Distance still matters!
Author: Heleen van Gorcum Student number: 1257013 July 2007
University of Groningen Faculty of Economics
International Economics & Business
Supervision:
Abstract
Content
Content
...3List of Figures
...4List of Tables
...4Chapter 1: Introduction
...6 1.1 Introduction...6 1.2 Problem Statement ...8 1.3 Research questions...9 1.4 Definitions...10 1.5 Research Design...111.6 Structure of the paper...12
Chapter 2: History of globalization
...132.1 Introduction...13
2.2 When did globalization start? ...13
2.3 The most recent of globalization...19
2.4 Conclusion ...21
Chapter 3: Discussion on the influence of distance on trade
...223.1 Introduction...22
3.2 Distance does not matter anymore...22
3.3 Distance still matters...25
3.4 Possible explanations for the observed increasing importance of distance in international trade. ...29
3.5 Modelling the influence of distance in international trade ...32
3.6 Conclusion ...35
Chapter 4: At First Glance, Comparing trade patterns
...374.1 Introduction...37
4.2 Construction of the graphs ...37
4.3 Data availability ...38
4.4 Different categories of countries...39
4.5 Results...42
4.6 Conclusion ...43
Chapter 5: Results regression analysis of trade patterns
...445.1 Introduction:...44
5.2 Construction of the model...44
5.3 Data sources and availability ...46
5.4 Results...47
5.5 Conclusion ...53
Chapter 6: The influence of distance during the most recent phase of
globalization.
...546.1 Introduction...54
6.2 Construction of the model...54
6.3 Results of the estimated regression equation ...55
6.4 Explaining the results...58
6.5 Conclusion ...60
7.1 Introduction...61
7.2 Answers to the sub questions...61
7.3 Answer to the research question ...65
7.4 Limitations and recommendations for further investigation...66
Appendix I: References
...67Appendix II: Data Sources
...71Appendix III: List of Countries included in the Analysis
...72Appendix IV: Graphs of the Trade Patterns 1995 and 2004 for Individual
Countries.
...74List of Figures
Figure 1: Development of Transportantion and Communication costs, 1920-1990...6Figure 2: Timeline of the Waves of Globalization. ...13
Figure 3: Spice and Coffee Mark-ups, Amsterdam vs. South-East Asia, 1580-1939. 15 Figure 4: Regions' Share of World GDP. ...16
Figure 5: Openness of Various Developed Countries, 1870-1995. ...17
Figure 6: Openness of Various Developed Countries, 1990-2005. ...18
Figure 7: Transportation Bill (Freight Only) devided by GDP...23
Figure 8: Trends in Average Applied Tariffs Rates in Developing and Industrialized Countries. ...24
Figure 9: Results from the Regression Equation by Brun et al...27
Figure 10: Agglomerations in Europe, "The European Banana"...30
Figure 11: Agglomeration and Economic Intergration in the Abscence of Labor Mobility...30
Figure 12: Trade Distance versus Transaction Costs...34
Figure 13: Bilateral Trade at Larger Distances has become relatively more important. ...40
Figure 14: Bilateral Trade at Larger Distances has become relatively less important.40 Figure 15: The Bilateral Trade Pattern did not change with respect to Distance. ...41
List of Tables
Table 1: Direction of Trade, UK and Europe, 1860, 1910 and 1996...7Table 2: Telephone Capacity & Internet Hosts, 1986-2000. ...19
Table 3: Growth of World Internet Usage, 2000-2007...20
Table 4: Overview of the Results of Different Studies performed. ...27
Table 5: Description and Measurement Units of Variables...39
Table 6: The Classification of Counties with respect to changes in their Trade Pattern. ...42
Table 7: Description and Measurement Units of the Variables...46
Table 8: The Results of the Estimated Regression Equation for Individual Countries for 1995...47
Table 9: The Results of the Estimated Regression Equations for Individual Countries for 2004...48
Table 10: The Results of the Estimated Regression Equation...51
Table 12: The Results of the Estimated Regression Equation...55
Table 13: The Quality of the Estimated Regression Equation...56
Table 14: Classification of the Countries based on the Results of the Estimated Regression Equation. ...58
Table 15: Overview of the Data Sources used in the Thesis. ...71
Table 16: List of Exporting Countries included in the Analysis. ...72
Chapter 1: Introduction
1.1 Introduction
In a globalizing world distances seem to vanish. Due to rapid declining transport costs and new developments in communication techniques the world is getting increasingly integrated. The rise of the internet in the last decade of the previous century enabled individuals to communicate, compete, share knowledge and operate globally at high speed and low costs.
Figure 1: Development of Transportantion and Communication costs, 1920-1990.
Figure 1 is obtained from the World Development Report of 1995 and shows a significant downward trend in costs of ocean freight, air transport, satellite and transatlantic calls. Figure 1 shows a sharp decrease from 1975 for the costs of satellites and for transatlantic calls from 1950. However, the graph also shows a flattening in the downward trend of transportation and communication costs and costs of ocean freight seems even to slightly increase again.
international bestseller. Friedman argues that due to the revolutions in transportation and communication techniques the established relationships between continents, countries, companies and individuals have shifted, which resulted in a tiny world where distances do not matter anymore. (Friedman, 2006)
The volume of trade is increasing rapidly. The exports of the Netherlands increased from 27.5 billion Euros in 1970 to €328 billion Euro in 2004 (source: Unstat). On the
other hand when looking at the main trade partners of the Netherlands, it appears that Germany, Belgium and the United Kingdom have remain the most important trading partners for the Dutch for decades.
Table 1: Direction of Trade, UK and Europe, 1860, 1910 and 1996.
Source: Baldwin and Martin, 1999
results in a decrease of bilateral trade of 9% (Disdier and Head, 2006). This might seem contradictory to what one would expect to happen in a globalizing world and to the believe that new developments in technology made distances between countries smaller.
1.2 Problem Statement
As can be concluded from the introduction above there is no consensus on this subject. Several studies have examined the influence of distance on trade flows and came up with different conclusions. Francess Cairncross and Thomas Friedman both wrote a book on the effects of the globalizing world. The titles, respectively, “The Death of Distance” and “The World is Flat” immediately give away their view on the globalization process (Cairncross, 1997; Friedman, 2006). However, several studies examined the influence of distance on trade, mostly using so called gravity analysis, and almost every study confirms that distance still plays a significant role in international trade and some studies even found an increase in the importance of distance (Brun et al., 2003; Disdier and Head, 2006).
All studies already performed on this subject are examining the overall relationship between distance and trade flows for a group of countries, while the effects for individual countries might be quite different. After all, the characteristics of the Dutch economy differ in many aspects from the economic situation of the United States or Japan.
