RIJKSUNIVERSITEIT GRONINGEN
Faculty of EconomicsDepartment of International Economics and Business
Master’s Thesis
THE INFLUENCE OF INTERNATIONALIZATION ON THE
PERFORMANCE OF BRITISH SMALL AND MEDIUM-SIZED
ENTERPRISES
L.A.M. Niers 1279998
February 2007
ABSTRACT
TABLE OF CONTENTS
1. INTRODUCTION………...5
2. LITERATURE REVIEW AND HYPOTHESES………...………..8
2.1. Introduction to internationalization………...……...…...8
2.1.1. The first studies on internationalization………..……...………8
2.2. Literature on SMEs………...9
2.2.1. SMEs and internationalization………...………9
2.2.2. Push and pull forces………...10
2.2.3. SMEs versus MNEs………...12
2.2.4. Asset uniqueness………...13
2.2.5. Forms of Internationalization………...13
2.3. Derivation of the Hypotheses………..…...14
2.3.1. Exporting and performance………...………...14
2.3.2. FDI and performance………..………...15
2.3.3. Combining different forms of internationalization………..…………18
2.3.4. R&D and internationalization………..………...19
3. METHODOLOGY………...27
3.1. Data source………...27
3.2. Sample selection………...27
3.3. The choice for the United Kingdom………...28
3.4. Research methodology………...29
3.5. Variables………...29
3.5.1. Dependent variable………...29
3.5.2. Independent variables………...…...30
3.5.3. Control variables………...31
3.6. Description of the interaction effects………...………...32
4. EMPIRICAL FINDINGS………...……….……34
4.1. Descriptive statistics………...……….34
4.2. Results of the regression………...………..…….37
6. CONCLUSION………...……….……43
6.1. Limitations and suggestions for future research………..44
6.2. Contributions, managerial implication and acknowledgement ………...45
1.
INTRODUCTION
Nowadays, the world experiences rapid globalization (Fillis, 2001). Due to this phenomenon, many firms become involved in international activities. Most research that has been conducted so far on this topic applies to the internationalization of large firms (McDougall and Oviatt, 1996). There is little known about the effects of internationalization on performance of small and medium-sized enterprises (SMEs) (Lu and Beamish, 2001). SMEs can be defined as companies with fewer than 250 employees. Specifically, this definition applies to medium-sized companies. Firms with less than 50 employees belong to the small sized category, according to the European definition of SMEs (Storey, 1994). Worldwide, there exists no generally accepted definition of SMEs. However, the American Small Business Administration (SBA) also defines SMEs as stand-alone enterprises with fewer than 250 employees.
It is worth considering the link between internationalization and performance for smaller firms, because eventually managers of smaller firms are interested in the question whether foreign expansion may positively affect firm performance. Acquiring sales outside the domestic market has been an objective of many SMEs and their governments (Audretsch, Roy, Kwaak and Bosma, 2003). However, the process of internationalization involves a difficult step, especially for SMEs. Expanding abroad implies numerous risks, which can be detrimental for SMEs, because they are limited in their resources compared to larger companies. Many SMEs, which obtain good results locally, fail in foreign markets.
I address the research questions in this thesis on the relationship between various forms of internationalization and the performance of SMEs. Specifically, the central objective in this research is to get insight in a sample of British companies about their performance and their degree of internationalization and the link between those variables. This paper attempts to obtain answers on the following research questions: (1) What is the relationship between a SME’s performance and its export intensity? (2)What kind of relationship exists between the level of FDI and SME performance? (3) Does internationalization due to FDI enhance SME performance if FDI is initiated by a high level of exports? (4) Will a greater investment in R&D lead to better SME performance, if FDI is considered as a form of foreign expansion?
British government actively supports the expansion-strategy of SMEs globally (Spence and Crick, 2006). Finally, although there are some studies focusing on datasets of different European countries, there has not been much empirical evidence regarding the internationalization of British SMEs so far.
