Changes in the relation between FDI and Economic Growth Do developing countries in fact benefit less from FDI since the shift in international investment policies?

Changes in the relation between FDI and Economic Growth

Michelle Kiene

University of Groningen

Faculty of Economics and Business

Master’s Programme: International Economics & Business

This paper analyses the effect of FDI on economic growth in developing countries from 1974-99. The assumption is that the potential of FDI to enhance economic growth through technological progress has declined substantially with the introduction of policy changes by the WTO and other international organizations. During the 1960s and 70s developing countries, as the Asian Miracle Economies, could still implement a wide range of policy measures in order to ensure positive effects from FDI. However, most of these instruments have been rendered illegal by the WTO agreements and developing countries face ever greater pressure to liberalize their markets since the mid 1980s. As a result, multinational enterprises can enter developing country markets without any restrictions that would ensure positive benefits to the local economy.

Acknowledgement

TABLE OF CONTENT

TABLE OF FIGURES ... 3

1. INTRODUCTION... 4

2. LITERATURE REVIEW... 6

2.1 EFFECTS OF FDI ON ECONOMIC GROWTH... 6

2.1.1 Mechanisms of FDI-led Growth ... 6

2.1.2 Impediments to FDI-led Growth... 8

2.2 DIFFERENT TYPES OF FDI... 12

2.3 POLICIES VS. FDI... 14

2.3.1 FDI policies in the Asian Miracle Economies ... 15

2.3.1.1 FDI dependent strategies: Singapore and the new tigers ... 16

2.3.1.2 Autonomous strategies: Taiwan and Korea ... 19

2.3.2 Policies to assure positive FDI effects... 21

2.4 CHANGES IN INTERNATIONAL INVESTMENT POLICIES... 23

3. RESEARCH DESIGN ... 27

3.1 THE ECONOMIC MODEL... 27

3.2 HYPOTHESES... 31 3.3 DATA... 33 3.4 METHODOLOGY... 35 3.4.1 Panel data... 35 3.4.2 Estimation techniques ... 36 4. ANALYSIS ... 42 4.1 DIAGNOSTIC CHECKS... 42 4.2 GMM ESTIMATION... 47

4.3 TESTING THE MAIN HYPOTHESIS... 49

4.4 DISCUSSION... 52

5. CONCLUSION... 56

REFERENCES ... 59

TABLE OF FIGURES FIGURE 1. FDI EFFECTS ON THE TECHNOLOGICAL PROGRESS IN THE DOMESTIC ECONOMY... 9

1. INTRODUCTION

In an era of globalization, characterized by rapid technological change and steep increases in international investment flows, multinational enterprises (MNEs) have become the major source of innovation and international competitiveness (Lall, 2002). In 2006 foreign direct investment (FDI) flows reached a peak of $1,306 billion (UNCTAD, 2007). Many developing countries, as e.g. the Asian Miracle economies, took advantage of FDI in order to enhance technological progress and accelerate overall economic growth. The Asian countries experienced remarkably rapid growth from the 1960s until the 1990s creating a 16-fold increase in income during this period. Interestingly, most of these economies have seen significant policy interventions in order to create FDI incentives and assure positive effects from multinational entry. Among those policies that contributed to the development of the Asian economies, infant industry protection and policies to enhance technology transfers to domestic firms played a vital role.

However, international institutions like the World Trade Organization (WTO) have set rules that minimize the possibilities of developing countries to use FDI policies and, in addition, force them to open up their markets. Consequently, multinationals is given the opportunity to exploit developing country markets and resources without any kind of restriction that would ensure benefits to the local economy. As a result, the potential of FDI to promote technological progress and enhance economic growth in developing countries is supposed to have declined substantially since the introduction of these policy changes. Wade (2003) claims that regulations, which limit technology transfers to domestic firms, and an increased pressure to liberalize markets preclude developing countries from experiencing economic growth and, hence, catching up with the developed world.

the effect of FDI becomes less positive after a policy change. As a second option, I carry out a ‘shifting regression’ analysis for different groups of countries, which allows observing the relation between FDI and economic growth over time for different regions. The results can then be compared to the policy changes in the respective country group. The results suggest that the overall effect of FDI on economic growth is negative for the whole sample over the entire time period. However, when analysing the effect over time for the different regions, it turns out that policy changes towards more liberalization have had a negative effect on FDI-led growth in Africa, but a positive one in Latin America and Asia.

2. LITERATURE REVIEW

2.1 Effects of FDI on Economic Growth

During the last two decades there has been a big surge in FDI flows, and although the largest proportion still goes to the developed countries, FDI has become increasingly important in total aggregate investment and overall economic activity in both, developed as well as developing countries (Narula & Portelli, 2004). Most policy makers, especially in developing countries, made great efforts to attract FDI due to the belief that it has positive effects such as productivity growth, transfer of technology, integration into international production networks, access to new markets and the introduction of management skills and know-how about new products and processes (Alfaro et al., 2002). All those factors are expected to lead to enhanced industrial competitiveness and overall economic growth. However, the debate on the effects of foreign investments on the host economy is still intense and heated up by the widespread fear that multinationals may, in fact, hurt the national economy and crowd out domestic investments. A common believe is that they increase the competitive pressure for domestic firms, have great bargaining power and can move their activities easily between different plants in several countries (Navaretti & Venables, 2004). The academic literature has paid great attention to the effects of FDI on the host country, particularly with respect to its potential for development through enhanced economic growth. So far, the analyses of the link between FDI and economic growth have not reached a common conclusion, yet the majority of the studies agrees on the fact that the growth effect of foreign investment depends on the host country’s characteristics such as the absorptive capacity. The following section is dedicated to explain the main channels through which FDI may enhance the host country’s economic growth. In addition, the factors that may hamper the country to benefit from these mechanisms as well as possible negative effects of FDI are described.

2.1.1 Mechanisms of FDI-led Growth

technological progress as Borensztein et al. (1998) argue. Direct technology transfers from a multinational to its foreign affiliate are a fast and efficient way of acquiring new knowledge, especially for firms that lack the capabilities to invest in R&D (Damijan et al., 2003). Navaretti and Venables (2004) find that FDI increases productivity in the host country since foreign firms are slightly more efficient than domestic firms and, furthermore, possess a whole set of different characteristics. Domestic firms might benefit from this bundle of capital, know-how and technology through linkages or externalities which allow that productivity gains spill over to the local industry (Narula & Portelli, 2004). These spillover effects might occur between firms that are either horizontally or vertically integrated with the MNE.