Although the globalization process might manifest itself in many different aspects of economy and life in general, the focus of this thesis will be on the change of influence of distance on trade flows on individual countries. There are several reasons to limit the analysis to the trade patterns of countries. First, globalization is a broad subject and it is impossible to examine all aspects in one thesis. Furthermore, trade has always had a large impact on debates concerning globalization. For example, people fear that as consequence of globalization, and the assumed shrinking size of the world, production will be outsourced to developing countries that can produce at lower costs than developed countries. People are afraid for the increasing amount of imports from e.g. Asian countries, because it competes with products produced at home. Thirdly, several studies examining the influence of distance have already preformed. Most of them using the gravity model, using trade flows corrected for GDP of the respective countries as a proxy of the masses of the countries. Examining trade flows in the empirical analysis of this thesis makes it possible to compare results.
1.3 Research questions
The main research question of this thesis is:
“To what extent does distance influence trade patterns of individual countries and did the influence of distance changed significantly during the most recent phase of globalization?”
To answer the research question stated above several sub questions are formulated:
1. How did globalization developed through history and in what way differs the most recent phase of globalization from previous ones.
Despite the fact that companies operate around the world for centuries, there does not seem to be consensus on the exact starting point of globalization. However most scholars do agree that globalization appears to hit the world in different phases. The answer to this sub question is supposed to give an understanding of the characteristics of globalization and why the most recent phase of globalization might have led to the discussion that distance does not matter anymore.
2. What are the arguments given in the discussion if distance still matters and what are the results of the studies already performed on this subject?
The statement that distance has become irrelevant or will become irrelevant in the near future has advocates and opponents. The answer to this sub question should give insights to the arguments given in favour and against the occurrence of the “death” of distance. Furthermore, it provides an overview and discussion of the studies already performed on this topic.
3. How well does distance explain the change in trade patterns of individual countries?
change visible at all? Furthermore, are these results the same for all countries or are there large individual differences?
4. Can one observe a significant change in the influence of distance on bilateral trade flows?
Finally, it will be examined if during the most recent phase of globalization one can observe a significant change in influence of distance on trade flows for individual countries. Are the results applicable to all countries, certain groups of countries (e.g. EU, NAFTA) or are there large individual differences and how can they be explained.
1.4 Definitions
To prevent confusion as to the exact meaning of the concepts used in this thesis a lists of definitions is given in this section.
Globalization: The term globalization is used heavily. However, what exactly is meant by globalization remains often unclear. Globalization is often used as an umbrella term. Globalization can comprise many fields, such as institutional and social fields of integration e.g. social policy of the convergence of consumer ‘tastes’. (O’Neill, 2004)
The definition in the Merriam-Webster dictionary given is:
glob·al·i·za·tion: The development of an integrated global economy market especially
by free trade, free flow of capital, and the tapping of cheaper labor markets (http://www.m-w.com/dictionary/Globalization).
Wikipedia, the free encyclopedia on internet, defines globalization as follows: “Globalization refers to increasing global connectivity, integration, interdependence in the economic, social, technological, cultural, political and ecological spheres”. “In economics is the convergence of prices, products, wages, rates of interests and profits towards developed country norms” (http://en.wikipedia.org/wiki/Globalization)
Williamson (2002). They examine one aspect of globalization and define globalization as the integration of international commodity markets. In a world without any trade restrictions or transport costs, international commodity markets would be perfectly integrated, which would result in the same prices for commodities at home as well as abroad. Unfortunately, in reality, transport costs and trade barriers are present and this results in difference between the prices charged at home and abroad. Globalization is defined as a decline in these price differences, which can be revealed in falling import prices, rising export prices and an increase in trade volumes. An evidence of present globalization occurs when trade-creating forces change domestic commodity prices first, before anything else can happen and as a result of the changes in domestic commodity prices this should lead to a rearrangement of resources and influence essential matters, such as the income distribution, living standards and quality of life. (O’Rourke and Williamson, 2002) When globalization is defined as commodity price convergence this made it possible to examine the globalization process empirically, while it mostly is a highly qualitative discussion. According to Williamson and O’Rourke it is price commodity convergence that is important but historians often look at different indicators, such as shipping technologies, trade volumes or port history. These indicators, however, can also result from changes in demand and/or supply at home as well as abroad and may not correlate with commodity price convergence. (O’Rourke and Williamson, 2002) Trade Costs: The definition used in this paper is the definition of trade costs given by Anderson and Wincoop in their paper “Trade Costs” (2004). Broadly defined trade costs include all costs in getting a good to a final user other than the marginal costs of producing the good itself. This includes: transportation costs, policy barriers, information costs contract enforcement costs, costs associated with the use of different currencies, legal and regulatory costs and local distribution costs. (Anderson, 2004)
Gravity Model: The gravity model is often used in social sciences to explain relationships between variables that show characteristics of gravitational interaction. Jan Tinbergen was the first who applied the model to explain economic relationships. In economics the model is used to predict bilateral trade flows using the GDP, as a proxy for the masses of the countries involved, and the distance between the two countries.
1.5 Research Design
research on the topic, the influence of distance on trade, where the focus in this thesis is on the effects on individual countries instead, what is examined already in most existing literature on this subject, the effects on whole group of countries.
To answer the first two sub questions existing literature and researches are examined. For the first sub question, literature concerning the development and history of globalization are examined in order to establish a clear understanding of the period under investigation and the reasons to focus on this specific period.
For the second sub question the focus is on existing studies and literature on the influence of distance on international trade. The goal of this question is to give an overview and analysis of the arguments given in favour and against the effects of globalization specific on the supposable ‘death’ of distance. The main part of the sources of information used consists of empirical studies, most of them using the gravity analysis to estimate the casual relationship between distance and trade, as well as more ‘popular’ literature on this subject, such as the international bestseller “The World is Flat” from the Thomas Friedman (2006) and the book “The death of distance” of Francess Cairncross (1997).
Sub questions 3 and 4 are answered in the empirical part of the thesis. In this part the data is collected from databases from the OECD, UNstat and CEPII. A limitation of this research is that it is highly dependable on the quality and the data given by these publicly available databases. The data is analysed by performing regression analysis using cross-sectional data and hypothesis testing. To perform this analysis the statistical programme Eviews was used.
1.6 Structure of the paper
Chapter 2: History of globalization
2.1 Introduction
It can not be denied that the world is getting increasingly integrated and a rising number of countries are facing the opportunities and challenges of globalization. In this thesis, globalization is examined in its economic meaning, since globalization is an important process, reshuffling the relationships between countries, companies, organizations and individuals worldwide. This chapter provides an overview of the development of globalization through history to put the developments and trends that can be observed today into perspective.
