The effect of outward FDI on domestic exports:
An empirical analysis for the Netherlands
Masterthesis
Author:
Marion (A.M.E) Kosse
Student number:
1460196
Date:
July 2010
1
st. Supervisor:
Prof. Dr. S. Brakman
2
nd. Supervisor:
Drs. H. J. Bezemer
University of Groningen
Faculty of Economics and Business,
Abstract
Trade and Foreign Direct Investment (FDI) have become more and more important for the Netherlands during the past few decades. But this brings also the fear that outward FDI has a
Content Content 3 Chapter 1: Introduction 5 1.1 Introduction 5 1.2 Problem Statement 6 1.3 Research question 7
1.4 Structure of the paper 8
Chapter 2: Literature review 9
2.1 MNE and horizontal/vertical FDI 9
2.2 Substitutability and Complementarity 11
2.2.1 Substitutability_theory / horizontal FDI 11
2.2.2 Substitutability_empirical results other researchers 12
2.2.3 Complementarity_theory / vertical FDI 13
2.2.4 Complementarity_empirical results other researchers 15
2.3 Gravity model 16
2.3.1 Gravity model_theoretical foundation 17
2.4 FDI versus trade models 20
Chapter 3: Hypotheses 23
Chapter 4: Data description 24
4.1 Dependent variable 24 4.2 Independent variables 25 4.3 Dummy variable 30 4.4 Interaction variable 30 Chapter 5: Methodology 32 5.1 Gravity models 32
5.2 Panel versus cross-section analysis 33
5.3 Fixed Effects Model versus Random Effects model 34
5.4 Hausman specification test 35
5.5 Breusch-Pagan Lagrange multiplier (LM) test 35
5.6 Correlation and multicollinearity 36
5.7 Heteroskedasticity 36
Chapter 6: Results 38
Chapter 7: Discussion 42
Chapter 8: The nature of the Dutch outward FDI 46
Chapter 9: Sensitivity analysis 48
9.1 Elasticity 48
9.2 Adding a constant 49
Chapter 10: Conclusions 50
10.1 Limitations and further research 51
Chapter 11: References and appendices 54
Appendix I: Dutch main trading partners 66
Appendix II: Commodity groups 67
Appendix III: Summary statistics 69
Appendix IV: Random Effects Model regression (without lagged FDI) 70
Appendix V: Correlation and multicollinearity 71
Appendix VI: The Hausman test 72
Appendix VII: Heteroskedasticity test 73
Appendix VIII: Fixed Effects Model regression 74
Appendix IX: The Breusch-Pagan lagrange multiplier test 75
Appendix X: Autocorrelation test 76
Appendix XI: Random Effects Model regression 77
Appendix XII: Random Effects Model regression with interaction variable 78
1. Introduction
1.1 Introduction
Since the days of the Dutch East India company, the Dutch economy has been heavily dependent on its firms´ overseas activities. Even after the ending of the colonial era, the Netherlands has continued to be a significant player in the world market (Hoesel & Narula, 1999). Indeed, some of the world´s largest multinationals are located in the Netherlands. In fact, there are 12 Dutch multinationals listed in the `Global 500 ´ of 2009, which is a rankinglist of the world´s largest corporations. Royal Dutch Shell and ING group, for example, are respectively the first and eigth in ranking in this list
(CNNmoney, 2009).
Businesses overseas are not only restricted to trading activities, but also include strategic alliances and networks as well as Foreign Direct Investment (FDI) by its multinational corporations (MNCs).
Based on figures of Eurostat, Statistics Netherlands shows that in 2008 the Netherlands was, after Germany, the second largest export country of the European Union. The Netherlands also hold a prominent position with respect to export value growth and trade surplus (Statistics Netherlands, 2009). This is not something new, export has always been important for the economy of the
Netherlands. This is also due to the fact that the Netherlands is a small open economy and according to Földvári (2006) there are two main factors why smaller economies tend to trade relatively more than large ones. First, smaller countries usually have a less diversified economic structure and therefore depend more on traded goods and resources. Secondly, the internal market is too small to reach economies of scale and external markets are needed for the domestic production to be able to decrease average costs.
Not only trade, but also Foreign Direct Investment (FDI) is very important in the Netherlands. In absolute as well as in relative terms, the Netherlands belong to the largest international direct investors in the world. According to the publications by The Central Intelligence Agency, the Netherlands is sixth in the list of world countries according to the stock of FDI abroad (The World Factbook, 2008). Moreover, the UNCTAD´s new ´Outward FDI Performance Index´ between 2005 and 2007, which measures the share of a country´s outward FDI in world FDI as a ratio of its share in world GDP, shows that the Netherlands is with a score of 3.697 on the 7th place in the list of world countries. This indicates that outward FDI is relatively very important for the Dutch economy compared to 124 other countries in the world that are on this list (UNCTAD, 2005-2007). This has to do with several
1.2 Problem statement
It is generally accepted that multinational activity enables firms to produce and operate more efficiently as a whole and increase market shares (Svensson, 1996). Even though transactions costs related to multinational operations are substantial, MNCs have a potential to be more profitable than pure domestic firms (Kokko, 2006). However, it is not obvious that these benefits also hold for the domestic economy as a whole. While higher profits and increased R&D are positive contributions to the home country, the increased market power may result in higher consumer prices and welfare losses. Moreover, the MNCs are more or less free in choosing the allocation of various gains, since the MNC has production facilities in different countries (Kokko, 2006).
These distinctions between the MNC and its domestic country, makes it very interesting to investigate the effects of FDI on the home economies. Especially the relationship between a firm´s foreign direct investment and its exports from its home country, or the relationship between total FDI and total exports from the home country, has been the subject of intensive public debate in the industrialized world and a considerable amount of empirical research. This discussion has got a resurgence of interest during the past decade, because of amongst other things, the global liberalization of trade and investment after the completion of the Uruguay round, the Chinese membership in the WTO and the regional integration processes in Europe, America, and the Asia-Pacific region (Kokko, 2006). Moreover the change in unemployment levels in Europe and the increase in outward FDI by European firms, have made the possible connection between the two also very interesting to investigate (Lipsey, 2002).
An important question in this debate is the extent to which FDI complements or substitutes for exports by the parent company or by other firms in the home country. This question is of importance since people fear that direct investment abroad replaces home country production and exports to foreign markets and, as a consequence, cause unemployment at home (Lipsey, 2000). This also influences the proposals for government action to hinder the growth of FDI, because if there is a chance that FDI substitutes for exports, it is very important to investigate to what extent. This is because trade (and therefore exports) are very important for an economy to grow, that has been a well-known fact for a very long time now.
countries to specialize and due to this specialization the countries are better in allocating the resources and in purchasing cheaper resources from other countries. Trade also increases the firm‟s market and its total domestic output and they therefore have lower average costs and are more able to increase their efficiency (Markusen, 1981). Moreover, the increased competition will reduce the inefficiency of a monopoly and force firms to stay competitive by being more innovative on production methods, technology, marketing and so on. Due to the increased competition weak firms will exit and strong firms will expand (Anderson, 2008).