The contribution of this paper is the focus on the internationalization of SMEs and the application on a British sample. As mentioned above, most research on internationalization has been dedicated to the internationalization strategies of multinational enterprises (MNEs). Academics agree that the effects of foreign expansion on the profitability of SMEs have not been studied sufficiently (Coviello and McAuley, 1999). Further, the studies that address the internationalization of SMEs are often theoretical instead of empirical (Fillis, 2001). Two important studies, however, have empirically tested the international expansion and its relationship to performance of smaller firms (Nkongolo-Bakenda, Anderson, Ito and Garven, 2005). These are the studies of Lu and Beamish (2001) on a Japanese sample of SMEs and the one of Majocchi and Zucchella (2003) on a sample of small Italian firms.
Lu and Beamish (2001) have revealed a negative relationship between profitability and exports for a sample of Japanese SMEs. Further, they have discovered the existence of a liability of foreignness (Hymer, 1976) on this sample of SMEs at the beginning of the internationalization process via FDI.
Majocchi and Zucchella (2003) conclude for a sample of Italian SMEs that performance is not determined by export intensity and the number of international agreements, but by the ability of firms to gain access to specific markets. Moreover, they find that performance tends to suffer when SMEs internationalize through FDI, a finding that also suggests a liability of foreignness. However, this negative effect can be offset by the international experience that SMEs develop through intense export activity, according to Majocchi and Zucchella (2003).
The aim of this research is to contribute to a better understanding of the internationalization problem through an analysis of the different internationalization modes on a sample of British SMEs. A national dataset is used to compare the results for Great Britain with those obtained by similar studies of SMEs in other countries, such as Italy and Japan respectively (Majocchi and Zucchella, 2003; Lu and Beamish, 2001).
different national sample of SMEs is analyzed, this paper contributes arguments on the main sources of competitive advantage for SMEs and the best form of internationalization in this case.
2.
LITERATURE REVIEW AND HYPOTHESES
2.1. Introduction to internationalization
Internationalization can be viewed from different perspectives. It depends on the questions that are going to be answered (Dunning, 1988). Examples of such questions regarding internationalization may be “Why do firms engage in foreign direct investment?” Or “Why do firms locate their activities in one country rather than another?” Or “Why do firms want to control its activities abroad?”
From these different questions, it follows that there is no one single operationally testable theory of international production. Scholars simply differ in their unit of analysis, that is they differ in the variable to be explained (Dunning, 1988). Only correct answers to particular questions exist, which together form a broad concept to explain internationalization.
In the following section, the most famous theories of internationalization will be discussed. For a long time, economists have provided different theoretical explanations regarding the driving forces behind international trade, because different research question were formed (Muelbacher, Dahringer and Leihs, 1999; Dunning, 1988).
2.1.1. The first studies on internationalization
Historically, the first studies on internationalization originate to economic researchers as Adam Smith (Sutherland, 1993) and David Ricardo (Hollander, 1979). Smith emphasized the importance of scarce resources and implemented an efficient distribution system accordingly. In his famous writing, the Wealth of Nations, he described that each country possessed unique resources in the production of goods. According to this theory, each country specializes in products or services that it can produce comparatively better than others do. Then each country exchanges these products or services for those, which are more efficiently produced in other countries.
that they are economic in nature, whereas today’s theories on internationalization concentrate more on the behavioral theory of the firm (Leonidou, 1989).
Vernon (1966) initiated this change in the approach to internationalization, due to the development of the product life cycle theory. The basic intuition behind this theory is that for many manufactured goods, comparative advantage shifts from one country to another. As mentioned, his contribution was to use a microeconomic concept, the behavior of firms, to help explain a macroeconomic phenomenon, namely the foreign activities Multinational Enterprises (MNEs, hereafter) (Dunning, 1988).
Three important theories on internationalization, which contribute to form my hypotheses, emanated after the theoretical standards described above. The first was the contribution of Stephen Hymer (1976). His arguments help to explain my second hypothesis. The second theory that will be discussed is the stage theory of internationalization of Johanson and Vahlne (1977). This theory is also known as the Uppsala School of internationalization and will be the underlying theory of the third hypothesis in this thesis. The last theory that will be addressed is the framework of Dunning (1988). He attempted to incorporate all the internationalization theories in one eclectic paradigm. This paradigm, known as the OLI – framework, will contribute to my last hypothesis.