2.2 When did globalization start?
Despite the fact that companies operate around the world for centuries, there is some discussion on the exact starting point of globalization. Most scholars however do agree that the globalization process appears to hit the world in different phases. Figure 2 shows an overview of the different opinions of several scholars on the starting point of globalization and the development of globalization through history. Some scholars assigned 1492, the year Columbus discovered America, as the starting point of globalization while others state that globalization did not start until the 19th century.
Figure 2: Timeline of the Waves of Globalization.
Friedman (2006) in his book “The World is Flat” makes the case for the year 1492, the discovery of America by Columbus, as starting point of the globalization process. Columbus’ discovery of America resulted in a growing awareness in countries in the ‘old’ world of trade opportunities with the New World. Countries were determined to exploit the newly discovered trade opportunities. Scholars who argue that the Industrial Revolution should be assigned as starting point do not deny that trade volumes started to increase before the 19th century, but claim that this was not the result of globalization since international trade was dominated by monopolies and there is no evidence, for instance, of commodity price convergence (Williamson and O’Rourke, 2002). Friedman distinguishes in total three different phases of globalization. The second phase started parallel to the Industrial Revolution around 1800 and lasted until approximately 2000. In this phase companies started to expand their activities worldwide. The developments in this phase were interrupted by the first and Second World War when countries and companies withdrew behind their own borders. The last phase distinguished, by Friedman, started around a decade ago when individuals started to operate and compete globally. According to Friedman, the developments, the developments in communications and transportantion techniques, in this phase led to ‘death’ of distance. (Friedman, 2006)
As can be observed from the overview in figure 2 Friedmans opinion about the starting point of globalization is not shared by most economists. Williamson and O’Rourke (2002), Crafts and Venables (2001), Baldwin and Martin (1999) all assigned the period of the Industrial Revolution that started at the end of 19th century, as starting point of the first phase globalization.
Figure 3: Spice and Coffee Mark-ups, Amsterdam vs. South-East Asia, 1580-1939.
Source: Williamson, 2002
Williamson and O’Rourke (2002) examined the price gaps for the commodities cloves, coffee and pepper. Figure 3 shows the mark-ups for cloves, coffee and black pepper for Amsterdam vs. South-East Asia. The prices for black pepper and coffee remained considerably stable until the French wars in 1850 when the mark-ups increased rapidly for that moment. After the French wars had ended the price convergence for these goods accelerated. Cloves form an exception, since the price convergence seems to already have started around 1750. Williamson and O’Rourke (2002) also made comparable graphs for the trade between England and South-East Asia, showing an increase in price convergence after the end of the French Wars. Although trade volumes did significantly increase during earlier periods, these graphs confirm the statement of Williamson and O’Rourke that globalization, defined as commodity price convergence, hardly started until the 19 century. For globalization to take off, it was necessary that the monopolies on long-distance trade, e.g. Dutch East Asia Company, did break down first.
The commodity price convergence is one way to look at globalization. Other possibilities are, for example, level of exports, the location of production and GDP, Factor price convergence and FDI. Crafts and Venables (2001) examined GDP and industrial output, share in population and share in manufacturers output.
Figure 4: Regions' Share of World GDP.
Source: Crafts, 2001
Looking at the location of production and GDP, Crafts and Venables (2001) classify the developments in the following three phases.
1. The 1st phase started in the late 18th and early 19th century, with the Industrial Revolution. The economies of the UK and the Western World started to develop rapidly and the economies of China and India crashed dramatically. This lasted through the second part of the 19th century. During this phase economic activity was mainly concentrated in the North Western World. The production of North Western World accounted for more than the half of world’s GDP and with only one-fifth of the world population.
2. The second phase is characterized by the rise of North America. This started around 1865, from the American Civil War and reaches its peak just after World War II. Observing the trend in GDP per capita, North America started to take over the first position of the UK. Latin America lagged behind during this period. Falling transport costs and innovations in communications continued to encourage the process of globalization.
Japan and other Newly Industrializing Countries in East-Asia. Foreign direct investment became increasingly important during this period. (Crafts, 2001) Baldwin and Martin (1999) examined the effect of the globalization process on trade by looking at the Trade-to-GDP ratios of countries. This ratio is often used as a proxy for the openness of a country. Hereby it is assumed, the higher the Trade-to-GDP ratio, the higher the integration level of the county in international trade.
Figure 5: Openness of Various Developed Countries, 1870-1995.
Total Trade to GDP 0 20 40 60 80 100 c.1870 c.1910 c.1950 c. 1995 Year % UK France Germany Sweden USA Japan
Source: Baldwin and Martin, 1999
Based on this results Baldwin and Martin (1999) assigned the Industrial Revolution as starting point of the first wave of globalization. Although they state that it is hard to assign a specific date, since the Industrial Revolution was the result of several small inventions and not just one technological break through. According to Baldwin and Martin (1999) the first phase of globalization was characterized by industrialization of North America and European Countries as a result of improved transportation and development of financial intermediaries. The second wave started around 1950, the globalization process was delayed by the First and Second World War. This can be observed from the graph by the declining Trade-to-GDP ratios for all countries. From around 1950 a clear increase in the Trade-to-GDP ratio’s can be observed for almost all countries except for Japan. According to Baldwin and Martin (1999) the trade in knowledge became increasingly important during the third phase of globalization. This also encouraged information exchange which resulted in a shift from long-run capital flows to short-run capital flows.
same countries from 1990 till 2005, except for Sweden. It is thus important to keep in mind that these graphs are not exactly comparable. Nevertheless they both give a clear indication of the development of the openness of the countries.
Figure 6: Openness of Various Developed Countries, 1990-2005.
Export-to-GDP Ratio 0 0,1 0,2 0,3 0,4 0,5 0,6 1990 1995 2000 2005 Year Export-to-GDP UK France Germany Netherlands USA Japan
Source: data from OECD database.
The trend shown for the period 1990-2005 is quite different from what would be expected when extrapolating the upward sloping trend in figure 6. Indeed, one can observe a rising trend from 1995 up to 2000, for the Netherlands and Germany, but for the other countries included in the analysis the Export-to-GDP ratio remained fairly stable. It can be concluded that the openness of countries did not increased rapidly during the last decade, despite what might be exepected considering the innovations in communication techniques, transport techniques and the ongoing integration of countries, e.g. EU and the NAFTA.
2.3 The most recent of globalization
The most recent phase of globalization seems to differ from the previous phases in several aspects. In previous phases, joining globalization was mainly preserved for countries and large companies. In the most recent phase it also became possible for individuals to operate and compete globally and get connected around the world. This was mainly the result of the rise of internet. The development of internet already started in the 1960s but the impact of this innovation on people’s life became only visible when it started to take off in the second half of the 1990s, when it rapidly became available for a large public.
Table 2: Telephone Capacity & Internet Hosts, 1986-2000.