Not only the producers, but also the consumers benefit from trade. Due to this increased competition, they are now able to make use of a greater variety of goods and services, better quality and lower prices (Anderson, 2008).
Even though trade seems to be good for the economy in general, the division of the gains can be uneven and there can be losers (Markusen, 1981).There are also more disadvantages of trade like the short-run structural unemployment that may arise when a country is specialized in one industry and prices in that industry will fall. Also the transport costs involved in international trade, will increase the production costs and therefore will decrease the efficiency.It is also hard for infant industries with higher production costs in the beginning, to compete with the already (more efficient) existing
industries (Markusen, 1981).
Although there are some disadvantages to trade, most of these disadvantages only apply in the short-run, since labour for example, can be trained in the long-run. It is therefore well-accepted that exports are beneficial for a country and it is of great importance to measure the effect that FDI has on domestic exports. To the best of my knowledge, this has not been done in the literature for the Netherlands, at least not for recent years.
1.3 Research question
The main objective of this study is to investigate, with the help of the gravity model, the effect of Dutch outward FDI on Dutch bilateral exports over the period 1996 till 2008. That is the effect of Dutch outward FDI in a foreign country on the Dutch exports to that particular country.
The general research question within this thesis can therefore be formulated as:
“What is the effect of Dutch outward Foreign Direct Investment on Dutch bilateral exports in the years 1996 till 2008?”
international perspective of the Netherlands can also been seen from the fact that the Netherlands has an open culture with a multi-ethinic and multicultural society. All these characteristics of the
Netherlands, make the Netherlands a very interesting country to use in my research when investigating the effect of home outward FDI on the home country‟s export.
Despite the long and rich tradition of the Netherlands in foreign direct investment and the prominent role of Dutch MNEs in the world economy, to my knowledge relatively little academic research has been done so far to study the relationship between Dutch outward FDI and Dutch exports. Most detailed studies of this relationship have focused primarily on Sweden and the United States, because those countries have published the most data on the outward investment by national corporations and therefore facilitated analysis at the macro level. This research will therefore be of great value to the already existing research on this topic.
Another reason why most researchers on this topic have focused on Sweden and the US, is because MNEs occupy dominant positions in both these economies, accounting for most of manufacturing employment, exports and R&D (Kokko, 2006). This also holds for the MNEs in the Netherlands and that is another reason why I have chosen the Netherlands for my research.
We will see at the end of this research that Dutch outward FDI had a negative but very small effect on the Dutch bilateral exports in the years 1996 till 2008. This negative effect may indicate the possible horizontal nature of the Dutch FDI. Authors who have tried to separate FDI into horizontal and vertical FDI found that the majority of world FDI is horizontal in nature and between the high-income developed countries. Moreover, they found evidence for horizontal FDI to substitute for domestic exports and vertical FDI to be a complementary to domestic exports.
1.4 Structure of the paper
This paper is organized as follows. After the introduction, chapter 2 discusses the existing literature supplementary to and useful for my research. It is about complementarity and substitutability between FDI and exports found in other papers, about the gravity model and about other theories related to this topic. The hypotheses are presented in chapter 3 and the data and methodology used in the study to test Hyothesis I will be found in chapter 4 and 5 respectively. These chapters are followed by chapter 6 and 7 where one finds the results and the discussion of these results. Chapter 8 describes the analysis and results for the investigation of Hypothesis II and chapter 9 contains the sensitivity analysis I have done in my research. This sensitivity analysis is followed by chapter 10 which contains the
2. Literature Review
2.1 MNE and horizontal/vertical FDI
Multinantional Corporations (MNC‟s) are firms that engage in foreign direct investment (FDI), which is investment where the firm gets a substantial controlling interest in a foreign firm or sets up a subsidiary in a foreign country (Markusen, 2002). Reasons for multinationals to invest abroad is for example to avoid trade barriers, to make use of cheap labor, to obtain information more easily and at lower costs, the ease of bringing products in line with the local demands and protecting their market shares when there are unfavorable developments in their home countries (Lipsey, 2002). Moreover, Markusen (1995) suggest that firm-specific assets may also play a role in the decision to become a foreign investor. The higher these firm-specific assets, the more likely it is for a firm to locate production abroad rather than exporting.
One can distinguish between horizontal and vertical foreign direct investment (Markusen & Maskus, 2002). According to Kokko (2006), it is relevant to make this distinction when exploring the home country effects of FDI. Vertical firms are multinationals that geographically split production by stages, mostly based on factor intensities. So, skilled-labor-intensive activities will be located in skilled-labor-abundant countries for example, in order to find low-cost locations for different parts of the production process. Helpman (1984) and Helpman & Krugman (1985) for example, have used vertical firms in their general-equilibrium model.
Horizontal firms on the other hand, are multinationals that duplicate a subset of its activities by setting up a foreign plant in addition to a home plant for some part of the production process (Navaretti & Venables, 2004). Examples of researchers who have used horizontal MNEs in their models are Markusen (1984) and Markusen & Venables (1998, 2000).
The main motive for horizontal foreign direct investment (HFDI) is access to foreign markets in order to avoid trade costs. Other motives are: better to be able to shape the final product to local tastes, to respond better to changes in local market conditions, to better shape the multinational‟s interaction with other competitors in the market and so on. HFDI will therefore tend to be flow to locations with good market access, where sales will be large enough to cover the fixed costs of the foreign plant. The main motive for VFDI is to reduce production costs by finding low-cost locations for parts of the production process (Navaretti & Venables, 2004).
The distinction between plant-level and firm-level economies of scale is important in the horizontal and vertical models. When a firm has large firm-level economies of scale, it means that the firm is large and therefore might have many sales in different countries. However, large plant-level
scale and small plant-level economies of scale (Navaretti & Venables, 2004). In case of HFDI, these plant-level economies of scale will be foregone as production is distributed across several plants. The horizontal firm has to make a trade-off between the increased sales and other benefits of access to foreign markets and these foregone economies of scale. In case of VFDI, the firm has to make the trade-off between having the benefits of producing in countries with low factor costs and economies of integration foregone (Navaretti & Venables, 2004).
The impact of trade costs will be different for horizontal and vertical FDI. In many studies, trade costs are assumed to consist of different components like transport costs, distance and trade policy barriers. They are particularly important for VFDI, since products at different stages of the production process may cross borders many times. Therefore trade costs are assumed to discourage VFDI, since they increase the costs of the trading components between different production units. However, in case of HFDI, trade costs seem to encourage investment. The higher the trade costs, the more financially attractive it becomes to invest abroad instead of exporting to that particular market.