2.2. Literature on SMEs
The internationalization theories described in the former paragraph have failed to address adequately the internationalization path of SMEs (Fillis, 2001). Therefore, I will now concentrate on an internationalization theory that suits better to the characteristics and competencies of SMEs, such as the push and pull forces theory. Furthermore, the characteristics of SMEs and the implications for them regarding foreign expansion will be discussed.
2.2.1. SMEs and internationalization
value for the enterprises involved? This is an interesting question for researchers and a crucial one for managers. So far, no clear conclusions on this matter can be drawn.
The growing interest in the internationalization of SMEs has led to new research on this phenomenon. Two distinct streams have already emerged in international new ventures (Lu and Beamish, 2001). The first are start-ups that are international from the beginning. The other stream treats the SMEs that have already been involved in international operations. This study belongs to the latter group, which mean that I will analyze only firms that already have expanded due to a specific form of internationalization.
2.2.2. Push and pull forces
A theoretical explanation of internationalization that fits especially to SMEs is provided by the construct of push forces and pull forces of internationalization. Etemad (2004) developed this framework, as an attempt to organize the complex dynamics in a way to face the exceeding complexities of the SME’s internationalization process, in a rapidly evolving environment. The framework that Etemad proposed consists of three theoretical dimensions, with each incorporating a number of different influences. The three dimensions are the push factors, the pull factors and the mediating factors and will be discussed in the following paragraphs.
The push drivers consist of forces that are internal to the firm and exert pressure from inside to internationalize. These forces arise when the domestic market becomes too small for a product. This occurs often in case of a niche product or a highly specialized product or when the domestic market becomes too competitive. The push factors contribute by pushing or accelerating the firm’s internationalization process to exploit international opportunities, also when domestic market inertia may have been limiting SME’s efforts (Etemad, 2004). Some examples of push factors are discussed in the following paragraph.
as R&D, can be spread internationally (Coviello and McAuley, 1999). Finally, a broader package of and easier access to resources is available internationally (Etemad, 2004).
The pull drivers are forces that develop when for example foreign buyers make unsolicited demands for a product or when a major customer requests that. In other words, the pull factors are external to the firm. They provide attractive incentives for the firm to internationalize, for example by emphasizing the benefits of larger and richer international markets. In this way, SMEs that initially operated in domestic markets are pulled into international markets, where they may be positioned in an unfilled market segment. Consequently, the firm is better able to match its supplies to international market demands. Specific examples of pull forces are described below.
In the first place, an example of a pull factor is the liberalization of international markets, which facilitates exporting and other ways of internationalization (Acs, Morck Shaver and Young, 1997). Due to absence of tariffs and quotas and other trade barriers, a better access to a cheap and specialized labor force can be secured. Another component of a pull factor are the current advances in Information and Communication and Transportation Technology (ICTTs), which vaporizes the impact of time and distance, facilitates networks operations and reduces costs (Aggerwal, 1999). A third example of a pull factor is the easiness of attracting resources of partners, which vanishes the limitations of size, shortens the internationalization process and in this way allows for survival in highly competitive international markets. Finally, as discussed, the pull factors make sure that the international needs of current buyers and suppliers are served (D’Aveni, 1994).
According to Etemad (2004), it are not separately the push and the pull forces, which contribute the most to the internationalization process, but rather the interactive factors between them. It is this mediated impact that influences the true course of a firm’s internationalization. For example, Coviello and Munro (1995) describe how the mediated impact of New Zealand’s small market size, which pushes the firm out, in combination with the attraction of the larger surrounding liberalized international markets, which pulls the firm abroad, have contributed both to the internationalization of New Zealand’s firms.
behavioral theories of the firm, such as organizational learning, game theory and transaction costs theory (Williamson, 1975; 1981). In that respect, this framework differentiates itself from other international theories in the way that it fits the international expansion process of especially small firms well.
2.2.3. SMEs versus MNEs
This paper tries to explore the relationship between performance and internationalization for a sample of British SMEs. This relationship has been thoroughly researched in the literature with regard to large multinationals, but not with regard to SMEs.
