real exchange rate appreciation to cause more exports once the exporting country becomes more competitive. The coefficients of GDP per capita of the exporting country as well as for the importing country are positive and significant,

(19)

19

like theory suggests, because the demand for imports in richer countries is higher and the supply of exports in the exporting countries is also higher.

TABLE 1

Estimation results for equation (1) Without country pair time trends

With country pair time trends

Euro_ijt 𝛾1 0.306** -0.135*

(membership in the Eurozone) (0.062) (0.053)

Euro1_ijt 𝛾2 0.021 -0.08

(outsiders export to the Eurozone) (0.063) (0.054)

Euro2_ijt 𝛾3 0.197** -0.034

(exports of Eurozone to outsiders) (0.06) (0.05)

FTA_ijt 𝛾4 0.346** 0.140**

(membership in a free trade area) (0.042) (0.028)

GDP_it 𝛽1 -0.156 -0.013

(income exporting country) (0.211) (0.454)

GDP_jt 𝛽2 0.544* 0.278

(income importing country) (0.219) (0.398)

GDPC_it 𝛽3 1.679** 0.804*

(income per capita importing country) (0.284) (0.401)

GDPC_jt 𝛽4 0.626* 1.278**

(income per capita exporting country) (0.268) (0.417)

RER_ijt 𝛽5 0.014** 0.008

(real exchange rate) (0.005) (0.004)

_cons -12.592* -18.105

(5.138) (11.071)

r2 0.806386 0.908387

N 12852 12852

Note: in parentheses are the robust for heteroskedasticity and serial correlation standard errors; *significant at 5% level; **significant at 1% level.

Unlike the variables presented so far, the 𝐺𝐷𝑃𝑖𝑡 coefficients are not significant. This is probably caused by the high correlation between GDP and GDP per capita1_{. These variables are} highly correlated because the logarithm of GDP per capita is simply the logarithm of GDP minus the logarithm of the population size. So, GDP appears in both variables. The coefficient of 𝐺𝐷𝑃𝑗𝑡 in the restricted model is positive and significant like theory suggests but in the unrestricted

1_{MSO (2003) encounter similar patterns with the sign and significant of the GDP coefficients. In response they}

(20)

20

model it is insignificant, again probably due to the correlation between GDP and GDP per capita. As for the differences between the restricted version and the unrestricted version, it is easy to see that almost all coefficients in the unrestricted version are different from the restricted model and less significant. This can be explained by misspecification of the restricted model where country pair specific time trends are not including causing an omitted variable bias. Controlling for this bias changes the estimation to be less significant.

Now I turn to discussing the 𝐸𝑈𝑅𝑂𝑖𝑗𝑡 coefficient. Clearly there is a big difference between the restricted model estimation and the unrestricted one. The 𝐸𝑈𝑅𝑂𝑖𝑗𝑡 coefficient is 0.306 corresponding with a 36% increase in exports due to the euro effect2_{. This result is similar to the} one of Bun and Klaassen (2002) which found that the euro increased exports by 37.6%. The estimation of 𝐸𝑈𝑅𝑂𝑗𝑡 in the unrestricted model is -0.135 corresponding with 12.6% decrease in exports due to the euro. This coefficient is significant and very different from the one of the other model. These findings are surprising because they do not follow the economic theory that is presented above, which assumed that the lack of exchange rate volatility should lead to an increase in trade. It is also surprising because none of the other studies in this field found a negative effect of the euro on trade.

That the estimation results are different from the one of other studies can be partly explained by the difference in the data. Unlike some other studies, my data contains more observations from after the establishment of the Eurozone. Another difference is that some researchers use trade flow data and not only export data. In this respect Bun and Klaassen (2007) found, in their sensitivity analysis, that changing the dependent variable from trade to exports reduces the euro effect on trade. To examine whether the euro coefficient is negative due to the sovereign debt crisis in the Eurozone countries I ran the same model but with the last year of the data being 2005, 2004 and 2003. The estimated euro coefficient was similar, so the Eurozone debt crisis is not the reason for the estimated negative effect of the euro on trade.

2_{Due to the fact that 𝐸𝑋𝑃}

𝑖𝑗𝑡 is the natural logarithm of exports, the different euro coefficients indicate the relative

(21)

21

With respect to the contradiction between the economic theory about the effect of currency unions on trade and my findings, I have thought about possible explanations of why the euro can reduce trade. Establishing a currency union is not the only aspect of the economic integration between the Eurozone countries. These countries also reduced barriers to trade other than their different currencies. One of the barriers, which I already control for in my models, are tariffs. I do it with the FTA dummy. But there are more barriers to trade such as different legal systems, government subsidies and other policies that favor local production. The countries that joined the Eurozone also reduced these other trade barriers. Non-EU countries also reduced those barriers to trade but not to the same extent. EU economies have become more integrated in many ways after a series of treaties and acts, such as the single European act. In addition, due to the single market between Eurozone members it is possible that some producers, instead of exporting their goods, establish factories or subsidiaries in the foreign market. This can result in less exports and more outsourcing. The firms choose to outsource because, in certain conditions, it is more efficient to outsource than to exports. Outsource can be done through foreign direct investment, that is establishing factories or service centers in the foreign country. The possibility to outsource, due to economic integration, can lead to less exports.