One should expect HFDI therefore to decrease in case of integration when trade costs are low. However, we have seen for example with the creation of the European Union, foreign direct
investment (primarily horizontal) to increase. An explanation for this empirical problem is the fact that the trade barrier motive (high trade barriers attract HFDI) and the market size motive (large foreign market size attract HFDI) interact with each other in affecting a firm‟s investment decision. The single European market reduces the cost of trading in the region but also expands the internal market, because it turns the separate small regions into a single large economy which makes it attractive for foreign firms to invest in. The latter is called the „export platform‟ motive, since investing in one of the small regions can be profitable in case of integration, because this region can now serve as an export platform to other parts of the integrated economy (Navaretti & Venables, 2004).
VFDI is expected to take place primarily between countries with different factor endowments and factor costs, while the „convergence hypothesis‟ (Markusen and Venables, 1996) suggests that HFDI is expected to flow between countries that are similar in either size or comparative costs.
Moreover, Markusen (2002) has shown that most of the world FDI flows between the high-income developed countries.
However, Hanson (2001) states in his paper that even though HFDI seems to be more prevalent than VFDI, the importance of VFDI has been increased during the last years.
2.2 Substitutability and Complementarity.
There is no consensus on the home effects of outward FDI. People are proud when their MNEs do well in Fortunes‟ rankinglist of the largest firms in the world, but they are afraid that this
internationalization leads to a reduction of home economic activity. Because due to outward FDI, production and employment that would have taken place in the home country, now take place abroad (Navaretti & Venables, 2004).
Whether outward FDI leads to an increase or decrease in domestic exports, depends on whether home and foreign activities are complements or substitutes to each other.
2.2.1 Substitutability_theory/horizontal FDI
Standard theory of the MNC assumes substitution. In this theory the decision as to whether to produce abroad is a matter of choosing among possible methods of serving the foreign market, like exporting, licensing or producing abroad. To make this decision, the MNC has to make a trade-off between the transportation costs and possible tariff costs that arise when exporting to a foreign market and the high fixed costs of producing in the foreign market. This means that a firm chooses to export for lower sales levels and will produce abroad in case of higher sales (Blonigen, 2001). One can expect that in these traditional theoretical models, where the firm‟s market share in the host country is assumed to be given (Svensson, 1996), FDI is a substitute for export to that foreign market, because FDI and
exporting are seen as alternative ways to penetrate the foreign market (Blomström et al., 1987; Head & Ries, 2001).
Foreign direct investment might not only be at the expense of the MNE exports, but also at the expense of the exports of the rival companies in the home country (Blomström et al., 1987).
Foreign direct investment can also have a negative influence on domestic exports when knowledge leakages occur. Some of the information about the MNCs´ proprietary technology and knowledge may spill over to local firms in the foreign countries (Blomström & Kokko, 1998). And once the foreign firms have learned to use these foreign technologies, they are able to capture some market shares from the MNC affiliates with possible decreasing domestic exports as a result (Kokko, 2006). However, it could also be the other way around, since the presence of a MNE in a foreign country is likely to generate various knowledge spillovers back to the home country (Kokko, 2006).
Although it is difficult to classify FDI into these neat categories (Lipsey, 2002), one can distinguish between horizontal and vertical foreign direct investment as mentioned earlier.
According to Navaretti & Venables (2004), most of the authors who have tried to separate the FDI data in horizontal and vertical, have shown that HFDI predominantly substitute for trade, while VFDI is a complement to trade. This substitution effect of HFDI on domestic trade is due to the fact that ´horizontal firms´ are mostly firms that want to serve the foreign local market instead of producing products and services for the export to the home country. That is also why the main motives for HFDI is getting access to large foreign markets and avoiding trade costs.
2.2.2 Substitutability_empirical results other researchers
Even though most of the empirical researchers found a complementarity effect of outward FDI on domestic output, there are also researchers who found a substitution effect. One of these researchers is Svensson (1996). He found substitution between Swedish FDI abroad and exports from Sweden. His results show that increased foreign production replaces parent exports of final products and
complements the exports of intermediates, but the net effect was found to be negative. In contrast to the many other studies his firm-level study also included exports to countries in which manufacturing affiliates have not been establish, in order to avoid sample selection bias. Moreover, he took foreign affiliates‟ exports to third countries into account in his research. According to Svensson (1996) these two differences with other studies might be the explanation for the positive or non-negative
relationships of many other researchers in contrast to his negative results.
While many researchers have used firm-, industry-, or country level approaches to investigate the relationship between exports and FDI, Blonigen (2001) have used product-level data. The advantages of using disaggregated product-level data is, according to Blonigen (2001), that one is able to
separately identify substitution from complementarity effects (in his case from vertical relationships). In his research he proved substitution effects between FDI and exports and that they often occur in large one-time changes, not gradual changes over time.
found a large substitution effect in these circumstances, since the firms facing this protection probably substitute foreign production for exports to avoid this protection.
Also Frank & Freeman (1978) found that U.S. FDI substituted for U.S. exports and that the net employment effect of FDI was negative.
There are also researchers who found both a complementarity as well as a substitutability relationship between FDI and domestic exports. For example, Head & Ries (2001) found net complementarity for the whole sample, however, the relationship varied across firms. The degree of substitution or complementarity they observed varied across firms in accordance with the ability of the firms to supply the foreign affilates with intermediate inputs. Those firms that are unlikely to supply the foreign affiliates with intermediates, are unlikely to have a stimulation of their exports due to this FDI. Lipsey (2004) argues that the effect of FDI on exports heavily depends on whether the type of
investment is horizontal or vertical; whether the target country is industrialized or developing; and whether the plant level or firm level economies of scale are predominant. So, while the overall prediction is that the home country does not lose exports or output when investing abroad, it may be possible to find individual cases where it does happen (Lipsey, 2004). Braunerhjelm and Oxelheim (2000) did one of the few studies discussing such differences. They argued that FDI and exports should be complements in industries based on Swedish raw materials, and be substitutes in industries where R&D and technology are the competitive assets (Kokko, 2006). Braunerhjelm et al. (2005) generalize these results by saying that we should expect complementarity in case of vertical FDI and substitution in case of horizontal FDI.
Moreover, Egger (2001) mentions in his article that the presence of adjustment costs can also play a role in determining the relationship between FDI and exports. As known from the investment literature, the presence of adjustment costs leads to sluggish adjustments of capital stocks (Lucas, 1967), which leads to a difference between the short-run and long-run (steady state) situations. Hence, there is a possibility for complementarity to occur in the short-run and substitution to occur in the long-run, or vice versa. Cushman (1988) for example, also included adjustment costs in his research and he found a substitutive relationship between FDI and exports. However, Egger (2001) himself did not find any clear-cut and significant relationship between outward FDI stocks and exports in his research.
2.2.3 Complementarity_theory/Vertical FDI
local production will decrease the costs of supplying that market. Depending on the size of the gain in market share and the importance of parent input in the affiliate‟s output, local production might raise or lower parent exports, because local production might require some input from the parent, such as components (Blomström et al., 1987).