Horizontal spillovers may occur through increased competition on the domestic market. Local firms that face increased competition by multinationals might adopt new technologies at a faster pace, for instance through licensing agreements. This, in turn, would raise the transfer of technological knowledge by MNEs to their foreign affiliates in the market and therewith increase the potential amount of spillovers to other local firms (Blomström et al., 2000). Hence, FDI would lead to technological progress through increased competition. Other important channels of horizontal spillovers are the demonstration of new products and processes by MNEs and the imitation by local firms using reverse engineering or industrial espionage (Kokko, 1992). The evidence on technology spillovers that arise between horizontally integrated firms is weak; a review by Görg and Greenaway (2002) that focuses on appropriate panel data studies concludes that most studies find a negative effect on the productivity of domestic firms. This might be explained by the fact that MNEs try to avoid spillovers of important technology and knowledge to their competitors (Narula & Portelli, 2004).

supplier industry to benefit from economies of scale and also may promote the set up of new companies. These effects are referred to as backward linkages since they occur in the upstream industry. Furthermore, improvements in the supplier industry also benefit domestic firms if they use the same inputs as MNEs. These forward linkages can also be found in the downstream industries of the MNE. Several studies find positive spillovers in particular for local firms in the upstream industries that stem from higher expectations of quality standards and reliability (Blalock & Gertler, 2003; Watanabe, 1983). However, the extent to which multinationals build linkages with local suppliers depends on the technological capacity of the domestic supplier industry and the size of the market, yet, they can be promoted by local content regulations (Alfaro et al., 2002).

Finally, employees that receive technological knowledge or acquire certain management techniques while working for a multinational could migrate to local firms and therewith provide for human capital spillovers. They could also use their obtained knowledge to open up their own business. Thus, FDI might also be a catalyst encouraging domestic firms to evolve in a certain sector (Blomström et al., 2000). Some studies show that MNEs in fact provide more training to their employees than local firms and that labour mobility includes movements from MNEs to domestic firms (Narula & Portelli, 2004; Görg et al., 2002).

The above mentioned mechanisms explain how FDI could have a positive effect on economic growth in the host country. However, there are several impediments that could prevent developing countries to make use of those mechanisms in order to benefit from FDI inflows.

2.1.2 Impediments to FDI-led Growth

F D I Spillovers Competition Demonstration + Imitation Human Capital Linkages Direct technology transfer horizontal vertical

provide a certain level of development in order to make use of the MNEs know-how. The study by Borensztein et al. (1998) shows that the potential growth enhancing effect of FDI depends crucially on the host country’s level of human capital measured by secondary school attainment. Moreover, a former study by Blomström et al. (1992) finds a positive effect of foreign investments on economic growth only for higher income countries, suggesting similarly that the capability to absorb new technology depends on the level of development. Finally, Alfaro et al. (2002) mention the development of local financial markets as a requirement to build linkages with local firms and facilitate technology transfer. Hence, it seems to be reasonable that Nunnenkamp (2004) argues that the benefits from FDI are easier to reap for less developed countries if FDI is more labour intensive and less technology intensive. However, as FDI provides for technology and human capital spillovers to the domestic industry, capabilities are built and the economy could benefit from higher technology FDI in the future. Figure 1 summarizes the potential effects of FDI on the host economy with respect to technological progress.

National capabilities

Figure 1. FDI effects on the technological progress in the domestic economy (source: own figure)

out domestic investments. Increased competition by MNEs might not only raise productivity in the industry and therewith induce the entry of domestic firms (crowding-in), but also lead to closing downs of domestic firms as they cannot compete with foreign firms (crowding-out). The effect on overall capital formation depends on the extent to which domestic firms are crowded out if foreign firms enter the market. The higher the technology level of the MNE compared to the technological capabilities of domestic firms, the higher the probability of crowding out through competition in the domestic market. Thus, positive effects of FDI depend crucially on the country’s capabilities and the ability to attract FDI with an appropriate technological level and in the right sector. Foreign investments that complement the domestic economy in sectors that lack the relevant technology not only possess a low probability of crowding out, but also provide little spillover effects for domestic firms. Moreover, competition can also take place in financial markets, if MNEs finance their investment by borrowing in the local market (Bloningen & Wang, 2005). Harrison and McMillan (2003) argue that domestic firms are crowded-out by foreign firms in the financial market since foreign firms are more profitable and liquid, thus they are a better investment for local banks. Domestic enterprises are generally considered more risky. Furthermore, artificially low interest rates in the host country are likely to increase local borrowing by MNEs and, consequently, aggravate domestic firms’ credit constraints due to credit rationing. In case that FDI crowds out local firms it can also lead to a decrease in employment, if not all employees are reemployed in the MNE (Navaretti & Venables, 2004). Empirical studies on the effect of FDI on domestic investments in developing countries offer mixed results. However, there is little evidence on crowding-out effects (Agosin & Machado, 2005); most of the studies find no evidence or even a positive effect on total investments, hence provide evidence for crowding-in effects (Borensztecrowding-in et al., 1998; Ramirez, 2006).

accumulation of foreign capital has a negative effect on economic growth in the long-run. As Bornschier (1979) argues, a high penetration of MNEs leads to ‘decapitalization’ in developing countries, which means a reduction of capital available to invest in the host economy. The reasoning is the following. Foreign investments absorb much of the host country’s capital in some advanced branches of the economy and therewith reduce capital formation in other sectors. The resulting industrial concentration leads to monopolization and high profits which are, due to limited possibilities of industrial expansion in the host country, invested in economic activities in other countries instead of being reinvested in the host country. Moreover, in addition to the usual outflows like dividends, royalties, license and management fees, the repatriation of profits is encouraged as high profits by MNEs are not appreciated and governments impose transfer restrictions. Consequently, high MNE-penetration is likely to have a negative effect on investment growth in the long-run and therewith on economic growth. Bornschier (1979) shows, in a cross-national analysis of 90 countries, that MNE-penetration in fact lowers economic growth in the long-run, partly through direct effects and partly by reducing investment growth.

pricing do not belong to the standard practices of MNEs and point to the necessity of supranational regulation of tax systems and monitoring. Finally, developing countries do not have the choice between foreign or domestic capital but rather between foreign or no capital (Dixon & Boswell, 1996).