SMEs cannot be considered smaller versions of larger enterprises: they fundamentally differ from MNEs. They differ in their resource endowments, organizational structures and processes as well as their management systems (Wyer and Smallbone, 1999). In addition, their strategic position and their internationalization policies differ substantially from those of the large firms. Therefore, the results of the research concerning multinationals cannot be automatically applied to smaller companies. SMEs may be even further differentiated from larger companies in terms of internationalization, next to the previous managerial, financial and operating characteristics. Examples are the personalized objectives of owners and the informal planning and control systems (Baird, Lyles and Orris, 1994).
Papadopoulos (1987) discusses firm characteristics unique to internationalized SMEs compared to larger firms. Weaknesses of SMEs include the lack of ability to grow and to gain expertise in the domestic market and the difficulty to find expansions abroad. Moreover, he argues that in contrast to larger MNEs, the SME has a limited range of international entry modes to choose from and a narrower international base from which international activities can be undertaken. This argument can be explained because financial constraints require a quick need for positive return on investment and in this line of reasoning few SMEs can afford to enter markets, which are so competitive that survival requires a long-term orientation. In other words, SMEs have a lower tolerance for temporal crisis (Castrogiovanni, 1996).
internationalization. This has been further researched by Hollander (1970) and Williams (1991) who found that there is no one to one functional relationship between the size of the firm and its level of internationalization. In addition, it has been discovered that a successful exporter does not have to be a big exporter (Reuber and Fisher, 1997; Coviello and McAuley, 1999).
It is argued that SMEs can compete successfully in the international market, specifically if they possess distinctive competitive advantages compared to their larger counterparts. These asset-based advantages play a critical role in the decision to internationalize (Hutchinson, Quinn and Alexander, 2005). Therefore, I will turn to this argument in the next paragraph.
2.2.4. Asset uniqueness
The key factor for successful internationalization strategies and competitiveness for SMEs from a resource-based point of view is the possession of unique assets (Barney, 1991). In this way SMEs have traditionally been thought as resource-constrained, because they are small compared to large multinationals. According to Barney, firms and especially SMEs need their resources to be unique in some way in order to gain a sustainable competitive advantage. In an international environment, when SMEs are forced to compete with MNEs and globally competitive firms, such a characteristic seems even more relevant. SMEs need to develop their own distinctive competencies to empower equally distinctive strategies that can lead to a stronger competitive position that enables to compete against other companies regardless of size (Etemad, 2004). Uniqueness, which can be reflected in assets or knowledge, should result in higher profitability, which in turn feeds additional international commitment and knowledge improvement via international learning processes (Kuemmerle, 2002).
Critical strategic resources are particularly reflected in intangible assets that are sources for sustainable competitive advantage for SMEs. These resources are extremely important for foreign SME-survival in order to build local competitiveness abroad (Buckley, 1988). Examples of intangible assets include research and development (R&D), technological know-how, reputation, patents and management skills.
2.2.5. Forms of internationalization
(Majocchi and Zucchella, 2003). For this reason, this paper differentiates between those two international avenues. Exporting and FDI will be separately linked to performance and further the joint effects of these two internationalization strategies will be examined.
The contribution of the various internationalization strategies to value creation has been the object of a number of empirical studies. The international business literature has focused mainly on the performance implications of exporting and FDI (Wolf, 1975). Although it is possible to argue that higher levels of international involvement would lead to better performance, the results of empirical studies do not reach definitive conclusions (Ramaswamy, 1992).
For example, regarding exporting, Majocchi and Zucchella (2003) do not find any support for their hypothesis that SME performance is positively related to export intensity, whereas Lu and Beamish discover a negative relationship between the performance of smaller firms and their exports. Considering these varying findings, it is interesting to re-examine this relationship once again for a new sample of British SMEs. This issue will be addressed in the subsequent paragraph.
2.3. Derivation of the hypotheses
In this section, I will derive my hypotheses. The hypotheses will explore the effect of different internationalization modes on SME performance.