Thousands of Voice Paths Thousands Transatlantic Transpacific Internet Hosts 1986 100 41 5.1(a) 1991 504 141.2 617 1996 2' 021.6 1' 098.6 12' 881 2000* 2'048.3 1' 889.1 29' 670\a * Projection of minimum capacities
Source: Cairncross, 1997
Table 2 gives a clear indication of the extend of the increase in telephone capacity and internet hosts. A large increase in transatlantic and transpacific voice paths as well as internet hosts is observable between 1991 and 1996. The table supports the statement that internet started to take of from the second half of the 1990s. Communication costs fell as well. The cost of a three-minute phone call from New York to London was in 1930 around $244.65. In 1990 the same phone call costed around $3.32 and today it costs less than $0.50 (Loungani, Mody and Razin, 2002).
Table 3: Growth of World Internet Usage, 2000-2007. Africa 626% Asia 241% Europe 198% Middle East 490% North America 115% Latin America 391% Oceania/Caribbean 142% World 203%
Source: Internet World Stats, https://www.internetworldstats.com
Table 3 gives a clear picture of the rapid growth in internet usage around the world in the past years. It might seem unrealistic that Africa has the highest growth of internet usage and Europe and North America appears to have the lowest growth rate. However, the growth rate is calculated for the period 2000-2007 and the internet started already to take off in Western world in the second half of the 1990s. The internet usage in Africa and in Latin America however started from zero in this period. Nevertheless, this table shows clearly the fast rise of internet worldwide. Of course when the internet would only be used by consumers for watching films at youtube.com this would not necessarily have an impact on trade, economy or the importance of distance. However, the computer and communication networks seems to bring today’s world closer to the ideal world where there a huge amount of information available, there are no transaction costs and no entry barriers. Internet contributes to transparency of the economics transactions. Books and CDs sold by the internet are on average 10% cheaper than in the traditional bookstore (The Economist, 2000). The rise of internet usage en the ICT developments also influences the way business is done by companies. It simplifies the process of finding suitable business partners, such as subcontractors, and the makes of just-in-time management possible. According to the Dutch Centraal Economisch Plan it was already observed in 2000 that companies decreased there volumes of stock. According to Jules Theeuwes in his article, “Future of the labor market” (2001), the larger the company, is not necessarily the better. The quality and size of the network the companies is part of becomes increasingly important. The influence of internet and new ICT developments encourages competition, simplifies entering new markets and lead to the decreasing importance of traditional economies of scale (Theeuwes, 2001). This may indicate that due to the new developments with regard to ICT and the internet, that started in the last decade of the previous century distance decreased in importance in business decisions during this most recent phase of globalization.
technologies made it possible to trade highly differentiated manufactured commodities and even services internationally. Furthermore, due to the developments in transportation and communication techniques companies try to take advantage of cost differences and fragment their production process in different components produced at different locations around the world. One believes that this has resulted in a considerable increase of economic interdependence of individual countries.
An other development that differentiates this phase from earlier phases is the rise of the Asian economies. In the previous phases international trade was dominated by the Western World, while in the present phase the economic importance of Asian economies, such as China and India increased rapidly.
2.4 Conclusion
Although there are differences in opinions with regard to the question when globalization actually started. However, among economists there seems to be consensus. Although Williamson and O’Rourke (2002), Crafts and Venables (2001) and Baldwin and Martin (1999) examined different indicators, they all come to the conclusion that data gives evidence that globalization took off in the beginning of the 19th century. Friedman (2006), however, argues that globalization already started when Columbus discovered America and that this discovery led to a mentality change and an increase in international trade. It should be remarked that this statement is not further illustrated by data.
The most recent phase of globalization, starting around 1995, is characterized by a sharp rise of the internet. This enabled individuals to act globally and knowledge can now easily be shared around the world. However, there are some indications that openness did not increased sharply during this phase of globalization.
Chapter 3: Discussion on the influence of distance on trade
3.1 Introduction
The globalization process results in the integration of the economies of individual countries worldwide. Globalization is not a new phenomenon and it can not be denied that the process affects countries, companies, organizations and even individuals worldwide. However, it is unclear what the exact effects of globalization are. Globalization can be seen as a broad concept e.g. institutional, social and economic integration. In this thesis, the focus is on economic globalization and more specifically on international trade. Crafts and Venables (2001) defined globalization as the changing costs of economic interactions across distances. Technical innovations in the area of transportation and communication technologies led to a considerable decrease in trade costs. When trade decisions would depend solely on these costs, this would make trade at larger distances more profitable than it used to be. Several studies examined the relationship between distance and trade and came up with diverse results. In this chapter, the different studies regarding the influence of distance on trade flows and their results will be discussed. In the following part of this chapter first the arguments of those who that support the statement that distance does not matter anymore now or in the near future, followed by a discussion of the studies that do not support this statement. Finally, I some conclusions are drawn.
3.2 Distance does not matter anymore
Francess Cairncross (1997) and Thomas Friedman (2006) both wrote a book on the effects of globalization on the world. The titles, respectively, “The Death of Distance” and “The World is Flat” immediately give away the perspective of these two authors on the effects of the globalization process. The innovations in transportation and communication technologies are seen as the main drivers of globalization.
Figure 7: Transportation Bill (Freight Only) devided by GDP.
Source: Bureau of Transportation Statistics, Annual Reports.
The largest decrease was between 1960 and 1990. The downward trend clearly flattened from the 1990s. However, the declining trend observed in the graph combines several changes: 1.The change in the real costs of moving goods. 2. Overall ratio of the goods shipped to GDP. 3. People might have decided to change the mode of transport of goods, e.g. using trucks instead of the railway transport in order to be more flexible, this might damper the overall decline in transport cost (Gleaser and Kohlhase, 2004). This might indicate that in the most recent phase of globalization the decline of transport costs was not a great driver for the increase in international trade after all.
Figure 8: Trends in Average Applied Tariffs Rates in Developing and Industrialized Countries.
Trends in average applied tariffs rates in developing and industrialized countries 0 5 10 15 20 25 30 35 40 1970 1980 1990 2000 2010 Year u n w e ig h te d in % developing countries Non-OECD Countries OECD Countries
Source: Tariff data is primarily based on UNCTAD TRAINS database.
Figure 8 shows especially for the developing a significant decrease in applied tariffs rates. The applied tariff rates for developing countries peaked at 38% in 1983 and are today around 15%. For the Non-OECD and the group of OECD countries the decrease is not as large as is the case for the developing countries. For the non-OECD countries even a slight increase in applied tariff rates can be observed for the first half of the 1990s. Nevertheless, for all three groups of countries, the applied tariff rates are decreasing and seem to converge. Using the definition of globalization stated by O’Rourke and Williamson (2002), that globalization can be defined as convergence of international commodity markets and that this is driven by the fall in transport costs and the decrease in trade barriers, such as tariff rates. Figure 8 implies that the process of globalization and integration is ongoing and might explain the expectation of the decreasing importance of distance in trade decisions.