In a more recent study Kentor and Boswell (2003) define a new concept of foreign capital dependence, namely foreign capital concentration. They argue that a high foreign capital concentration, which is the proportion of FDI stocks owned by the single largest investing country, affects the autonomy of the host country in implementing favourable policies. The study reveals that foreign capital penetration and concentration had a significant negative effect on economic growth from 1970-1985. However, after globalization policies in the 1980s only the concentration of foreign capital continues to have a significant effect. This finding suggests that the structure of FDI has become more essential for economic development than the overall level of investments, and again stresses the importance of industrial policies.

In conclusion, countries should not focus exclusively on the attraction of FDI since its benefits to the host economy are not self-evident. In order to reap the benefits of foreign capital flows, policies should also aim at improving local conditions and attracting the appropriate type of investment in sectors of the industry where it creates beneficial effects. A good understanding of the channels through which FDI might enhance technological upgrading and affect economic growth is important in order to adapt adequate policies to benefit from foreign investments.

2.2 Different types of FDI

production process is geographically fragmented in the way that different stages of the production take place in different countries. The motivation behind this sort of investment is that the firm can take advantage of lower factor costs in other countries.

Despite their different motivations, both types of FDI have similar potential effects on the host economy. They both include direct technology transfers to the local affiliate and have the potential to increase productivity through linkages and spillovers. Yet, only HFDI has an effect on product market competition since VFDI does not aim at selling the output on the local market but rather exporting it to other production stages in different countries. As described above, increased competition in the product market may lead not only to crowding out of national firms, but also to increased productivity in local firms in order to withstand the multinationals entrance. These competition effects can only arise from HFDI.

In sum, the likelihood that FDI has positive effects on the host country’s economic growth depends on the technological level of the domestic industry compared to the MNE, the extent to which MNEs build vertical linkages to local firms as well as the type of FDI. Policies to attract beneficial FDI have to take into account country characteristics as well as the various effects that the different types of FDI might have on the economy.

2.3 Policies vs. FDI

Different types of FDI do not only have different implications for the host economy, they also follow different incentives since they have different motivations (Navaretti & Venables, 2004). While HFDI seeks large and growing markets with a high potential to make profits, VFDI is attracted by low factor costs that allow minimizing production costs. Some country characteristics have even contradicting implications for the two types of FDI. High trade costs, for instance, would increase the amount of HFDI as foreign firms would avoid trade costs by supplying the market directly through an affiliate. VFDI, on the other hand, would be discouraged by high trade costs as the output is not determined for the local market but is exported, and inputs are probably imported as well. As a consequence, VFDI only takes place if trade costs are not too high so that the company can still benefit from low factor costs.

Another relevant determinant of FDI is regional integration. While the increase in market size through regional integration probably attracts HFDI from other regions, this type of FDI is reduced among the integrated economies due to very low trade costs. In contrast, VFDI would be encouraged within the integrated region if there are countries with important differences in factor costs, which is the case in the area of the North American Free Trade Agreement (NAFTA). The European Union, by contrast, has been a region with relatively similar countries until the EU enlargement in 2004. Furthermore, beneficial effects from agglomeration might also play a role in attracting FDI. The proximity to other firms in an industrial cluster can provide for knowledge spillovers, and firms can benefit from specialized factor markets as well as already established customer and supplier networks and infrastructure.

changed by any kind of policies; for instance, market size and factor costs. Nevertheless, if a country meets the necessary preconditions for an investment, policies that create incentives may have an influence on the decision where to locate in case there is a number of similar countries. Tax policies and trade policies may play a decisive role in attracting investments. Furthermore, FDI incentives include tax reliefs, exemptions from import or export tariffs, direct government subsidies to finance the investment or subsidized infrastructure and services (Navaretti & Venables, 2004). However, the general economic and political environment is of great importance, too. The success of any of the policies in attracting FDI and, moreover, creating beneficial effects for the host economy depends on how they fit the country’s characteristics. In the next section the implementation of different policies towards FDI in the Asian Miracle Economies is described in order to give an example of how policies can contribute to create beneficial FDI effects.

2.3.1 FDI policies in the Asian Miracle Economies

industrialization in these economies, usually referred to as the Asian tigers, was followed by further developments in the South East Asian economies of Malaysia, Indonesia, Thailand and the Philippines, experiencing rapid growth and economic transformation since the mid 1980s.

Some of the common characteristics in the strategies of the Asian miracle economies are described by Stiglitz (1996) who mentions macroeconomic and political stability as a requisite to create a good business climate as well as flexible government policies that aim at complementing markets rather than replacing them. The governments in East Asia created, for instance, capital markets and institutions that guaranteed effective investments to promote economic growth. A high emphasise on education and the provision of good infrastructure rendered private investments more profitable and also facilitated technology transfer and development. Moreover, governments supported industries with good long-term growth prospects and high returns on Research and Development (R&D) while at the same time encouraging exports and competition (Stiglitz, 1996).

However, there are several differences in the strategies pursued by the Asian miracle countries to achieve rapid technological change and economic growth. While Hong Kong did not see much government intervention, Japan, Singapore, Taiwan and Korea used a wide range of policy measures to support the domestic industry. These policies included the regulation of technology licensing and FDI so that national enterprises could benefit from spillovers and successfully absorb foreign technology (Chang, 2006). Furthermore, Singapore, a relatively small city-state at the beginning, strongly encouraged foreign investments, contrary to the infant industry protection strategy followed in the other countries. This strategy had also been followed later on by the new Asian tigers.

2.3.1.1 FDI dependent strategies: Singapore and the new tigers

know-how and skills as well as institutions and policies are needed (Lall, 2002). Hence, in order to encourage MNEs to upgrade their technologies, the government had to play a vital role in providing specialized training and education, increasing national supplier capabilities and developing more advanced services offered by technological institutions.

Furthermore, it is crucial for developing countries to encourage national R&D activities in order to build a comparative advantage and become internationally competitive (Lall, 2002). By supporting education and training as well as infrastructure in certain areas, and offering financial incentives, the Singaporean government channelled FDI flows into sectors with a high growth and spillover potential. Linkages with local suppliers were encouraged through economic development agencies that assisted local firms in building the necessary capabilities (Coe & Perry, 2004). These programmes provided financial incentives for MNEs that voluntarily helped to upgrade local supplier’s capabilities. As a result, incentives to channel FDI and promote technology transfer played the most important role in order to build national capabilities. These capabilities, in turn, contributed to encourage the upgrading of imported technologies and achieve international competitiveness. In important sectors of the industry state-owned enterprises (SOEs) were set up instead of attracting MNEs, so that all large firms today are either multinationals or SOEs, like the well-known Singapore Airlines (Chang, 2006).

shares for acquisitions of local companies. Furthermore, they used negative lists to provide an overview of sectors in which FDI is prohibited or in which exist foreign equity limits. Other measures to control FDI inflows include restrictions on the right to own land and to bring in expatriate staff. Export-oriented foreign investments, by contrary, were promoted by means of automatic approvals, the right to own land and other exemptions from the above mentioned restrictions. These privileges were granted to foreign enterprises that were export oriented as well as those that provided technology transfers (Thomsen, 1999). Nevertheless, domestic capabilities remained poor.