There are various reasons for complementarity to occur. Lipsey and Weiss (1984) for example, state that the foreign production of one product may increase the demand for that product but may also increase the total demand for all of the firms‟s products. This might happen through among others the commitment-to-market effects on consumers and the more efficient and quicker way of delivery and distribution. Moreover, a sales office may provide valuable services to the foreign customers that cannot be given efficiently by contracting with local agents. And by establishing as a local producer, the MNC probably familiarizes a foreign market with the parent company‟s reputation and name (Head & Ries, 2001). Brainard (1997) uses the term “proximity advantages” for the fact that the foreign production and sales of a MNC‟s product have the advantage of promoting exports of goods produced by the firm in the domestic country. Moreover, it is possible that the presence of MNC affiliates abroad facilitates the diffusion of information about other producers from the home country what might have a positive effect on the domestic employment and export level (Kokko, 2006).
Besides this demand complementarity, complementarity can also occur by vertical production relationships. Investment in a foreign country may increase exports of intermediates of the home country to the host country (Blonigen, 2001). However, it is very difficult to disentangle what drives the complementarity; the demand complementarities or the vertical relationships.
Grubert & Mutti (1991) and Lipsey & Weiss (1984) found the vertical relationships to be the reasons for complementarity. They found that the strongest positive effects of FDI on exports was found for intra-firm exports, which suggests that FDI increases exports of intermediates.
Horst (1979) offers another explanation for complementarity to occur. He does not reference to exports of intermediate goods, but to ancillary activities. According to him, one has to make a distinction between manufacturing and ancillary activities of foreign affiliates. While the former activities tend to substitute for home exports of final goods, the latter may complement the home country exports. Examples of these ancillary activities are technical assistence, suiting the products to the local preferences, distribution and repairs. These ancillary activities may stimulate the exports of other final goods from the home country (Horst, 1979).
Moreover, in case of vertical FDI (by exploiting lower factor costs abroad), the vertical investment makes the MNE more competitive abroad as well as at home. This probably creates new employment and raises exports (Kokko, 2006).
VFDI is FDI with the aim of finding low-cost locations for parts of the production process, this FDI flows especially from high-income to low-income countries. The finding that VFDI complements domestic exports, contradicts with the general thought that investment in cheap-labor countries has a negative effect on home economies, especially on the labor market. A reason for this finding could be that VFDI reduces the production costs for the multinational as a whole, and therefore increasing total output and employment of complementary activities at home (Navaretti & Venables, 2004).
2.2.4 Complementarity_empirical results other researchers
Swedenborg (1979, 1982) is a well-known researcher in the field of FDI. In her research she
investigated the relationship between foreign production and the exports of individual Swedish parent companies. She found a positive relationship between exports and foreign production, but the
significance of this relationship is never tested (Svensson, 1996). Moreover, she also did research by distinguishing her data between exports shipped by Swedish companies to their foreign affiliates and exports sold to third parties. In this case she statistically proved that the exports to their foreign affiliates are complementary to FDI and the exports to third parties are substitutes to FDI, as she already expected (Swedenborg, 1979). However, in Swedenborg´s latest paper she concludes that the net effect of foreign production is probably close to zero (Swedenborg, 2001).
The complementarity between exports and foreign production, is also found by Blomström et al. (1987). They not only looked at levels of exports at one time, but also at changes in exports over time. They stated that the predominant relationship between foreign production by Swedish and U.S. affiliates and the exports of Sweden and the U.S. to these foreign countries, is something between neutrality and complementarity (Blomström et al.,1987). Besides that, the effect was insignificant in some industries. Lipsey and Weiss (1981, 1984) concluded that production by affiliates in a country and U.S. exports are complementary. Meanwhile, exports from other industrial countries were negatively influenced by the existence of U.S. manufacturing affiliates.
Bergsten et al. (1978) have also used U.S. data to show that there is a (weak) complementary effect between investment abroad and exports. However, this complementary effect is seen up to a certain level, since most initial investment goes into marketing ans assembly. However, beyond that level further foreign investment starts to replace exports (Svensson, 1996). Moreover, Lipsey et al. (2000) found in their study on Japanese firms, that Japanese affiliate production substitute for exports by rival countries while promoting the exports of Japan. Fontagné & Pajot (2002) came to the same conclusion while focusing on France exports and the Productivity Commission (2002) found this positive effect on the exports of Australia.
reached similar overall complementariy conclusions (at aggregate, industry and firm level) when investigating the effect of FDI on exports.
Most forms of FDI replace some previous home country production and exports of finished goods. On the other hand, FDI tends to increase exports of intermediate goods from the domestic parent firm or other domestic suppliers to the foreign affiliates. The net effect of these two determine whether FDI and exports are complements or substitutes and it is likely to vary from case to case (Kokko, 2006). However, one can note that most econometric studes which investigate this relationship, concluded that the complementarities tended to outweight the substitution effects. Therefore most governments stimulate FDI by providing policies which encourage national firms to invest abroad (Kokko, 2006).
In my research I will use the gravity model to determine whether Dutch outward FDI have stimulated or have been complementary to Dutch bilateral exports in the years 1996 till 2008. In other words, the effect of Dutch outward FDI in a foreign country on the Dutch exports to that particular country will be investigated. Because the gravity model consist of factors that has to do with spatiality and geography, this model will be very usefull to investigate this relationship. The gravity model includes distance for example, which is a very important factor in choosing between exporting to or investing in a foreign country as the right mode of supply. In the next two sections the gravity model will be explained extensively.
2.3 Gravity model
Gravity models have become very popular in the last four decades in empirical analysis of bilateral trade and foreign investments (Földvári, 2006). It has been used to explain econometrically the ex post effects of economic integration agreements, national borders, currency unions, language, and other measures of trade costs on bilateral trade flows (Rose, 2004).
The gravity model is based on four assumptions (Anderson, 2008): expenditure on goods from all sources is equal to income from sales to all sources, markets for all goods clear, products are
differentiated by place of origin and all countries have the same tastes for goods. In its simplest form in a frictionless world, when free trade is assumed so that all countries have identical prices, the gravity equation states that the bilateral trade between two countries is directly proportional to the product of the countries‟ GDPs. This means that larger countries will tend to trade more than smaller countries (Feenstra, 2004).
Until recently researchers have used a gravity model akin to Newton‟s Law of Gravity which describes the bilateral trade flows from an origin i to a destination j to some supply-, demand- and trade
and the trade stimulating or restraining factors are for example distance as a proxy for trade costs and trade preferences and other factors that affect trade barriers between that pair of regions (Brenton et al., 1999). Anderson & van Wincoop discuss some important trade barriers in their article of 2004. Among these are language barriers; policy barriers (tariffs and non-tariff barriers (NTBs)); currency barriers (also discussed by Rose & van Wincoop, 2001) and information barriers (discussed by Portes & Rey, 2002).