The new communication techniques, such as the possibilities of the internet, enabled people to communicate worldwide at very low costs and enabled companies to market their products worldwide. A decrease in transport costs enabled companies to transport their products over large distances at very low costs. Furthermore, these new techniques also made it possible for a wide range of intangible products, such as knowledge and services, to be easily spread and produced globally.
available as well as the availability of a certain level of infrastructure. Especially in developing countries, good and low-cost communications possibilities can free these areas from there isolated position. Furthermore, Cairncross predicts a convergence of income among countries around the world. As productivity rises in developing countries their wage level will follow. As an increasing number of companies become footloose they will locate their production where they can get the best trade-off for skills and wages. This will result in the redistribution of wages.
Convergence of income is also predicted by the neoclassical model. Assuming ‘death of distance’ and zero trade costs, countries will specialize in the industry in which they are factor abundant and as a result factor prices will equalize around the world leading to a convergence of income per capita.
According to Cairncross (1997), the irrelevancy of distance will result in a wide and fast dispersion of ideas, knowledge and products worldwide stimulating rapid innovation. As a result of the convergence of economies there will be more price transparency and low inflation, but in order to make this happen trade in goods and services should be fully liberalized as well as the communication market. Furthermore, globalization will encourage global peace, since countries are getting increasingly integrated and interdependent of each other.
Friedman (2006) emphasizes in his book ‘The World is Flat’, that due to the revolutions in communications and transportation, the old, established relationships between countries and continents have shifted. Products and even services can nowadays be produced almost anywhere in the world and more countries, e.g. China and India, are becoming serious competitors for the Western world, not only in manufacturing goods but also in areas that require more knowledge, such as software development. Friedman, however, does not state that automatically, as a result of the globalization process, income around the world will converge. Countries have to be aware of the process and have to be aware of their weaknesses and their strengths to be able to keep up in the process of globalization. Otherwise, countries such as the United States and the European countries might even fall behind, despite there position nowadays.
3.3 Distance still matters
importance of distance in bilateral trade, despite of the increasing integration of economies.
Table 4: Overview of the Results of Different Studies performed.
Study Year: Distance coefficient Year: Distance coefficient
Disdier & Head (2006) 1969: - 0.63 1996: - 0.95 Berthelon (2004) 1985-1989: -1,498 1995-2000: -1.606 Brun et al. (2005) 1962: -0.86 1996: -1.34
Frankel (1997) 1965: -0.48 1992: -0.77
Figure 9: Results from the Regression Equation by Brun et al.
Source: Brun et al, 2005
the sample in the rich Western world and the developing countries. The outcome shows that for the sample of developing countries the increasing importance of distance is still visible. However, this is not observed for the sample of developed countries. The Rich Western countries were able to benefit from the decline in importance of distance in international trade, while poor countries were not able to benefit from the most recent wave of globalization (Brun et.al. 2003). This seems contradicting to the statement of Francess Cairncross that globalization will lead to the convergence of economies. This conclusion seems to be in line with the predictions made following the line of reasoning of the New Trade Theory. The New Trade Theory predicts that there is a selection of a few countries, favoured by their location that will experience rapid transition (Venables and Crafts, 2001). This will not lead to a convergence of economies of different countries, but instead to an increased gap between different groups of countries (Crafts, 2003).
One of the consequences of globalization is that new technologies and innovations, especially in the area of communications, made it possible to outsource or even some mass-produce services, while services were always seen as internationally non-tradable. Can a change in the impact of distance be observed or is this trend exaggerated as well? At the moment only 7% of total services are traded internationally (Lejour, 2004). International trade in services is much more difficult than international trade in goods, since trade in services is often hampered by differences in regulations between countries, cultures etc. Nevertheless, distance seems to form less a barrier in trade in services than in trade in goods. Examining intra-EU trade, goods as well as services, for the period 1999-2001 a distance coefficient of - 0.922 for services and a distance coefficient of - 0.988 for goods was found. Certain categories of services, e.g. communication and financial services were able to benefit significantly from the new available technologies in communications. This enabled them to be traded internationally more easily. Nevertheless, considering the outcome of the gravity analysis, Le Jour and De Piava Verheijden (2004) conclude that distance is still of importance in trade decisions.
3.4 Possible explanations for the observed increasing importance of distance in international trade.
Simply the statement that distance still matters in international trade decisions and even became more important over time did not satisfy researchers. Therefore multiple explanations were sought to explain this distance puzzle.
First, consider the developments in with regard to trade costs. It is not hard to imagine that in the case where trade costs are very high, it will be unprofitable to trade with other countries/regions and economic activity will be scattered. Companies will stay focused on their home market. Transporting and selling their products in different markets or countries will simply not be a profit-generating activity. At the other extreme, in the case when trade costs are extremely low, companies do not care where to locate their production facilitates. Location and distance will not matter. At
Figure 10: Agglomerations in Europe, "The European Banana".
Source: Brakman, 2006
As is already mentioned above, it is realistic to assume that between countries labor is less mobile than within countries. Although, one might expect that international mobility is increasing, due to further economic integration and the changing perspective of people on distances, since new communication technologies and the rise of internet enables them to keep in touch at much more ease with people all around the world. However, this does not seem to be already today’s reality.
Figure 11: Agglomeration and Economic Intergration in the Abscence of Labor Mobility.
Source: Puga 1999
divided. At low levels of trade costs, firms will decide to locate their production location where the immobile factors are the cheapest. As a result they will spread their activities across regions. At intermediate levels of trade costs firms can still save on their trade costs by locating their production near other producers, suppliers or large markets and agglomerations will emerge. Due to the limited labour mobility the workers can only be attracted from the other domestic (e.g. the agricultural sector) sector by offering higher wages. As long as savings on transport costs outweighs the costs of high wages agglomeration will occur. When trade barriers between countries decrease significantly the savings on transport costs do no longer outweigh to cost of offering higher wages and production is likely to be transferred to non-industrialized countries, characterized with a low wage level, to capture the advantage of low factor prices. In this case, economic activity seems to become widespread again (Surico, 2003).