Obviously, by offering cheap labour and promoting export-oriented FDI the new tigers attracted predominantly vertical FDI, which possesses little incentives to upgrade imported technologies. At the same time, restrictions on foreign investments and compulsory technology transfers might have discouraged technology intensive investments into the region. Consequently, the countries’ technology levels remained relatively low. Moreover, a major problem in the East Asian countries was that only weak linkages have been built between the multinationals and the local economy. Imports made up to 90% of the value of exports in 1994 reflecting the isolated existence of the foreign industry (Lall, 2002). As a consequence, industrial upgrading through technology transfer has been very limited. One explanation is that export-oriented multinationals often purchase their inputs from other foreign firms since they want to produce at international competitive prices. Thomsen (1999) also suggests that the policies designed to increase the cooperation between foreign affiliates and local firms, such as joint ventures and local content requirements, might have had an adverse effect on the technology transfer from foreign investments. Technologies seem to be transferred faster to wholly owned subsidiaries than to joint ventures. Finally, government policies aimed at building new sectors by means of foreign investments instead of attracting FDI in sectors in which domestic capabilities already existed (Thomsen, 1999).

attracted predominantly low-technology and VFDI and, in addition, did not promote linkages and domestic capabilities. Given the increased competition from low-wage countries they face the risk of loosing their foreign high-value technological activities and therewith their international competitiveness (Lall, 2002). Thus, the very attraction of FDI is not sufficient to promote sustained economic growth; governments have to ensure that foreign investments support the technological progress and build national capabilities. For that purpose policies should help to attract a favourable sort of FDI in sectors with already existing capabilities and promote linkages to transfer technology in order to prevent the development of foreign-dominated industrial enclaves.

2.3.1.2 Autonomous strategies: Taiwan and Korea

Instead of relying on FDI in order to promote technological change and economic development some other Asian countries built a strong domestic industry under the protection of tariffs and other trade protections. Japan, often taken as a model of economic development in the East Asian economies, used subsidies for exports, investments, R&D and infrastructure to develop some key sectors while strongly protecting its infant industry, just like other developed countries as the United States, Germany and Great Britain did (Chang, 2002). Policies used in Korea and Taiwan, as part of the Asian miracle economies, were similar to those in Japan and give examples of autonomous strategies to ensure sustained economic development and growth. Contrary to the other Asian tigers the policies of these two countries were very hostile towards FDI; the local market was highly protected by means of tariffs and quantitative measures. Nevertheless, they achieved the building of domestic capabilities making use of imported technology (Lall, 2002).

Hyundai and Samsung. These emerging chaebols1 were given high incentives to export and compete in the domestic market and the government put pressure on them to set up supplier networks (Lall, 2002). The study by Cyhn (2001) suggests that OEM contracts constituted the major instrument to learn from foreign technologies, rather than FDI and licensing. Under OEM contractors domestic suppliers were provided significant assistance in producing specialized goods for a multinational that were then sold under the MNE’s brand name. By this means, domestic capabilities were built and domestic firms were given export opportunities that they otherwise would have never had due to their small size and low experience as well as their distance to the markets. In 1990 about 70 to 80% of Korean exports in the electronic industry were through OEM contracts. In addition to the interest of MNEs in their suppliers’ technological competency, OEM contracts had the advantage that the local suppliers could save marketing and distribution costs which could in turn be invested in their technological capacity (Cyhn, 2000). Consequently, MNEs seem to have played a major role in enhancing technological progress in Korea through the development of domestic capabilities; even though this was achieved by the use of OEM contracts rather than FDI.

Taiwan’s industrial policies were similar to those in Korea in terms of subsidies to the local economy and trade protection. However, it promoted the upgrading of technological capabilities in small and medium-sized firms (SMEs) in contrast to the large corporations in Korea. As a consequence of the absence of large enterprises, the Taiwanese government had to be more active in developing new technologies since small companies often lack the financial resources. As a consequence, more than 60% of R&D expenditure was afforded by the government rather than by the private sector. Taiwan also had to be more open to foreign investments for this reason; policies aimed at attracting FDI in weak sectors of the industry and ensuring technology transfers as well as supporting SMEs to access and adapt the new technologies. At the beginning of the millennium, Taiwan’s spending on R&D activity was second highest in the developing world (Lall, 2002; Chang, 2006).

1

2.3.2 Policies to assure positive FDI effects

Policies that are implemented in order to assure the positive effects of FDI on the host economy with respect to enhanced economic growth through technological progress should aim at attracting the appropriate type of FDI in the right sector and promoting linkages to local firms in order to assure the building of national capabilities and encourage the upgrading of imported technologies. The ways the Asian countries achieved positive effects on their economy are very different; nevertheless, there are some general inferences that can be drawn from their experience. Figure 2 shows what kinds of policies worked in which way to achieve technological progress and, hence, economic development.

Figure 2. Policies to ensure positive effects of FDI (source: own figure)

First of all, in order to attract the sort of FDI that complements the existing local industry, specific incentives, like the support of education and infrastructure in certain sectors as well as financial incentives, seem be more effective than restrictions on FDI. Furthermore, the promotion of linkages is vital for local firms in order to benefit from technology transfers. This can be achieved through government programmes that offer incentives for MNEs to

Policies

Incentives to build linkages and

transfer technology. The promotion of linkages by mandatory technology transfer provisions seems to be less effective since it may prevent high technology activities to locate in the country. In general, the support of domestic industries is important in order to build national capabilities, which increase the potential of technology spillovers and encourage technological upgrading. The development of national capabilities facilitates linkages and enables local firms to take advantage of the imported technology. At the same time, the development of national capabilities increases competition and puts pressure on MNEs to upgrade their imported technologies. Furthermore, increased national capabilities decrease the probability that domestic firms are crowded out by FDI. In sum, the building of national capabilities with the aid of policies that attract the appropriate sort of FDI in the right sectors and create incentives to transfer technology is central for developing countries in order to reach sustained economic growth through technological upgrading.