2.3.1 Gravity model_theoretical foundation
Despite the gravity model‟s popularity and empirical success, this model was long said to be lacking strong theoretical foundations (Ivus & Strong, 2007). More recently, different authors have therefore tried to develop theories which give an explanation why these gravity equations fit the data so well. For example, Deardorff (1998) has shown that the gravity model can be derived from the H-O model, the model that assumes trade to be based on relative factor abundances. When countries produce the goods and services that require a relatively large amount of the factor in which they are relatively abundant, it implies that trade tends to occur between a rich country and a poor country (Choi, 2002). However, it is proved that this comparative advantage theory contradicts the empirical findings done on this subject. An example is the research by Leontief (1953) who wanted to test this H-O model empirically by looking at the United States. He found that the capital-labour ratio of the U.S. exports is smaller than the capital-labour ratio of the U.S. imports. Because the U.S. is seen as capital abundant, this finding contradicts the predictions of the H-O model. This phenomenon is known as the Leontief paradox. However, in his article Deardorff (1998) argues that the H-O model does is consistent with the gravity equation, at least in some of the equilibria that it permits.
According to Ivus & Strong (2007) the gravity model can be based on the Walrasian general
equilibrium theory with demand and supply forces. The aggregate income of the importer is a proxy for the demand in the destination region and the aggregate income of the exporter can be viewed as a proxy for the supply in the home market. Distance (as a proxy for trade costs) drives a wedge between this demand and supply, resulting in a lower equilibrium export flow. Anderson (1979) provided a theoretical basis for the gravity model by assuming constant elasticity of substitution and goods that are differentiated by country of origin.
consumption patterns. Thus, nations with similar demands and similar per capita income levels, would develop similar industries and therefore trade with each other in differentiated, but similar products (Bernasconi, 2009). Helpman & Krugman (1985) made an other important contribution to the gravity model, they assumed increasing returns to scale in production when deriving the gravity model. In the same way, Evenett & Keller (1998) derived the gravity model from both the Hecksher-Ohlin model and the increasing returns to scale hypothesis, under both perfect and imperfect product specialization.
It is clear that the same basic gravity equations can be derived from several trade theories. However, the fact that the gravity equation can be derived from different trade theories, makes among others Anderson & van Wincoop (2003) and Deardorff (1998) quite skeptical about using the gravity equation for the justification of any of the trade theories. According to Anderson & van Wincoop (2003), the theoretical foundations of the gravity equations are still unsatisfactory.
This unsatisfaction with the theoretical foundation of the gravity model was soon restored when Anderson & van Wincoop (2003) accounted for the omitted price-terms in gravity equations, which they referred to as the ´multilateral resistance´ terms. Because of the lack of reliable data normally price-terms are not included in a lot of empirical studies, while Anderson (1979), Anderson and van Wincoop (2003), Bergstrand (1985), Deardorff (1998), and Feenstra (2004) all suggest that the traditional gravity equation is probably misspecified due to the omission of multilateral price terms. In reality, the export flows between country i an j are influenced by the prices of goods from the other countries in the world. When there are border effects, transport costs or tariffs for example, prices are not equalized across countries anymore. It is therefore very necessary to take account of the overall price indexes in each country (Baier & Bergstrand, 2009).
However, it is still a challenge to estimate the gravity equation where price effects are included. Feenstra (2004) mentions three different approaches in his book to solve the problem. The first approach is the one used by Baier & Bergstrand (2001) and Bergstrand (1985, 1989) and, where they estimate the gravity equation by using published price index data. In this approach, the growth of trade depends on changes in transport costs, changes in the sum of GDP, changes in the relative country size and changes in the prices of each country, measured with GDP deflators. However, Feenstra (2004) uses this approach and tested this equation with OECD countries and he found that the entire increase in trade was only due to the sum of GDPs, transport and tariff costs and there was no role at all for the convergence in relative country size. So, his conclusion was that relative country size fails to be economically important in explaining the growth in trade among the OECD countries.
for some less developed countries and because most studies use constant price data, it is not very convenient to use (Földvári, 2006). That is why Feenstra (2004) mentions a second approach to measure the price effects and in this approach estimated border effects are used. This approach was also used by Anderson & van Wincoop (2003). Instead of using data to measure prices, they consider ´iceberg transportation costs´ and therefore make a distinction between c.i.f. prices and f.o.b. prices due to distance and other factors. To solve for the unknown prices in the equality where the value of output of the firm (using f.o.b. prices, before the quantity has ´melted´) is equal to the expenditure of consumers (using the c.i.f prices and where the quantity has already ´melted´), one can use the market-clearing conditions to obtain an explicit solution for the prices. However, Anderson & van Wincoop (2003) make use of the implicit solution to these prices and therefore they finally got an equation where bilateral trade between countries depends on their GDPs and their implicit price indexes. Anderson & van Wincoop (2003) referred to the price indices as the ´multilateral resistance´ terms. The third and simpler alternative to estimate the gravity equation, is to use fixed effects (country-specific dummies) to take account of the unobserved price indexes (Feenstra, 2004). With this
approach one can use the Ordinary Least Squares method and because of this simplicity this approach has been used by a number of researchers, among others Eaton & Kortum (2002), Harrigan (1996), Head & Mayer (2001), Head & Ries (2001), Hummels (1999), Redding & Venables (2004) and Rose & van Wincoop (2001). In this approach source and destination region fixed effects are used and the coefficients of them estimate the multilateral indexes.
The fixed-effects method might be considered as the preferred empirical approach, because it gives consistent estimates of the average border effect between countries and it is very easy to implement.
Accounting for the omitted price-terms, was seen as a large improvement for the theoretical foundation of the gravity model. Of course the gravity model has still some drawbacks. Baier & Bergstrand (2009) mention one in their article. They state that the traditional gravity model ignores the fact that the export of country i to country j is affected by the trade costs between region i an j relative to those of the rest-of-the-world (ROW). Moreover, the economic sizes of the ROW countries and the prices of their goods matter as well and are also ignored in the gravity model.
Even though the gravity model has some drawbacks, the gravity equation is still a very important tool for international trade economists and policymakers (Földvári, 2006). This is due to the improvement by Anderson & van Wincoop (2003) and since the fact that the gravity model is empirically highly successful with a strong explanatory power and has a high degree of flexibility, a high policy relevance and is convenient to use.
Based on these reasons I will also use the gravity model in my research to explain the relationship between Dutch outward FDI and Dutch bilateral exports.
2.4 FDI versus trade models
In my research I will use a gravity equation where export is the dependent variable and FDI is one of the explanatory variables. This method is some combination of the existing FDI and trade models. An example of a model that explains FDI is the OLI (Ownership Location Internationalization) framework of Dunning (1981). Even though the OLI framework of Dunning (1981) is very good in explaining the conditions necessary for direct investment to occur, this model has been lacking in explaining the key trends of FDI over the past three decades. According to Brenton et al. (1999), this model does not, for example, clarify the increased volume of the two-way investment between rich industrial countries when trade barriers decrease. Moreover, this model does not say much about the choice among alternative modes of entry, such as licensing versus exporting versus joint venture (Markusen, 2002).