Céline Carrere and Maurice Schiff (2004) also concluded in their study that the importance of distance increased over time. They gave as an explanation that a change in the influence of distance is not the result of an overall decrease in trade costs, but that this is the result of a change in the different components of trade costs. Some trade costs are related to distance but this does not apply to all trade costs, e.g. port labour costs, storage costs. The transportation costs unrelated to distance are called dwell costs. (Carrere and Schiff, 2004). A decrease in both kinds of costs will result in an increase of international trade but will have opposite effects on the influence of distance. For example, a reduction in dwell costs will result in a proportional larger decrease in transport costs for countries at small distances and will give an incentive to trade with countries situated nearby, while a decrease in trade costs related to distance, e.g. gasoline prices, will encourage trade at large distances. Looking at it this way, it is not that important if total trade costs change, but which components of the total trade costs change, this change will influence the importance of distance in international trade flows (Carrere and Schiff, 2004). This adds to the results of the study of Berthelon and Freund (2004), who also examined the different components of trade costs and came to the conclusion that due to change in these different components of trade costs 25 industries became more sensitive to distance. This applies mainly to homogeneous goods. Differentiated goods on average did benefit more from the fall in communication and transportation costs, which can be explained by the higher importance of search costs for these kinds of goods (Berthelon and Freund, 2004).
for example, still seems to be an important element to be able to establish trust and understanding and to ensure a successful transaction. This might hamper trade on large distances. Furthermore, as a result of falling transportation and communication costs, the mobility of products did increase, but this does not necessarily have to result in an actual movement of the goods or even production facilities. The ‘old’ situation might still be the most preferred one. (Leamer, 2006)
A change in the influence of distance on trade over time nevertheless can have important consequences for economic growth and welfare. A change in technologies, for example in communication or transportation, can influence the attractiveness of a country at large distance compared to countries located at smaller distances. This can have important consequences for the country involved. When distance becomes more important in international trade, countries can decide to trade with partners closer to home. Trading partners at larger distances can be confronted with a restriction on their access to goods or technology and on their trading opportunities. (Carrere, 2004)
3.5 Modelling the influence of distance in international trade
Most empirical studies examining distance as explanatory factor in international trade patterns, as discussed in the above section used the gravity model. The reason that the gravity model is used so often is that it shows a good fit and appears to be stable over time. The gravity model was first applied for economic purposes by Jan Tinbergen in 1962 and shows the following form: Fij=G*(Mi*Mj)/Dij, where F is the trade flow, G
is a constant, M is the economic mass of respectively country i and j, and D is the distance between those two countries. The model is often converted to a linear model to be able to use the ordinary least square method. This resuls in the regression equation, ln(bilateral trade flow)=α+βln(Yi)+βln(Yj)-βln(Dij), where Yi is the GDP of
country exporting country, Yj is the GDP of the importing country and Dij is the
distance between the two countries. This equation holds, in general, for most countries despite differences in economic levels, political structure, geographical location (Van Marrewijk, 2002). However, the use of the gravity model was criticised, because these studies gave no theoretical justification of this model, only an intuitive justification. Nevertheless, several successful attempts were made to provide formal theoretical foundations for the use of the gravity model.
models is suspect (Deardorff, 1995). In my opinion the gravity model is a fairly simple way to estimate reasonably accurate the bilateral trade flows. For example, the R2 of the estimated regression models in the studies discussed in chapter 3 are reasonable high. The model is sufficient to examine trends in bilateral trade, such as the importance of distance in bilateral trade. However, of course also the gravity model has its drawbacks of which one should be aware. The simple model that is stated here assumes that there is free trade, which means that there are no trade barriers or transport costs and thus countries have identical prices (Feenstra, 2006). Of course, in reality trade barriers do exist, resulting in different prices for the same goods in different countries. Despite the fact that the gravity model usually performs really well, one should be aware of the fact that its strong assumptions.
The gravity model is not the only model used to model the influence of distance on trade. Venables and Limao (1999) developed a Heckscher-Ohlin-von Thunen model of international specialization. They agree that the gravity model is indeed able to explain international trade flows, but according to these authors the gravity model fails state anything about its implications. The goal of their paper is to make examine the effects of the costs of distance on different economies and to examine the effect of the globalization. In the Heckscher-Ohlin-von Thunen model the assumption of the existence of a central location and several more remote locations of the location theory of Von Thunen is used. Just as in the Hechscher-Ohlin model it is assumed that that locations have fixed endowments of several types of factors of production and goods exhibit different factor intensities and labor is internationally immobile (Venables and Limao, 1999). Furthermore, this model exhibits a more general production structure, than is generally used in the HO-model, which means that products in the centre can also be produced elsewhere, and intermediate goods to be able to study final assembly activities do exist. There are some important limitations with this model; it assumes constant returns to scale, perfect competition and no market imperfections. As a result the world is divided in different zones, all characterized with differences in inter- and intra-zone economic structures, depending on their location and the endowments. That results in the outcome that zones located at larger distances from the centre receive lower prices for their exports to the centre, and pay higher prices for their imports to the centre.
Venkat and Wakeland (2006) examined the effect of trade over distances by using an agent-based model with transaction costs. The reason behind this method is that international trade is rapidly expanding but in the end distance might still limit international trade as a result of transaction cost related to distance. These transaction costs might increase in the future due to increase of the price of oil as a result of the depletion of the oil reserves or the growing concerns concerning greenhouse gas emissions. The model is used for two scenarios, one where there are no regional differences (Globally Mixed Random) and one where the world is divided in a western half and an eastern half and the zones exhibit differences in endowments of the two goods in the model (Local Comparative Advantage Random). The world exists of traders trading two goods, all attempting to maximize their utility using the Cobb-Douglass utility-function. Traders are rational, non-strategic and myopic and, as in reality, repeated actions should result in trade patterns. The transaction costs are calculated as follows: distance*quantity of goods bought* unit distance costs (Venkat and Wakeland, 2006). For this thesis we are especially interested in the effect of increasing distance costs on average trade distance.
Figure 12: Trade Distance versus Transaction Costs.
Ave-LCAR = Average Local Comparative Advantage Random
StdDev-LCAR = Standard Deviation Local Compartive Advantage Random Ave-GMR = Average Globally Mixed Random
StdDev-LCAR = Standard Deviation Globally Mixed Random Source: Venkat and Wakeland, 2006
is divided in two zones, with different endowments, the trade distance does not fall as fast as in the first scenario (GMR), because traders have to search further to find trading partners, due to the different endowments of the goods. However, when unit trade costs rise further, the trade distance falls beneath the level as in the case of no regional differences, because trading opportunities decrease significant. In fact, this is a highly simplified and artificial model that can not be tested against real-world data. Nevertheless, the model does show that distance-based trade costs can have a significant impact on trade (Venkat and Wakeland, 2006).
The problem with the Hecksher-Ohlin-von Thunen model as well as the agent-based model with transaction costs of Venkat and Wakeland and the gravity model is that the trade patterns estimated using the models do not resemble the real world trade patterns. This is a result of the fact that in order to model, in this case world trade patterns, simplifications have to be made. The gravilty model is relatively simple does a remarkable good job in explaining trade patterns.The R-squares usually obtained by studies using the gravity model are reasonable high. Since more attention is given to the theoretical justification of this model, it is used more often in studies investigating international trade patterns. As can be seen in the discussion on the influence of distance on international trade in chapter two most studies performed to estimate the influence of distance on trade patterns are using the gravity model.. For these reasons, the gravity model is chosen to be base for the empirical analysis in the following chapters of this thesis.