In addition, the experiences of Korea, Taiwan and Japan suggest that the protection of infant industries is important to develop strong national firms that can compete with multinationals. Even though Singapore seems to be the perfect example for liberal FDI strategies to benefit a developing countries economy, in fact, the domestic industry did not benefit much from FDI in the beginning. Coe and Perry (2004) argue that the relatively low technology level compared to Korea and Taiwan is a result of the crowding out of weak domestic firms in the manufacturing sector when MNEs entered the market. Consequently, subsequent upgrading policies did not have the same success they could have had in the absence of crowding out effects. Thus, the building of national capabilities while protecting infant industries from competition with MNEs seems to be reasonable. Korea’s experience even demonstrates that taking advantage of multinationals’ technologies is possible without allowing for FDI. OEM contracts can be used to plug into their international production networks and get access to technology and knowledge while reverse engineering and adaptation make the development of national capabilities possible.

in more advanced countries. As a result, foreign firms dominate the most dynamic sectors of the industry. The main problem is that these MNEs are not linked to the local industry; thus, generating no spillovers to local firms that would increase national capabilities. Furthermore, exports are based primarily on natural resources and primary commodities, the slowest growing industrial segment, while capital and labour intensive goods are imported. These are possible explanations why growth rates in the LA countries could not reach the level of the Asian countries.

The reason why other developing countries did not follow the examples of active governments in the East Asian countries, and reach comparable levels of economic growth, might be that international regulations rendered many of the described policies illegal since the beginning of the new globalization era. Multilateral agreements and treaties push for ever greater liberalization while limiting the options for developing country governments to support the domestic industry and ensure positive effects of foreign investment flows. As Wade (2003) claims, this ‘shrinking of development space’ is lead by developed country governments that try to kick away the ladder by which they climbed up their self in order to lock in their position at the top. Multinational companies from developed countries are given ever more possibilities to enter developing country markets without any restrictions that would ensure benefits to the local economy. In the following section, the changes in international regulations regarding FDI since the early 1990s are explained to show how the possibilities of developing countries to take advantage of FDI changed since the rise of the Asian Miracle Economies.

2.4 Changes in International Investment Policies

As one of the most important international organizations the World Trade Organization passes multilateral agreements that oblige governments “to keep their trade policies within agreed limits to everybody’s benefit”2. When, in the early 1990s, an increasing portion of international trade began to take place within international production systems of MNEs, the

2

agreements began to cover regulations on FDI. The three main agreements from the Uruguay Round in 1994 treat this issue and resulted in a considerably constraint of the possibilities that governments have to regulate foreign investments. As a result, today’s developing countries are prevented from making use of certain industrial policies that helped former developing countries to facilitate technological upgrading and economic growth.

First, the Agreement on Trade-related Aspects of Intellectual Property Rights (TRIPS) requires the introduction of minimum standards for intellectual property rights. As a consequence, technology transfers from MNEs to the host economy have become much more limited since strategies like imitation and reverse engineering, which were used extensively in some Asian economies, have been prohibited. Moreover, patent and copyrights raise the costs of accessing knowledge and therewith help to lock in the position of the developing countries at the bottom of the technological ladder (Wade, 2003).

Second, the Agreement on Trade-related Investment Measures (TRIMS) prohibits performance requirements on foreign firms since they are interpreted as distortions of trade and investment. These include local content and other requirements that force multinationals to purchase inputs from local suppliers as well as export requirements (Wade, 2003). Yet, as we have seen above, it is important to build vertical linkages between MNEs and the national industry in order to benefit from FDI, and local content requirements might have the potential to facilitate those linkages (Blomström et al., 2000). Export requirements also seem to be reasonable in light of the fact that the East Asian countries achieved rapid economic growth with the aid of rapid growing exports. Besides these explicit prohibitions, the fear of developing countries to be taken to the Dispute Settlement Mechanism (DSM) or to loose foreign aid prevents them from using other performance requirements with regard on joint ventures, technology transfer and R&D (Wade, 2003).

and forced out of the market when MNEs enter with liberalization, just like in the case of Latin America.

In general, liberalization and integration into the world economy seem to have become “a substitute for a development strategy” as Rodrik (2001) claims (Wade, 2003, p. 630) while possibilities for developing countries to take advantage of FDI with the aid of suitable policies have shrunk. In order to channel FDI and promote linkages with local firms the governments have to rely on general incentives, like the support of education and training in certain sectors. Any kind of restriction for FDI or protection of industries is not longer allowed. The example of Singapore suggests that such a FDI-welcoming strategy might have favourable effects; however, we also saw that there are problems if the existing industry is too weak. It has to be taken into account that Singapore is a city-state and therefore generalizations have to be handled with care. All the other Asian tiger states used some sort of restriction or protection to ensure positive or at least avoid negative effects of FDI on the local industry.

Moreover, not only the WTO promotes liberalization as the one and only remedy to achieve international competitiveness and economic growth. The International Financial Institutions also obligate developing countries to open their markets for international trade and investment through so called Structural Adjustment Programs (SAPs) (Welch & Oringer, 1998). During the 1980s and 90s SAPs have been imposed on many developing countries as a condition to aid agreements mandating policy changes that include decreased government spending, reduced tariffs and the liberalization of foreign investment regulations. The main criticism on these programs is that they do not promote sustainable economic growth since the general neoliberal principles only benefit a narrow elite in the private sector while contributing to the contraction of the domestic industry. Just like in the case of the WTO agreements the main benefits are supposed to accrue to MNEs from developed countries that get access to developing country markets.

1993, on the other hand, come more often to the conclusion that FDI does not have any spillover effects on domestic firms, or even negative productivity effects.3 Consequently, as the possibilities of developing countries to benefit from FDI have been reduced, I would expect that the potential of FDI to enhance economic growth in these countries has shrunk.

In the next section, I will describe a model that explains economic growth and includes technological progress as the main channel through which FDI might have an effect. Several studies found a significant positive effect of FDI on the host country’s growth rate when taking into account local conditions that influence the effect of FDI (Alfaro et al., 2002; Borensztein et al., 1998; Hermes & Lensink, 2003). However, the time periods examined reach from 1970 until 1995 only. Since most of the changes in economic policies with regard on foreign investments occurred since the mid 1990s, with the introduction of the WTO agreements, I would expect to find changes in the relation between FDI and economic growth for later points in time.