Recently, there have been developed models that only consider the ownership and locational
advantages of the OLI framework and does not take into account the internationalization advantages. In these models, for example used by Brainard (1997) and Markusen & Venables (1996), the two advantages have been integrated in general equilibrium trade models where multinationals arise endogenously. In these models two-way FDI is able to occur between countries and both in the OLI approach and in these models, trade and FDI are seen as substitutes (Brenton et al., 1999). Brenton et al. (1999) have used the gravity model to investigate the impact of the integration between the EU and the CEECs on FDI and also Markusen (2002), Markusen & Maskus (2002) and Markusen &Venables (1996), have used several methods to explain FDI trends.
There have been also developed several trade models and theories. An example is the New Trade Theory. This theory was developed in order to try to explain the most significant trends in the post World War II data. It wanted to explain especially the increase in the ratio of trade to GDP; the increase in trade between industrialized countries; and the fact that trade among industrialized
of the increase in trade between industrialized countries, but it is not capable of explaining the large increase in the ratio of trade to income.
There are many reserachers who have used the gravity model in their research to explain trade flows and they all extend the basic gravity model in different ways for different purposes. For example, Anderson & van Wincoop (2003) extend the basic gravity model by including a remoteness variable in addition to the direct bilateral distance variable. To explain bilateral trade between a country and a specific trading partner, they incorporate the average distance of an exporting country from its trading partners except from the specific trading partner himself, which they call remoteness. It is assumed that the remoteness of an exporter from the rest of the world has a positive effect on bilateral trade (Anderson & van Wincoop, 2003).
Aitken (1973) included in his gravity model a dummy which indicates whether countries participate in a preferential trading arrangement or not, to investigate the effect of EEC and EFTA on European trade. Since this study of Aitken (1973), the gravity model has been used very often to test the effect of preferential trading arrangements on trade flows.
Helpman (1987) was the person who incorporated the role of differential country size in the gravity equation.When a region consists of two countries and the size of this region is fixed, it would mean that two countries of unequal size are not able to trade that much compared to two countries of equal sizes. Therefore the Theorem Helpman (1987) states that the volume of trade relative to GDP is larger, the more similar the countries are in size (in income levels). However, it is proved that this only holds for OECD countries, not for non-OECD countries. The reason for why the basic gravity model does not work for non-OECD countries, is that the basic gravity model assumes that countries are
specialized in different goods. The trade between industrialized countries is indeed often in different goods, but that does not hold for most of the trade between developing countries which consists of basic agricultural goods or low-skilled commodities. So, in that case there is no reason at all for the gravity equation to hold (Feenstra, 2004).
Another application of the basic gravity model came from McCallum (1995) who compared
intranational trade between Canadian provinces to international trade between Canadian provinces and U.S. states. In this equation bilateral trade between two regions depend on the output of both regions, their bilateral distance, and whether they are separated by a border. A lot of researchers use
McCallum´s (1995) gravity equation in their research. For example, Frankel et al. (1998) estimated the gravity equation similar to McCallum´s (1995) to determine the impact of trade unions on trade and Rose (2000) to determine the impact of monetary unions. Moreover, McCallum´s (1995) gravity equation is also used to determine the impact on trade of particular economic integration agreements (EIA) like free trade agreements (Tinbergen, 1962), different languages, adjacency, migration flows, equity flows and FDI flows and a variety of other factors (Anderson & Wincoop van, 2003).
that the so-called “border effects” lead to a lot more internal trade in Canada than external trade. The reason for why the Canada-US border has such a large impact, is given by Anderson & van Wincoop (2003). It is because border effects have an asymmetric effect on countries of different sizes, and in particular, have a larger effect on small countries. To avoid this bias, Anderson & van Wincoop (2003) propose to rederive the gravity equation and thereby including trade barriers, like tariffs or transport costs, right from the beginning.
3. Hypotheses
As discussed in Chapter 2, there is no consensus on the effects of outward FDI on home exports. Whether outward FDI leads to an increase or decrease in domestic exports, depends on whether home and foreign activities are complements or substitutes to each other. On the one hand, outward FDI can replace domestic exports of final products, but on the other hand it can complements the exports of intermediates from the parent country to the foreign affiliates. There are many more reasons for substitutability or complementarity to occur as discussed in the literature review (chapter 2). Even though some researchers like Svensson (1996) and Blonigen (2001) have found a substitution effect of outward FDI on domestic exports, most of the researchers in this field concluded that the
complementarities tend to outweight the substitution effects. Therefore most governments stimulate FDI by providing policies which encourage national firms to invest abroad (Kokko, 2006).
In line with this, I hypothesize the following for the Netherlands:
Hypothesis 1: Dutch outward foreign direct investment had a positive impact on Dutch bilateral exports over the period 1996-2008.
Although it is difficult to classify FDI into these neat categories (Lipsey, 2002), one can distinguish between horizontal and vertical foreign direct investment as discussed in chapter 2. According to Navaretti & Venables (2004) in general terms a relationship of complementarity is found between outward FDI and export by most of the authors. However, when it is possible to split foreign direct investment into HFDI and VFDI, several authors like Blonigen (2001), Braunerhjelm et al. (2005) and Head & Ries (2001), showed that complementarity appears to hold for VFDI and HFDI seems to be a substitute for trade (see chapter 2). These results are line with theoretical predictions. If this logic is true, I expect the Dutch bilateral exports to increase more due to outward FDI in low-income countries than due to outward FDI in high-income countries. I will therefore test the following hypothesis:
4. Data description
4.1 Dependent variable
Log of Dutch bilateral exports (lnEXPjt)
The dependent variable in my research will be the log of Dutch bilateral exports to its main
tradingpartners (j)1. The export values (Free on Board, f.o.b.) are expressed in US dollars and in order to get a linear relationship between the variables in the gravity equation, I will take the logarithm of the bilateral exports. Most empirical studies assume a log-linear or log-log functional form for gravity equations in order to get a linear relationship (Földvári, 2006). Moreover, taking the natural log of the variables also reduces the influence of large values (Azémar & Desbordes, 2009).
The trade data I use is collected by the Organisation for Economic Cooperation and Development (OECD) and mostly follow the UN recommendations. Trade data is therefore considered to cover all goods which add to or substract from the resources of a country as a result of their movement into or out of the country (OECDStat_Metadata). This implies that goods in transit are not included in the Dutch export figures.
Following Blomström et al. (1987) and Lipsey & Ramstetter (2001), I take the total bilateral Dutch exports in my research and not only the bilateral exports of the Dutch foreign direct investment parent companies. This is because FDI might not only be at the expense of the MNE exports, but also at the expense of the exports of the rival companies in the home country (Blomström et al., 1987;
Braunerhjelm (1991). On the other hand, when MNCs produce more abroad, subcontractors at home may be able to raise their exports. To include this effect of FDI on domestic rival companies, total bilateral Dutch exports are used instead of Dutch MNEs bilateral exports.