2.3.1. Exporting and performance
The general aim of this research is to find out whether internationalization can be regarded as a crucial factor in explaining SME performance. In effect, I differentiate between two kinds of internationalization: exporting and internationalization through FDI, which will be addressed in the next section.
Earlier in this thesis, the classical theories argued that internationalization in the long run would be beneficial. Exporting is the first step to internationalization, serving as a platform for future international expansions. Exporting is the lowest and most flexible level of commitment to international marketing. This approach entails minimum efforts and costs, compared to other forms of internationalization (Dibb, Simkin, Pride and Ferrel, 2001).
with fast access to foreign markets, with little capital investment required but at the same time, SMEs have the opportunity to gain valuable international experience.
A couple of economic benefits can be gained by exporting. Advantages such as economies of scale and scope can be the result of exporting. These economies come from larger sales volumes and extended production in a broader market. Furthermore, a presence in international markets can lead to increased market power and gains from the diversification of revenues (Ramaswamy, 1992).
These potential advantages of exporting suggest that exporting is positively related to the financial performance of SMEs. Hence, the first hypothesis that will be tested, states:
H1: A positive relationship exists between a SME’s performance and the level of export intensity.
2.3.2. FDI and performance
FDI, is the other form of internationalization. FDI can be defined as the creation of wholly owned subsidiaries abroad. The benefits of this form of internationalization are location-based advantages, such as the access to a cheap labor force and internalization advantages, which means benefiting from an internal market. Because of these advantages, FDI has frequently been proven to be associated with higher firm performance. Explanations for this accrue from spreading overheads over more nations, especially in R&D intensive industries (Kobrin, 1991; Talmann and Li, 1996). Moreover, greater organizational learning (Kobrin, 1991) and access to cheaper resources in foreign countries contribute to better performance (Porter, 1991; Jung, 1991). Finally, cross-subsidization, price discrimination and arbitrage potential are easier accomplished with a larger geographic scope.
Although FDI certainly has specific advantages, it also has some negative aspects that should be considered. For example, more financial and managerial resources have to be invested abroad compared to a simple export-strategy. Further, once investments in FDI have been taken place, these are difficult to reverse. Therefore, FDI can be considered less flexible compared to exporting, for example when political instability and fluctuating market conditions in host countries are experienced.
globally orientated companies. Empirical investigation has shown different relationships between these variables and there is a large empirical discrepancy between the sign and the shape of performance and internationalization in the form of FDI (Contractor, Kundu and Hsu, 2003).
Early studies addressing the internationalization expansion debate, show a negative linear result between performance and internationalization (Brewer, 1981; Collins, 1990). These inquiries have produced inconsistent and contradictory findings (Ruigrok and Wagner, 2003). Researchers agree nowadays that the omission of internationalization costs in these inquiries merely explain the ambiguous results (Gomes and Ramaswamy, 1999; Hitt, Hoskisson and Kim, 1997; Sullivan, 1994).
Therefore, scholars started to discuss both the advantages and the disadvantages of internationalization of firms investing in FDI, leading to non-linear results. Some examples of these studies show only positive performance beyond certain levels of foreign investment (Geringer and Hebert, 1989; Hitt et al., 1997). The threshold in this relationship dates back to Hymer (1976). The main observation Hymer (1976) wanted to explain was the fact that foreign firms wish to control operations outside their national boundaries. He expressed his dissatisfaction with the classical theories of indirect capital transfers to explain the foreign operations of firms (Dunning, 1988).
In the first place, he acknowledged the existence of market imperfections such as risk, uncertainty, volatile exchange rates and transaction costs. If these imperfections were included in the classical theory, much of its predictions would be wrong, such as the flow of capital in response to interest rates movements. These imperfections changed the way how firms managed their foreign operations, according to Hymer.
Secondly, Hymer brought forward the suggestion that firms not only transferred financial capital abroad, but a whole package of resources, such as management skills and entrepreneurship. Firms were trying to optimize the economic rent on the total package of these resources and the way they were organized.