3

3. RESEARCH DESIGN

3.1 The Economic Model

The economic growth model used in this study is based on the new growth theory developed in the 1980s. This theory differs from neoclassical growth models and presumes technological progress to be endogenously determined. In addition, it includes human capital and mathematical explanations of technological progress to describe long-run economic growth. Admittedly, there are various models that try to explain economic growth. The model that is chosen here is described in Barro and Sala-i-Martin (1995) (chapter 6) and considers technological progress as an expansion in the number of varieties of producer and consumer products. A change in the number of varieties is regarded as a fundamental innovation, analogous to opening up a new industry. Hence, this approach seems to be appropriate in order to represent technological progress in developing countries. Technological advances due to quality improvements of existing products, on the contrary, may represent the upgrading process in more advanced economies. Borensztein et al. (1998) further develop the former framework and include FDI as the main channel of technological progress. The model can be summarized as follows. The production function takes the form

Y = A H α K 1-α (1)

where Y is output, A reflects exogenous influences on the productivity level, H is human capital and K denotes physical capital. In contrast to labour as a production factor, the concept of human capital also takes into account the quality of the labour force in terms of technological skills and knowledge. Physical capital consists of a variety of capital goods that are produced by domestic and foreign firms. If N indicates the total number of capital goods produced, then n gives the number of varieties produced by domestic firms and n* the number of varieties produced by foreign firms:

N = n + n* (2)

has to be adapted. This technology adaptation is supposed to be costly and, therefore, Borensztein et al. (1998) assume a fixed setup cost (F) before a new variety can be produced. These costs are smaller, the higher the proportion of products produced by foreign firms in the economy (n*/N), because MNEs provide knowledge that facilitates the adoption of advanced technologies. Consequently, FDI might accelerate the adoption of new technologies, therewith increase the country’s capital accumulation and enhance economic growth. This effect depends on the ability of the country to make use of the MNEs knowledge as described above. Furthermore, the setup costs are supposed to be smaller, the smaller the number of capital varieties in the domestic economy compared to the number of varieties in more advanced countries (N/N*). This indicates that for more backward countries (lower N/N*) the imitation of products is cheaper than the innovation of new products. Thus, the smaller the number of varieties produced domestically, the higher imitation possibilities are and the smaller setup costs to introduce new varieties are. This relation is also known as the convergence or ‘catch-up’ hypothesis stating that more backward countries tend to grow faster since the costs of adopting new technologies are lower. However, technological backwardness can also hamper countries from benefiting from FDI if the technology gap is too big. As explained above, there exists a minimum threshold of human capital and development in order to benefit from FDI. As a result, setup costs F can be written as

F = F (n*/N, N/N*), where δF/ δ(n*/N) < 0 and δF/ δ(N/N*) > 0 (3)

The smaller the costs of introducing new capital goods, the faster the adoption of new capital goods happens. Consequently, the growth rate of output can be expressed as a function of the setup costs (F), human capital (H) and a number of other variables affecting economic growth (A):

g = f (A, F(n*/N, N/N*), H) (4)

transactions, are considered to be direct investment. Furthermore, FDI is recorded in terms of assets (for the economy of the direct investor) and liabilities (for the economy of the direct investment enterprise) in the Balance of Payments statistic. Hence, a country’s net FDI stock is the difference between the stock of FDI assets and liabilities reflecting the value of accumulated stocks of foreign-owned assets. If we consider FDI stocks as a measure of the fraction of capital goods produced by foreign firms, (n*/N) captures the potential positive effect of FDI on technological progress. Moreover, the catch-up effect of more backward countries (N/N*) is captured by an initial GDP variable. A lower level of GDP is associated with a greater difference in income levels compared to advanced economies, thus, a lower initial GDP level reflects a higher income gap.4 The higher the income gap, the higher imitation possibilities are and, consequently, the higher the effect of FDI on economic growth is. This is captured by an interaction variable between the initial GDP level and FDI. Moreover, the level of human capital and the development of the financial market are expected to influence the effectiveness of FDI on economic growth, as explained in section 2.1.2. Consequently, the interaction terms with these variables are included as well.

The equation to be estimated is basically the following:

where GDPGR is the growth rate of real per capita GDP in country i at time t. FDI gives the amount of foreign direct investment, GDPIN gives the country’s initial per capita income, SCHOOL is a measure of educational attainment in order to capture the quality of the human capital stock and PRIVCRED measures the size of the financial market. Human capital is also assumed to increase the speed of convergence, thus, the effect of the initial GDP level on economic growth. Barro and Sala-i-Martin (1995) include, therefore, the interaction term GDPIN*SCHOOL in their regression, however, this term is found to be insignificant by

4

Another possibility to capture the catch-up effect is to include the income gap measured by the initial GDP relative to the GDP in the United States at the same point in time (See e.g. Blomström et al., 1992)

GDPGRit = β0it + β1FDIit + β2GDPINit + β3(FDI*GDPIN)it + β4SCHOOLit

+ β5(FDI*SCHOOL)it + β6PRIVCRED it + β7(FDI*PRIVCRED) it

+ β8Cit + εit

Borensztein et al. (1998) and therefore not included in the equation. Finally, C captures a number of other control variables that were found to have a significant effect on economic growth in the relevant literature (Barro and Sala-i-Martin, 1995; Borensztein et al., 1998; Alfaro et al., 2002; Carkovic & Levine 2002; Hermes & Lensink 2003; Yang, 2008). These include the total population of a country, measures for government spending, gross domestic investment, macroeconomic stability, openness, the black market premium and population growth.

actually be detrimental to economic growth in less developed countries (Rodríguez & Rodrik, 2000). Therefore, I measure trade openness in terms of exports as a share of GDP, which are expected to increase economic growth. The black market premium on foreign exchange is a proxy for market distortions and therefore expected to decrease economic growth. Finally, population and population growth are controlled.

3.2 Hypotheses

Besides the control variables captured in the vector C, equation (5) comprises seven independent variables that are assumed to have the following effects on economic growth.