This hypothesis can find theoretical roots in the theory of international trade. According to this theory, foreign direct investment (FDI), which is a form of outsourcing, substitutes for exports. Markusen et al (1995) write about this substitution in their article “Direct foreign investment and the multination firm”. They introduce the concept “location advantage” which is one of the reasons why firms engage in FDI. This advantage affect the production decision of the firm, that is, to engage in FDI rather than to export. The drivers of the location advantage are lower transportation costs, cheaper labor in the foreign country, tariff jumping, and other more elusive factors such as customer access. My main point is that according to the authors there is a substitution between exports and FDI.

Another consideration for the firm to engage in outsourcing is the restriction of FDI in the foreign country. Golub (2003) studied the restrictions to foreign direct investments in OECD countries. According to his paper, some of these restriction are on foreign ownership, obligatory

(22)

22

screening and approval procedures, limitations of foreign nationals to take part in the firm management and also informal unintelligible public or private restrictions. In respect to FDI restriction in the countries of my data set, the OECD publish an FDI restrictiveness index where countries are given a grade from 0 to 1 base on their FDI restriction measures. Where 1 is completely closed to FDI and 0 is completely open. The average grade for the Eurozone countries in my sample is 0.041 where for the other countries it is 0.1. The average for the EU countries that did not adopt the euro is 0.059. Hence the Eurozone countries in my data set are 247 percent more open to FDI than the other countries and 41 percent more open than the rest of the EU countries. These statistics suggest that it is possible that Eurozone firms engage in FDI more than firms outside the Eurozone due to the low FDI restrictions. It also suggest that outside courtiers should engage more in FDI in the Eurozone countries. Thus, the substitution between FDI and exports might contribute to the negative euro effect on exports that I found in my research.

Furthermore, this can partly explain the negative estimation of the euro effect on trade with outside countries. It can explain it through the substitution between exports and FDI of the outside countries. The Eurozone is more open to FDI. So, exporters from outside the Eurozone establish factories and service centers inside the Eurozone that cause a reduction in exports.

4.1 Sensitivity analysis

In this section I check for the robustness of my results by changing the estimation methods. I do so by adding other controls variables to equation 1, such as the lags of exports and the interaction term between the Israel dummy and the 𝐸𝑈𝑅𝑂1_𝑗𝑡 dummy. I also leave out the GDP per capita from the control variables. Tables 2 presents the estimation with these changes. The first three columns show the estimations with only one change at a time, the forth column show the estimation with all the changes and the fifth column show the long run effect, of some chosen variables, of the estimation with all the changes.

(23)

23

TABLE 2

Estimation results for changes in equation (1) with country pair time trends

With Israel interaction

Without GDPC With lags All changes Long run Euro_ijt γ1 -0.136* -0.124* -0.056** -0.050** -0.166 (Eurozone membership) (0.053) (0.052) (0.019) (0.019) Euro1_ijt γ2 -0.102 -0.072 -0.046* -0.043* -0.143

(exporters outside Eurozone) (0.054) (0.054) (0.02) (0.019)

Euro2_ijt γ3 -0.032 -0.026 -0.025 -0.02 -0.07

(Importers outside Eurozone) (0.05) (0.049) (0.018) (0.018)

FTA_ijt γ4 0.142** 0.139** 0.063** 0.063** 0.209

(free trade area) (0.028) (0.028) (0.013) (0.012)

GDP_it β1 -0.033 0.708** -0.183 0.253** 0.84

(exporter's income) (0.452) (0.164) (0.166) (0.051)

GDP_jt β2 0.283 1.426** -0.083 0.568** 1.87

(Importer's income) (0.399) (0.11) (0.142) (0.056)

GDPC_it β3 0.825* 0.483**

(exporter's GDP per capita) (0.4) (0.154)

GDPC_jt β4 1.272** 0.723**

(Importer's GDP per capita) (0.418) (0.139)

RER_ijt β5 0.010* 0.009 0.001 0.001 0.003

(real exchange rate) (0.004) (0.004) (0.002) (0.002)

EURO1ISR_it γ5 0.149 0.014 0.046

(Israel exporting to Eurozone) (0.114) (0.055)

lag1 δ1 0.611** 0.616**

(firs lag of exports) (0.049) (0.049)

lag2 δ2 0.088* 0.087*

(second lag of exports) (0.043) (0.043)