3.6 Conclusion
face-to-face contact might be very important as well, but hard to establish at large distances.
The gravity studies performed on this subject mostly examined the changing influence of distance on trade for a whole group of countries. Brun already showed in his paper “Has Distance Died? Evidence of a Panel Gravity Model”, that different results are obtained when examining developing and developed countries separately. The globalization process is often compared to a tsunami that overflows the world. Countries are supposed to drag into it whether they like it or not. However, countries differ greatly in their characteristics, is it not likely that they react differently or are influenced differently by the globalization process? Looking at distance, could it be possible that the change of the influence of distance in international trade differs among countries. After all, Brun already showed us that difference exists between developed and developing countries. In the remainder of this thesis, the influence of distance of individual countries is examined to see if there exists great differences among countries and if one can observe similarities in the results obtained.
In this chapter also a few models, using distance as explanatory variable, are discussed. Most studies performed on distance and trade used the gravity model. Although there is some criticism about the theoretical justifications of the model, it still seems to be appropriate model to use as a starting point for the empirical part of this thesis.
Chapter 4: At First Glance, Comparing trade patterns
4.1 Introduction
The results of the studies using the gravity model give a general result for all the countries involved in the analysis. Globalization is a process affecting all countries around the world, but does it affect countries in similar ways? Do countries that show similar patterns also have similar characteristics? One could imagine that, for example EU member states, show similar changes. The goal of this chapter is to get an impression of the influence of distance on bilateral trade patterns for individual countries during the most recent phase of globalization, which started during the last decade of the previous century. This is done by drawing a graph of the bilateral trade pattern with respect to distance for two different years, 1995 and 2004, for individual countries. 1995, the year around which the most recent globalization phase started to take off, and 2004, the year for which the most recent data are available, are selected for further investigation.
4.2 Construction of the graphs
The reason behind drawing the graphs is that graphs can provide a quick visual representation of the trends happening in the relationship between distance and the trade ratio (Exports/GDP) during 2004 and 1995. The graphs are drawn for every country separately, which makes it possible to classify the countries according to the trends observed in their respective graphs and to quickly examine if, and which similarities or differences can be observed among these countries. Furthermore the results obtained from the analysis of the graphs form a starting point from which regression analysis can be performed.
The data for the distance is obtained from the database of the Centre de D’Etudes Prospectives et D’Information Internationales (CEPII). The distances in this database are estimated using the great circle formula. This formula uses the latitudes and longitudes of the capitals to calculate the distance between them.
Great circle formula: acos ((lat1) * cos (long1) * cos (lat2) * cos (long2) + cos (lat1) * sin (long1) * cos (lat2)* sin (long2) + sin (lat1) * sin (lat2)) * Earth radius
Where:
1= exporting country 2= importing country
lat1= latitude of the capital of the exporting country lat2= latitude of the capital of the importing country long1= longitude of the capital of the exporting country long2= longitude of the capital of the importing country
This is a simplification which allows us to neglect the existence of natural trade barriers. (Van Bergeijk, 1990) This is a common method, however, one has to be aware of its limitations. The economic centre of a country does not always correspond to the capital of a country. It is very difficult to accept, especially for large countries, such as the US, that there is only one economic centre. Unfortunately, the data for exports can not be separated towards these different economic centres. Nevertheless, the measure for distance of the CEPII database does also incorporate a measure for internal distance. This internal distance, dii =2/3√(area/π), is based on the area of a
country and represents a measure for the average distance between the consumers and producers within a country to partly overcome the disadvantage of point-to-point measuring (Head, 2002; Mayer, 2006).
4.3 Data availability
during the most recent phase of globalization. The conclusions that can be drawn from these graphs can serve as a base for further examination in a model. In the end 29 exporting countries are selected, all OECD countries and China (Except for Luxembourg). For these countries the exports to 45 countries is examined. Unfortunately, the countries from the African continent are somewhat underrepresented. Nevertheless, the 45 countries involved in the analysis represent approximately 75% of world exports. An overview of the specific countries included in the analysis is given in the appendix. Luxembourg forms an exception as OECD country. Luxembourg, as exporting country, is excluded from the analysis, since Luxembourg reported up to 1999 only the total of exports of Belgium and Luxembourg. Using this data would result in double counting. Subtracting the exports reported by Belgium is not an option as well, since one would miss the exports of Luxembourg to Belgium.
Table 5: Description and Measurement Units of Variables.
Variable Description Unit
Eij Value of exports from country i to country j
thousands of US dollars
Yi GDP of country i At current prices in US dollars Dij Distance between capitals of
country i and j, including a measure for internal distance
In kilometres
4.4 Different categories of countries
The results when examining the changes of the trade patterns of the different countries can be assigned to three different categories. In the next paragraph the implications of each category will be discussed separately. These categories are:
1. Countries for which trade at larger distances has become relatively more important. 2. Countries for which trade at larger distances has become relatively less important. 3. Countries for which there are no clear changes visible in their trade patterns.
1. Trade at larger distances has become relatively more important to a country.
Figure 13: Bilateral Trade at Larger Distances has become relatively more important.
Ireland -5,00% 0,00% 5,00% 10,00% 15,00% 20,00% 0 5 10 15 20 Distance in 1000 kilometres Exports/GDP Exports/GDP 1995 Export/GDP 2004 Logaritmisch (Exports/GDP 1995) Logaritmisch (Export/GDP 2004)
The Export-to-GDP ratios are shown on the Y-axis and the distance on the X-axis. Since the distance between the capitals of the countries did not change there are always two observations shown above each other. There are two possible outcomes, the observation of the Export-to-GD{ ratio of 1995 lies above the observation of 2004 or it lies beneath the observation of 2004. In figure 13 a line is drawn through the bilateral trade ratios for 1995 as well as 2004 for Ireland. The curve decreases not as fast in 2004 as in 1995. The Export-to-GDP ratios decreased for small distances and increased for large distances. For Ireland and the countries showing comparable graphs it can be concluded that trade with countries located at large distances has become relatively more important compared to trade with countries located nearby.
2. Trade at smaller distances has become relatively more important to a country.
Figure 14: Bilateral Trade at Larger Distances has become relatively less important.
In figure 14 the curve drawn for the year 2004 is steeper and decreases faster than the curve drawn for the observations in 1995. For the countries showing comparable graphs it can be concluded that bilateral trade at small distances has become relatively more important than trade at large distances, since a larger proportion of there total bilateral trade occurs at smaller distances. Furthermore for Germany it can also be concluded that the openness of the country on average has increased, since the curve of 2004 lies entirely above the curve drawn for 1995.