In this thesis I will only focus on the export flows of goods, not on the export of services, since good data on services in the Netherlands are lacking. The exportgoods can be subdivided into 97 different commodity groups which in total covers 4346 products.2 One can find these commodity groups in Appendix II at the end of this paper.
For the bilateral Dutch export figures, I have used the OECD International Trade by Commodity Statistics (ITCS) database (OECD, Stat Extracts). This database consists of values and quantities of exports and imports by partner countries and by commodity or industry. The commodities are
1
The 52 main tradingpartners of the Netherlands are based on information from the DNB and are given in appendix I.
2
available at the most detailed level of the SITC (revision 3 with 4346 products) and the Harmonised System (HS 1988).
4.2 Independent variables
Log of Dutch net outward FDI flow (lnFDIflowjt)
In this thesis I will investigate the effect of Dutch net outward FDI on the bilateral Dutch exports with its major trading partners. Therefore Dutch net outward FDI is one of the most important independent variables in this research. I use net outward FDI instead of gross outward FDI, because gross outward FDI figures reflect the sum of the absolute value of in- and outflows and therefore do not account for disinvestment. In order to get a linear relationship between the variables, I have to take the log of the Dutch net outward FDI. Due to the inclusion of disinvestment in the FDI figures, the net outward FDI flows have negative values in some years and it is therefore impossible to take a logarithm of those values. These values are therefore excluded from my database.
FDI data are from the Foreign Direct Investment Statistics database provided by the OECD (OECD, Stat Extracts) and are presented in millions of national currencies, so in my case in millions of euros. The FDI data comply with the guidelines of the OECD Benchmark Definition of Foreign Direct Investment, 2008 (OECDStat_Metadata). According to these guidelines, FDI is defined as: “A category of cross-border investment made by a resident in one economy (the direct investor) with the objective of establishing a lasting interest in an enterprise (the direct investment enterprise) that is resident in an economy other than that of the direct investor” (OECD, 2008).34 The reason for the direct investor to invest abroad is to establish a strategic long-term relationship with the direct investment enterprise with the aim of having a significant degree of influence in the management of the direct investment enterprise. Direct investment is evidenced when the direct investor owns directly or indirectly at least 10% of the voting power5 of the direct investment enterprise. A strict application of the 10% threshold is recommended in order to achieve global consistency of FDI statistics and to facilitate international comparisons. Direct investment is not only limited to equity investment, but also relates to reinvested earnings and inter-company debt (OECD, 2008).6
Moreover, there are different types of FDI. An investor may purchase or sell existing equity in case of Mergers & Acquisitions, an investor may do a new investment in case of greenfield investment and extensions of capital is also one type of FDI, where additional new investments are made as an
3
For the defintions of „direct investor‟ and „direct investment enterprise‟, see OECD Benchmark Definition of Foreign Direct Investment, 2008.
4
Direct investment by intermediate holding companies (Special Financial Institutions) in the Netherlands is not included in the FDI figures (OECD, 2008).
5
For the definition of „voting power‟, see OECD Benchmark Definition of Foreign Direct Investment, 2008.
6
expansion of an established business (OECD, 2008). There are even more types of FDI, but in my thesis I will not make a distinction between these different types of FDI, I will use the aggregate FDI figures.
There are three different kinds of FDI statistics: investment positions (stocks), financial transactions (flows) and associated income flows between enterprises which are related through a direct investment relationship. FDI flows provide information for FDI activity within a given reference period (year, quarter, month) , while FDI stocks indicate the levels of investment held at the end of the reference period. Direct investment stocks are not only affected by changes in price, exchange rates and volume, but they are also affected by financial transactions prior to (and during) the period. And therefore I have chosen to use FDI flows in my research instead of FDI stocks, to only include the effect of investment during the period on the Dutch exports and not the effect of eventually financial transactions prior to the period.
The OECD Benchmark Definition recommends to measure the direct investment acccording to their market value. It places all assets at current prices and therefore allows for comparability and
consistency between flows and stocks of assets of different enterprises, industries and countries, as well as over time (OECD, 2008).
In the fourth edition of the OECD Benchmark Definition of Foreign Direct Investment (2008), two analytical approaches to presenting FDI data are mentioned:
For aggregated FDI assets/liabilities data; and
For detailed FDI data on a directional basis analysed separately by partner country and by economic activity. This approach is called the „directional principle‟.
In my thesis I make use of the directional principle. According to this principle, loans that are granted by subsidiaries to the parent company will be subtracted from the FDI values, because these loans are considered to be a disinvestment by the direct investor. Moreover, investments into fellow enterprises7 abroad (dependent on the residence of the ultimate controlling parent of the resident fellow
enterprises) will be added to the FDI figures.8
7
For the definition of „fellow enterprise‟, see OECD Benchmark Definition of Foreign Direct Investment, 2008.
8
FDI for the reporting country (RC) according to the directional principle is summarised as follows:
Outward investment = Investment by direct investors of the RC in direct investment enterprises abroad minus
Dutch net outward FDI stocks (FDIstockjt )
In order to calculate the ratio of the percent change in exports to the percent change in net outward FDI to get the elasticity which is used as another method discussed in the sensitivity analysis (see chapter 9), I also need to have data on the Dutch net outward FDI stocks. Dutch net outward FDI flows will be divided by the net outward FDI stock of the starting year. As mentioned above, FDI stocks indicate the levels of investment held at the end of the reference period. Even as the data on the FDI flows, the data on the FDI stocks is from the Foreign Direct Investment Statistics database provided by the OECD (OECD, Stat Extracts), they comply with the guidelines of the OECD Benchmark Definition of Foreign Direct Investment, 2008 (OECDStat_Metadata), and are presented as market values in millions of euros.
Log of Dutch and foreign Income (lnGDPnlt & lnGDPjt)
Income is the most used variable in gravity equations as a measure for country size. It is common to use the gross domestic product (GDP) of a country for this measure (Anderson & van Wincoop, 2003; Bergstrand, 1985, 1989; MC Callum, 1995). In this research, both the log of GDP of the Netherlands as the log of GDP of the Dutch major trading partners are used.
The GDP of the home country is expected to have a positive influence on exports, since an increase in GDP means more domestic production and when this increase in supply is more than the increase in domestic demand, more products will be exported to avoid domestic price decreases. Moreover, when prices have been decreased already due to the relative high supply, the country becomes more
attractive for foreigners to import from which increases the domestic exports.
The positive effect on domestic exports also holds for the GDP of the host country. The greater the total foreign income, the greater the demand in the host country and the more Dutch exports will be (Svensson, 1996). Moreover, it appears that higher income countries trade more in general, because of for instance the superior transportation infrastructure (roads to the interior, container ports, airports etc.) (Head, 2003).