_cons -18.072 -53.381** 2.619 -17.594** (11.123) (4.469) (3.865) (1.926) r2 0.908475 0.90755 0.94738 6 0.947075 N 12,852 12,852 12,240 12,240

Note: in parentheses are robust for heteroskedasticity and serial correlation standard errors; *significant at 5% level; **significant at 1% level; the long run effect calculation for the 𝐸𝑈𝑅𝑂1𝑗𝑡 coefficient, for example, is done

using the following equation: _{1− 𝛿}𝛾2

1−𝛿2

In the previous subsection I assumed that the euro effect on exports from outsiders, in general, is similar to the one on Israel in particular. This assumption might not be valid due to the special characteristics of Israel. For example, it is the only small open economy that is not a member of the European Union in my data set other than Switzerland. So that the effect of the euro on trade with big and closed economies might be different than the one on small and open

(24)

24

economies. The effect of the euro can be different between these two groups of countries for many reasons. Perhaps it is easier for exporters in big economies to sell their products in the domestic market once they become a bit less competitive to the consumers in the Eurozone.

To estimate the euro effect on exports from Israel I include the variable EURO1ISR_it. This variable is the interaction between two dummy variables. It takes the value of one for exports from Israel to Eurozone countries and zero otherwise. As you can see from the estimation results of the column headed by “with Israel interaction” the coefficient of this parameter is not significant. So, my conclusion is that the euro effect on exports from Israel is not different from the one on outside countries and according to my estimation the latter is not statistically different from zero.

Another change to the estimation of equation 1 is leaving out the GDP per capita control variables. I do so because of the high correlation between the two variables. Like I mentioned above, according to gravity model theory, the effect of GDP of both exporting and importing countries should be positive. This is one of the basic principles of the gravity model and that is why it is important for me to make sure that I estimate the GDP coefficients without a bias. The estimation results from the second column show that both 𝐺𝐷𝑃𝑖𝑡 and 𝐺𝐷𝑃𝑗𝑡 are strongly positive and significant, as expected from the gravity model theory.

The last change I made was to include two lags of exports in my model. As Goldstein and khan (1985) showed us, exports are dynamic due to the lag between the decision of the exporter the export and the actual export flow. Exports flows are dynamic because they are influenced by the sunk costs of the exporter. For example an exporter that invest in establishing distribution centers or in marketing campaigns will export more in the next period than in the current one. Another reason that exports are dynamic is that consumers in the importing country acquire a taste for a specific good, meaning that they will demand more of this good also in the next period and the one afterwards.

In the third column I include the estimation for only the significant lags, that is the first and the second lags. As predicted the coefficients of both lags are positive. Now, the long run effect of the euro on exports from outside countries is -14%. Hence, the euro has decreased

(25)

25

exports from outside countries to the Eurozone by 14% this estimation is in line with my baseline results. To complete the dynamic model estimation I also added two lags of the GDP, GDP per capita and real exchange rate variables. The long run coefficient of 𝐸𝑈𝑅𝑂1𝑗𝑡 is then -0.158, a very close results to the one without the other lags.

5. Conclusion

This thesis focused on the question what the effect of the euro on exports from Israel to the Eurozone countries is. To answer this question I used bilateral exports data from 18 countries in a time span of 41 years from 1970 to 2011. To estimate the euro effect on exports, this study uses the gravity model with exports as my dependent variable, different euro dummies and other control variables that are common in this field of research such as GDP, GDP per capita, the real exchange rate and a free trade area dummy. I also included time fixed effects, country pair fixed effects and country pair specific time trends. I found that the euro effect on exports from Israel to the Eurozone is not statistically different from zero leading me to believe that the euro effect on exports from Israel is in economic terms not high.

This result is positive for the Israeli economy, which is a small and open economy, dependent on high levels of exports to the Eurozone. It is good news because it means that there is no evidence that exports from Israel to the Eurozone countries have changed significantly and are neither expected to change in the future due to the adoption of the euro. Another finding of my research is that the euro effect on exports from outside countries to the Eurozone in general is also not statistically different from zero. This means that I did not find evidence that the euro caused trade diversion.

As for the euro effect on trade within the Eurozone, my estimation is that the euro decreased trade within the Eurozone by 12.6%. This is an unexpected result because it is contradictory to the economic theory and to findings of other papers in this field. In respect to the contradiction with other researches it might be due to the fact that my study is among the few that uses country pair specific time trends and export data over a longer period of time after the euro was introduced. In respect to the contradiction with economic theory I have presented a hypothesis that can explain this surprising result. According to my hypothesis, the substitution

(26)

26

between FDI and exports in the economic interaction between the Eurozone countries can partly explain the negative euro effect on exports. A suggestion for further research will be to investigate the substitution between FDI and exports within the Eurozone more in dept.