3. There are no clear differences observed in the trade pattern between the two years.
Figure 15: The Bilateral Trade Pattern did not change with respect to Distance.
China -2,00% 0,00% 2,00% 4,00% 6,00% 8,00% 0 5 10 15 20 25 Distance in 1000 kilometres Exports/GDP Exports/GDP 1995 Export/GDP 2004 Logaritmisch (Exports/GDP 1995) Logaritmisch (Export/GDP 2004)
4.5 Results
The results of the graphs made for the countries can be divided according to the three categories mentioned above. The graph for each specific country shown in the appendix as well as all an overview of the countries included in this analysis.
Table 6: The Classification of Counties with respect to changes in their Trade Pattern.
Countries for which bilateral trade flows at large distance became relatively less important.
Countries for which no change can be observed with regard to the geographical dispersal of bilateral trade flows.
Countries for which bilateral trade flows at large distance became relatively more important.
Austria Canada Australia
Belgium China Greece
Germany Denmark Ireland Hong Kong China Finland Italy
Hungary France Switzerland
Japan Iceland United Kingdom
Korea Spain United States
Mexico Sweden
Netherlands New Zealand
Norway Poland Portugal Turkey
Table 6 shows that results based on the graphs drawn for the different countries are diverse. All three categories are well represented in the table above. Most countries are assigned to the category that states that the influence of distance increased comparing 1995 with 2004. This seems to add to the conclusion of the studies using the gravity models that distance still matters. However, one has to keep in mind that this could also be a result of the selection of countries included in the analysis.
Despite the expectation that the EU-member states might show similar changes in their trade pattern, simply due to the fact that they are members of the European Union does not seem to hold. In table 1 shows that the different European countries are assigned to all three categories.
Up till now, the discussion of the results focused on the rich Western world. However, Asia made an important catch-up and became an important player on the world market. How did the most recent phase of globalization influence the trade flows in these upcoming Asian countries according to the graphs? Surprisingly, for China hardly any difference can be observed in their bilateral trade pattern comparing the trend lines of 1995 and 2004. Of course the trade volumes did increase in the last decade but the relative importance of countries located at different distances did hardly change. For Japan trade at short distances became relatively more important in the last decade. This is not very surprising, since Japan already was a developed modern country situated close to the new upcoming markets as China and India.
4.6 Conclusion
Chapter 5: Results regression analysis of trade patterns
5.1 Introduction:
The graphs drawn in the previous chapter are very useful to get an impression of the changes in trade patterns with respect to distance. However, to be able to draw some more specific conclusions about the possible changes in trade patterns for the different countries included in the analysis e.g. about the strength of the relationship, the graphs are not sufficient. To overcome this limitation in this chapter a regression analysis is performed to model the relationship between the Export-to-GDP ratio and distance and to determine the magnitude of these relationship. In this chapter, first the estimated regression equation will be explained. The regression analysis will be performed for each country individually.
5.2 Construction of the model
In this chapter the relationship between distance, explanatory variables and the trade ratio (Exports/GDP), the dependent variable, is estimated.
The equation used to estimate this relationship will be as follows: Eij/Yi = c Dij α Yjβ α0>0, α1<0
Where:
Eij = the exports of country i to country j.
Yi = The GDP of country i
Dij = the distance of the capital of country i to country j, with a measure of internal
distance included.
Yj = The GDP of country j
The model seems to be nonlinear and in this case it would be difficult to use the least squares method to estimate the equation. However, although the model is nonlinear in its variables it appears that by taking logarithms the model is linear in its un-known parameters. (Carter Hill, 2001)
Ln(Eij/Yi )= c+ αLn(Dij) + β Ln(Yj)+ e α>0, β<0
Where ln(Eij/Yi) is the log of the dependent variable, Export/GDP ratio, and α ln(Dij)
distance included. Ln(Yj) is the GDP of the imported country. Furthermore, c is the
intercept and e = random error term, which is assumed to be normally distributed; e~N (0,σ2). (Carter Hill, 2001)
Examining the model above, it can be concluded that regression equation stated is not the traditional gravity model. Instead, the Export-to-GDP ratio is estimated. Based on the available data, this decisions was made because inflation would complicate the comparision of the results at different moments in time.
However, one should keep in mind that this specification has some drawbacks. Robbert Feenstra (2004), gives critique in his book “Advanced International Trade”. First of all, the gravity equation assumes that countries are completely specialized in their outputs and tastes are identical and homothetic and there is free trade world wide. Prices will be identical as well. Robbert Feenstra (2004) mentioned in his book an other disadvantage of the gravity equation stated above, the exports of the exporting country are divided by its GDP at the left side of the equation. Estimating the dependent variable in this way constrains it at unity. A drawback is that there might be some fixed effects. In the estimated regression equation stated above there is only one measure of cross-border trade, namely Eij. This results from the assumption
that transport costs are symmetric. According to Feenstra (2004), it might be appropriate to included fixed effects since the fixed-effect method takes into account average effects across countries. Furthermore, it is assumed that a good produced in a country is sent to all countries in porportion to the purchasing country’s GDP. This implicitly assumes that the country shares are approximately constant over time (Feenstra 2004). This however does not need to be the case in reality and it is something to keep in mind, when comparing the results of the estimated regression equations of 1995 with 2004. Unfortunately, since a model is always a simplification of the real world, every model will has its drawbacks and limitations.
regression equations 1/ (Eij/Yi) = α Dij + β Yj + c + e and 1/ (Eij/Yi) = α Dij + β log
(Yj) + c + e were also estimated. The goodness-of-fit for these two models appeared
to be a lot lower than for Ln(Eij/Yi ) = c + αLn(Dij) + β Ln(Yj)+ e, the regression
equation used now. The R2 was considerably lower for the other equations. Therefore, it was decided to use Ln(Eij/Yi ) = c+ αLn(Dij) + β Ln(Yj)+ e for further research.
Table 7: Description and Measurement Units of the Variables.
Variable Description Unit
Eij Exports from country i to country j
In thousands of US dollars
Yi GDP of country i At current prices in US dollars Dij Distance between capitals of
country i and j, including a measure for internal distance
In kilometres
Yj GDP of country j At current prices in US dollars
5.3 Data sources and availability
5.4 Results
The regression analysis is performed for each individual country separately to obtain an impression of the differences between the countries. The results of this regression analysis for the coefficients α, β, the constant and the corresponding R2 andadjusted R2 for 1995 and 2004 are respectively are shown in table 8 and 9.
Table 8: The Results of the Estimated Regression Equation for Individual Countries for 1995.