First, the country’s initial income per capita GDPIN is supposed to control for initial conditions and captures the income gap between developed and more backward countries. Hence, a higher value of GDPIN is expected to have a positive effect on the GDP growth rate of the country, i.e. β2 > 0. The interaction term FDI*GDPIN , on the other hand, is expected to

have a negative effect on GDP growth, since a lower GDP level is supposed to increase the effect of FDI on economic growth (catch-up hypothesis), β3 < 0. Third, the country’s human

capital stock SCHOOL reflects the capability to implement superior technology and is thus expected to have a positive effect on the GDP growth rate, i.e. β4 > 0. The interaction term

FDI*SCHOOL captures the assumption that a higher level of human capital increases the effect of FDI on economic growth. The variable is therefore expected to have a positive effect on GDP growth, i.e. β5 > 0, since a higher level of human capital facilitates the adoption of

new technologies brought in by MNEs. Next, the financial market variable PRIVCRED is supposed to have a positive effect on economic growth, β6 > 0 , since the availability of credits

to the private sectors facilitates economic growth. Furthermore, the possibility for local firms to benefit from FDI is higher with more advanced financial markets. Hence, the interaction term FDI*PRIVCRED is also expected to be positive, β7 > 0. Finally, the first independent

variable in the equation, namely FDI, is the crucial variable and of special interest in the analysis. In accordance with the above stated literature, I expect to find a positive effect of FDI on economic growth, i.e. β1 > 0; at least for the period before the introduction of more

As mentioned above, developing countries today do not have the same possibilities to take advantage of FDI as the former developers, in particular the Asian Miracle Economies. It has become much more difficult for developing countries’ governments to ensure that the entry of MNEs leads to technological upgrading of the domestic industry. Since technology transfer is supposed to be the main channel of FDI-led growth, it would be expected that the relationship between FDI and economic growth changes with the introduction of policy changes with regard on FDI. The most important policy changes in this respect have been promoted by the WTO and the IFIs, therefore I use entry dates into the WTO and the implementation of SAPs as well as other International Monetary Fund (IMF) programs as proxies for changes towards more liberal FDI policies.5

Before any policy changes with regard on FDI occurred the effect of FDI on economic growth is expected to be positive, i.e. β1 > 0, since governments had the possibility to

implement a wide range of policy instruments to take advantage of FDI inflows. After a policy change takes place, this relation is expected to be less positive because policy instruments to promote economic growth through FDI have been reduced to a minimum, thus β1 decreases.

Hence, the central hypothesis of the paper is the following.

The correlation between FDI and economic growth is less positive after a policy change.

In order to test this hypothesis, I will use two different methods. First, the inclusion of dummy variables indicating a policy change will be used to create interaction terms with FDI which are supposed to be significantly negative, since a policy change is expected to reduce the effect of FDI. Second, a shifting regression analysis is used to identify breakpoints in the relation between FDI and economic growth, i.e. relatively sudden alterations in the size and sign of the relationship. These breakpoints are expected to coincide with the policy changes indicated by WTO entry and implementation of IMF or World Bank (WB) programs. The methods are explained in detail in section 3.4.2.

5

3.3 Data

The sample used in this paper to explore the relation between FDI and economic growth over time covers 62 developing countries from 1970 to 1999. Unfortunately, for some variables data were not available after 1999; however, the data set includes the period from 1994-99 in which I expect the impact of FDI on economic growth to decline substantially, since the agreements from the Uruguay Round entered into force in 1994. The observation of changes due to the implementation of more liberal FDI policies with the introduction of WB and IMF programs is also allowed for since these already started in the 1980s. Furthermore, the sample consists of developing countries6 only, since the effects of FDI have been described with respect to their potential to enhance economic development in more backward countries. As Bloningen and Wang (2005) point out, the pooling of developed and developing countries leads to insignificant effects of FDI on economic growth, since the growth enhancing mechanisms are different for less developed countries. When pooling data for both groups of countries they find that the result of Borensztein et al. (1998), that FDI affects economic growth only if education levels are high enough, is no longer valid. This relationship does not exist in developed countries, where education levels are generally high. Thus, pooling data obscures the mechanisms through which FDI affects economic growth in developing countries. Finally, the availability of data reduced the sample size to a set of 62 countries for which data could be obtained for all variables used in the analysis.

Unfortunately, data on FDI stock are unavailable; therefore, FDI flows are aggregated over 5-year periods. This procedure is also followed by Borensztein et al. (1998) who use aggregates for the two decades 1970-79 and 1980-89. Since this would result in the loss of a large part of my data, I take the aggregate of the previous 5 years for every single year, starting in 1974. Thus, the value for FDI in 1974 is the aggregate of FDI flows from 1970-74, for 1975 the FDI value is the aggregate of FDI flows from 1971-75 and so forth. A disadvantage of taking the aggregate of FDI flows is that the first year for which a five-year aggregate value is

6

available is 1974, since FDI flow data only start in 1970. Therewith, the analysed time period is reduced to 1974-99. Hence, the dataset consists of 62 countries that are observed over 26 years. Another data constraint consists in the fact that secondary school education is only available as 5-year averages; hence, in order to have annual data, the average values are taken for every single year in the 5-year period.7

The data on FDI flows come from Easterly (2001) who used Global Development Finance (GDF) and World Development Indicators (WDI) measuring the net inflows of investment as percentage of GDP. Furthermore, data on real GDP per capita and the GDP growth rate as well as data for government consumption and openness are obtained from the Penn World Table 6.2 (Heston & Summers, 2006).8 Initial GDP is included as lagged variable since the inclusion of an initial GDP value that does not change for the whole period could bias the results. Moreover, per capita GDP is measured in constant US dollars in order to adjust for variations in purchasing power over time. As a measure for educational attainment in SCHOOL the average years of secondary schooling for the population aged 15 and over is used, which is provided by Barro and Lee (2000). This group is more appropriate for developing countries than the alternative of the over-25 age group since educational levels are generally lower than in developed countries. Moreover, data for private credit come from Beck et al. (2000) and include private credit granted by deposit money banks and other financial institutions, also measured as a ratio to GDP. Finally, data on population growth and the black market premium come from Easterly (2001) whereas inflation rates and data on gross fixed capital formation are collected from WDI 2007.

In order to identify policy changes, there are several dummy variables that will be included in the analysis. The first variable is WTOMEMBER which is 1 if a country is a member of the WTO and 0 otherwise. The data come from the World Trade organization

7

Another option is to take 5-year averages for all variables, but this would result in a reduction to 6 observations for each country.

8

(n.d.). Furthermore, the variable WBADJUST indicates if a World Bank adjustment program was in effect for at least 5 months in a particular year, and IMFSAF is the same for Structural Adjustment Programs under the IMF. World Bank data are obtained from Boockmann and Dreher (2003) while data on IMF programs are provided by Dreher (2006). Besides the adjustment programs, variables for other programs that were conducted by the IMF are included, comprising Extended Fund Facility (IMFEFF) arrangements, Poverty Reduction and Growth Facility (IMFPRGF) arrangements and Standby Arrangements (IMFSAB). These data also come from Dreher (2006).