(27)

27

References

Anderson, J. E., & Van Wincoop, E. (2001). Gravity with gravitas: a solution to the border puzzle (No. w8079). National bureau of economic research.

Anderson, J. E. (1979). A theoretical foundation for the gravity equation. The American Economic Review, 106-116.

Baldwin, R. E., & Nino, V. D. (2006). Euros and zeros: The common currency effect on trade in new goods (No. w12673). National Bureau of Economic Research working paper.

Baldwin, R. E., Di Nino, V., Fontagné, L., De Santis, R., & Taglioni, D. (2008). Study on the impact of the euro on trade and foreign direct investment. European Economic and Monetary Union Working Paper, (321).

Berger, H., & Nitsch, V. (2008). Zooming out: The trade effect of the euro in historical perspective. Journal of International Money and Finance, 27(8), 1244-1260.

Blanka K., Palerm A. & Thomsen S. (2010). OECD’s FDI Restrictiveness Index: 2010 Update, OECD. Bun, M. J. G. and Klaassen, F. J. G. M. (2002). Has the Euro Increased Trade? Tinbergen

InstituteDiscussion Paper No. 02-108/2, University of Amsterdam.

Bun, M. J.G. and Klaassen, F. J.G.M. (2007). The Euro Effect on Trade is not as Large as Commonly Thought. Oxford Bulletin of Economics and Statistics, 69(4), 473-496.

Central bureau of statistic Israel. (2013). Statistical Abstract of Israel. 2013-No.64. Available at: www.cbs.gov.il/reader/shnatonenew_site.htm

De Nardis, S., & Vicarelli, C. (2003). Currency unions and trade: The special case of EMU. Review of World Economics, 139(4), 625-649.

Emerson, M., Gros, D., & Italianer, A. (1992). One market, one money: an evaluation of the potential benefits and costs of forming an economic and monetary union. OUP Catalogue.

Evenett, S. J., & Keller, W. (1998). On theories explaining the success of the gravity equation (No. w6529). National bureau of economic research.

Flam, H. & Nordstrom, H. (2003). Trade volume effects of the euro: aggregate and sector estimates, mimeo, Institute for International Economic Studies, Stockholm University.

Goldstein, M., & Khan, M. S. (1985). Income and price effects in foreign trade. Handbook of international economics, 2, 1041-1105.

Golub, s. (2003). Measures of restriction on inward foreign direct investment for OECD countries. OECD Economic Studies No. 36, 2003/1

Helpman, E. & P. Krugman (1985). Market Structure and Foreign Trade, MIT Press, Cambridge, MA. Hooper, P., & Kohlhagen, S. W. (1978). The effect of exchange rate uncertainty on the prices and volume of international trade. Journal of International Economics, 8(4), 483-511.

(28)

28

IMF (2014). Report for Selected Countries and Subjects. World Economic Outlook Database. Available at: www.imf.org/external/pubs/ft/weo/2014/01/weodata/weorept.aspx?sy=2014&ey=2014&scsm=1&ssd= 1&sort=country&ds=.&br=1&c=436&s=NGDPD%2CNGDPDPC%2CPPPGDP%2CPPPPC&grp=0&a=&pr.x=7 5&pr.y=20

Kenen, P. (2003). Making the Case for the Euro. The International Economy, 51-54.

Klaassen, F. (2004). Why is it so difficult to find an effect of exchange rate risk on trade? Journal of International Money and Finance, 23(5), 817-839.

Markusen, J.M., J.R. Melvin, W.H. Kaempfer & K.E. Maskus (1995). Direct foreign investment and multinational forms. In: Markusen, J.M., J.R. Melvin, W.H. Kaempfer & K.E. Maskus. International Trade: Theory and Evidence, McGraw-Hill, New York, Chapter 22.

Micco, A., Stein, E. & Ordonez, G. (2003). ‘The currency union effect on trade: early evidence from EMU’, Economic Policy, Vol. 37, pp. 315–356.

Rose, A. (2000). ‘One money, one market: the effect of common currencies on trade’, Economic Policy, Vol. 30, pp. 9–45.

Soffer, Y. (2007). To Join or not to Join a Currency Union: the Case of Israel. Bank of Israel, Foreign Exchange Activity Department, The Foreign Exchange Discussion Paper Series, No. 1/07 February 2007 Viner, J. (1950). The customs union issue. New York, Carnegie.

WTO (2013).Trade Policy Review: Israel, Report by the Secretariat: Israel. WT/TPR/S/272 /Rev.1.

( .ע ,ןמדירפו .י ,איבל 5002 ,רקחמה תקלחמ ,לארשי קנב .לארשי לש ץוחה רחסו ילאירה ןיפילחה רעש .) תרדס ןוידל םירמאמ 5002.50 .'בונ ,

(29)

29

The effect of the Euro on trade between Israel and the Eurozone countries