In my case it is better to use GDP instead of GNP, since GDP reflects domestic income and does not include goods and services produced by national corporations in other countries. This makes sure that our results are not driven by potential built-in correlations between our domestic income measure and outward FDI (Herzer, 2008). Moreover, in order to get a linear relationship between the variables, I take the log of GDP in this research.
summation of final consumption expenditure, gross fixed capital formation, changes in inventories and exports less imports (OECDStat_Metadata).9
The gross domestic product used in my research is from the World Economic Outlook Database of the IMF and is based on the purchasing-power-parity (PPP) valuation (IMF, WEO Database 2009), because converting each country‟s GDP into a common currency, made comparisons between countries possible. PPPs are currency converters that equalise the purchasing power of the different currencies. One can also convert the national currencies into a common currency by using exchange rates, but according to the OECD these convertions give a misleading comparison of the volumes of goods and services in GDP (OECDStat_Metadata). I use therefore the GDP based on the PPP valuation.
The GDP figures are given in billions of current international dollars.
Log of foreign Income per capita (lnGDPCAPjt)
Another independent variable in my research is the income per capita of the Dutch major trading partners. This is as a measure for economic productivity, measured as the GDP per capita (per person). Again, I take the log of this variable to get a linear relationship. As discussed in Blomström et al. (1987) the GDP per capita may have a positive as well as a negative influence on trade (exports). In general, in case of demand-side influences, one would expect a positive effect when the income elasticity of demand is high and a negative effect when the income elasticity is low. However, there are also supply-side influences that this variable could represent. For instance when the GDP per capita is high, it could mean that it is strongly associated with a strong comparative advantage in skill-intensive products and therefore the import demand will be low for these products. It depends on which influence is dominant to determine whether the effect of GDP per capita on exports is positive or negative (Blomström et al., 1987).
The figures for GDP per capita come from the database of the International Monetary Funds, called the World Economic Outlook Database (IMF, WEO Database 2009). As was the case for the GDP figures, the GDP per capita figures are also based on the purchasing-power-parity valuation and they are given in current international dollars (units).
Log of Distance (lnDISTj)
Following among others Brenton et al. (1999), Egger (2001), Head (2003) and Lipsey et al. (2000), we use distance as a proxy for transport costs. Exports are expected to decrease to more distant markets, since the trade costs involved in exporting to those markets will increase. Distance also indicates the
9
Two other measures of GDP are: 1) output less intermediate consumption (i.e. value added) plus taxes less subsidies on products (such as VAT): 2) the income earned from the production by adding employee compensation, the gross operating surplus of enterprises and government, the gross mixed income of
time elapsed during a shipment, so the probability for perishable goods to survive depends negatively on the time in transit. Moreover, distance indicates communication costs, transaction costs and synchronization costs. The latter occurs when factories combine several inputs in the production process, so they need those inputs on time to prevent bottlenecks to occur (Head, 2003). Therefore we expect exports to be inversely proportionate to distance.
Distance in my research is the bilateral geographic distance between the Netherlands and its major trading partners (j), which is time-invariant. For this variable I use figures out of the ´dist_cepii.xls´ database of the CEPII (Centre d'Etudes Prospectives et d'Informations Internationales). It is common to model distance according to the ´Great Circle´ formula (Anderson & van Wincoop, 2004; Head, 2003) which uses latitutes and longitudes of the most important cities/agglomerations (in terms of population) or the geographic coordinates of the capital cities (Mayer & Zignago, 2006). In most cases the main city of a country is also the capital city of that country, however, this does not hold for all the countries used in my sample.10 For 8 out of the 52 countries in my sample, the capital was not
populated enough to represent the ´economic center´ of the country (Mayer & Zignago, 2006). This is the reason why I have chosen to use the distance figures measured following the Great Circle formula, from the economic center of the Netherlands to that of the trading partners and not between the capitals of those countries. This is what Baier & Bergstrand (2009) did as well.
The bilateral distance figures are provided in kilometers (CEPII, dist_cepii.xls database).
Log of foreign Index of Economic Freedom (lnEFIjt)
Following Brenton et al. (1999), I also include in the regression equation an Index of Economic Freedom (EFI). This index ranks countries according to their economic freedom and it takes into account 10 factors of „economic freedom‟.11 Higher index values stand for freer countries.12 This EFI is annually provided by The Heritage Foundation and The Wall Street Journal (The Heritage
Foundation & The Wall Street Journal, 2010).
This index can be seen as an indicator of the „market-friendliness‟ of economic policies in the host country. On the one hand we can expect Dutch bilateral exports to increase to countries where the EFI increases, since freer countries are more easily to reach and favourable countries to export to. On the other hand, when foreign direct investment and exports are assumed to be substitutes, Dutch bilateral
10
Those cases where the economic center differs from the capital are: South Africa (The Cap), Germany (Essen), Australia (Sydney), Brazil (São Paulo), Canada (Toronto), United States (New York), Nigeria (Lagos) and Turkey (Istanbul).
11
To measure economic freedom and rate each country, the authors of the Index study 50 independent economic variables. These variables fall in the following 10 categories (factors of economic freedom): business freedom, trade freedom, fiscal freedom, government spending, monetary freedom, investment freedom, financial freedom, property rights, freedom from corruption and labor freedom (The Heritage Foundation & The Wall Street Journal, 2010).
12
exports could also decrease to countries where the EFI increases. This is because countries that has a high EFI index are ´freer´ countries and they therefore not only attract foreign trade but also foreign FDI. When the increase in FDI decreases the exports in case they are substitutes, the effect of EFI on Dutch eports may also be negative. It is therefore very interesting to include this EFI-variable in my regression equation and see which effect it has on the Dutch bilateral exports.
4.3 Dummy variable
Member of the Economic and Monetary Union (EMUjt)
In order to explore the possible relationship between bilateral exports and EMU-membership, a dummy for host country membership in the EMU is included with a value of one if the Dutch partner is an EMU-member and a value of zero when the partner is a non-member.
Most studies (for example: Choi, (2002) and Frankel & Rose (2000)) have shown that two economies that have a Monetary Agreement and therefore share the same currency, trade more with each other than countries that do not share a common currency. This is because countries that do not share a common currency have higher risk uncertainty due to the exchange rate volatility which hinder trade flows. I expect therefore EMU membership of the Dutch partners to have a positive influence on the Dutch bilateral exports.
Facts about EMU membership are from the Dutch Central Government website (Central Government, PO box 51).
4.4 Interaction variable
Log of Dutch net outward FDI flow * Low-income countries (lnFDIflowjt*Lowincj)
In this research I will also test the theory of HFDI and VFDI on my data sample. This theory states that when distinguishing between HFDI and VFDI, the HFDI will substitute for domestic exports and VFDI will be a complement to domestic exports. Since VFDI is FDI with the aim of finding low-cost locations for parts of the production process, we might assume that the Netherlands invest vertically in developing countries (low-income countries). I will therefore include the interaction variable
and upper middle), or high income13. In the end I use the dummy variable Lowincj with a 1 standing for low- and middle income countries and a 0 for high income countries.
13