Since the adjustment program from the World Bank and the IMF, called Structural Adjustment Facility (SAF), obtained a fairly bad connotation, due to its consequences for developing countries as described in section 2.4, it was renamed Poverty Reduction and Growth Facility (PRGF) in 1999. Under this program countries develop Poverty Reduction Strategy Papers that are supported by the IFIs, but as a precondition they are still required to make changes in their economic policies as describes above. Hence, this program has the same implication for FDI-led growth as the SAF. Whereas the SAF and PRGF have the objective to overcome long-term economic growth difficulties, the Standby Arrangement (SBA) is designed to overcome short-term difficulties in the balance of payments. These programs last only one to two years, but they are also associated with the implementation of policy changes if a country wants to get access to IMF financing. An Extended Fund Facility (EFF) is basically the same as an SBA, with the difference that it addresses balance of payments difficulties stemming from structural problems and, therefore, need a longer period of adjustment (IMF, 2003). As a result, all the programs can be used equally in order to indicate a policy change towards more liberal FDI strategies. They all have the same implications for FDI-led growth. An overview of the definitions of all variables is given in table 3 in the appendix.

3.4 Methodology

3.4.1 Panel data

time. As can be seen in table 2 many studies that focus on the link between FDI and economic growth use panel data9; this happens because panel data have a series of advantages over time-series or cross-section data.

First of all, and most important, panel data might be more useful than cross-sectional data since it takes into account within-country differences in addition to between-country differences. Hence, panel data make it possible to control for unobserved individual heterogeneity (Brüderl, 2005). The rich information due to the combination of cross-section and time-series data also facilitates the reduction of omitted-variable bias since it allows controlling for omitted variables that are constant over time for a given individual but vary across individuals (or are constant across individuals at a certain point in time but vary through time). Furthermore, estimates will be more efficient as panel data provide a large number of observations and, hence, have more degrees of freedom and may better fit the informational requirements of the model. The greater variation in the explanatory variables also contributes to solve problems of collinearity which are often encountered in time-series analysis (Hsiao, 1986; Hill et al., 2001). Finally, panel data analysis gives information on intra-individual differences and changes trough time which allows detecting relationships that one would not be able to detect with a single cross-section or time-series data set.

However, limitations of panel data analysis arise with regard on the quality of the data Most of the times the data are not perfect, the analysis has to take into account attrition and sample selectivity in order not to impose arbitrary and false assumptions on the model (Hsiao, 1986).

3.4.2 Estimation techniques

With respect to econometric techniques used in the literature to estimate the effect of FDI on economic growth, table 2 shows that there are several different methods. The difficulty when combining time-series and cross-sectional data is that the intercept and response parameters vary across countries and over time. In order to make the model operational,

9

simplifying assumptions have to be made. There are several models that take this into account. First, the Seemingly Unrelated Regressions (SUR) model assumes that the parameters only vary across countries but are constant over time. By using information provided by the correlation between the errors terms, the SUR technique estimates the various equations for the different cross-section units jointly in a generalised least square framework. Hence, the SUR provides a possibility to pool the data while accounting for individual differences in behaviour. This technique has been used by Barro and Sala-i-Martin (1995) as well as Borensztein et al. (1998). However, the contemporaneous correlation of the error terms has been found to be small, therefore, the SUR estimates are only used to control the results obtained from estimation with instrumental variables.

Other techniques that take into account the panel structure are provided by the Fixed Effects (FE) model and the Random Effects (RE) model. In both models only the intercept parameter varies across countries (and/ or over time) and thus, captures all differences in behaviour, between countries and over time. The FE model has turned out to be useful in several studies (De Mello, 1999; Yang, 2008). De Mello (1999) acknowledges that the inclusion of country-fixed effects improves time-series analysis since it takes into account unobservable determinants of growth that are country-specific. However, FE models assume homogeneity of the different panel groups for a common slope to be imposed in pooled regressions, which can result in serially correlated disturbances in dynamic panels with a large time dimension.10 Inconsistent parameter estimates then tend to an overestimation of average short-run effects and underestimation of average long-run effects (De Mello, 1999). The introduction of regional dummy variables that divide the panel into relatively homogenous groups can reduce this bias. The RE model, on the other hand, assumes the intercept to be a random variable that consists of the population mean intercept and an unobservable country-specific random error, and therefore allows inferences from the sample that are representative for a larger population of countries. Moreover, the RE model has the advantage that time-invariant variables can be included among the regressors, however it requires the explanatory

10

variables to be uncorrelated with the country-specific error term. A Hausman test gives information about whether to use a FE or a RE model by testing the requirement for a RE model of no correlation between the regressors and the error term.

Almost all studies cited in table 2 use Instrumental Variables (IV) to control for endogeneity issues. Endogeneity occurs if an explanatory variable is correlated with the error term, which can occur if there is reverse causality between the dependent and an independent variable. Since some explanatory variables in the model, such as the magnitude of FDI, are likely to be influenced by higher growth rates, they may, in fact, cause endogeneity problems due to reverse causality. Instrumental variables are a common method to deal with endogeneity problems and get unbiased estimators. The assumptions are that the instrument correlates highly with the explanatory variable while it does not correlate with the error term. Yet, one should note that the latter assumption cannot be tested and, therefore, Brüderl (2005) argue that IV estimation may produce contradictory results since conclusions rely on untestable assumptions.

Finally, the Generalized-Method-of-Moments (GMM) is an estimation technique that uses instrumental variables and was developed for dynamic panel data. Carkovic and Levine (2002) use the GMM since it provides a panel estimator that controls for the endogeneity of all explanatory variables and, furthermore, accounts for the bias induced by the inclusion of initial GDP in the growth model. However, Brüderl (2005) argues that dynamic panel models, including a lagged dependent variable on the right-hand-side of the equation, complicate estimation since the estimation techniques described above cannot be used anymore. This is due to the correlation between the lagged dependent and the error term(s) that result in biased FE- and RE-estimators. In order no to complicate the estimation of the model, I will, therefore, start with simpler non-dynamic FE- and RE-estimation techniques. Endogeneity can be controlled for by means of instrumental variables. I also include the lagged value of GDP in the model, but this does not result in a dynamic panel model since the dependent variable is GDP growth. Hence, FE- and RE-estimators will not be biased.