Because of these improved assumptions, Hymer argued that firms, which successfully want to own foreign affiliates, must have some kind of advantage compared to domestic firms. This advantage should be sufficient to outweigh the disadvantage of competing with indigenous foreign firms in the country of production. Examples of disadvantages for firms considering FDI are language barriers, lack of knowledge of the local economy and cultural differences. Furthermore, foreign firms may experience discrimination by foreign governments, foreign consumers and foreign suppliers. Finally, foreign firms are disadvantaged, because they are exposed to foreign exchange rate risk (Hymer, 1976).
This disadvantage for foreign firms was named the liability of foreignness, which means that foreign firms face high fixed costs in the form of fixed investments and experience gathering. In other words, the liability of foreignness means that global entrepreneurs suffer from higher initial costs than local competitors do.
More recently, scholars have found empirical evidence for another relationship between FDI and performance, the sideways “S” hypothesis, also called the three-stage theory of international expansion (Contractor et al., 2003). The first half of the sideways “S” represents the liability of foreignness, where performance is declining because of international learning costs. Later on, performance will increase if ownership advantages are exploited internationally (Tallman and Li, 1996). Finally, performance will fall again if coordination costs outweigh the benefits of internationalization (Hitt et al., 1997). However, this theory applies better to larger, well diversified international firms, because smaller firms often have not reached the stage of global coordination problems, because they simply do not have the resources for that (Lu and Beamish, 2001). Since this research focuses on small firms, Hymer’s liability of foreignness seems still best applicable to the relationship between FDI and performance for SMEs. I therefore hypothesize that SMEs experience a negative slope between performance and low levels of FDI, but a positive slope at higher levels of FDI:
2.3.3. Combining different forms of internationalization
Further, I want to research the link between export-activities and FDI-investment. According to the Uppsala Theory of Internationalization, (Johanson and Vahlne, 1977; 1990) firms gain experiential knowledge in stages. This step-by-step approach states that firms accumulate knowledge as a basis for entering foreign markets. Firms are presumed to start exporting, the first stage of internationalization, by targeting physically close countries. Later on, due to the accumulation of experience, firms can commit greater resources to countries that are more physically distant. In the second stage of internationalization, when more confidence in the internationalization-process is obtained, firms start to establish sales subsidiaries abroad, which mean a form of FDI. The last stage of this internationalization theory is overseas production in self-owned production facilities (Johanson and Vahlne, 1977). In this way, the firm will be better equipped to overcome the hurdles of internationalization, because they have already gained international experience due to its exports activities (Johanson and Vahlne, 1977).
This theory is in line with other internationalization theory that views foreign expansion as the result of the incremental decision that gradually lead to the involvement of international operations (Kogut and Chang, 1991). Pennings, Barkema and Douma (1994) view the firm’s expansion as part of a process of organizational learning, in which organizations evolve as they accumulate experience. The implication of this is that the firm’s age and prior experience regarding internationalization have a positive impact on performance.
This argument is recently confirmed by Spence and Crick (2006), who in their research are comparing British and Canadian firms. They argue that most firms in both countries took a relatively cautious route towards the different modes of market entry. This means that initial expansion abroad occurred via agents or distributors. The availability of financial resources played a major part in decision-making, according to Spence and Crick (2006).
(1987), who argued that the major tenets of the stage theory of Johanson and Vahlne are only derived from two empirical studies.
Because of the different arguments that can be brought forward in favor and against the stage theory of internationalization and the lack of empirical evidence proving this theory, scholars researching this phenomenon have re-initiated studies on this topic. Examples of these are authors like Lu and Beamish (2001) and Majocchi and Zucchella (2003), who focus on small firms and research the relationship between exporting, FDI and performance, the variables that conceptualize the stage theory. Surprisingly, also their findings are contradicting. Whereas Majocchi and Zucchella (2003) find empirical evidence for the stage theory of internationalization for an Italian sample of SMEs, Lu and Beamish (2001) do not find a positive relationship of FDI on performance, when FDI is initiated by high exporting. Their finding disproves the theory of the Uppsala School of internationalization. This discrepancy in theoretical reasoning and the confusion in empirical evidence lead me to reconsider this relationship once again. I will therefore empirically test this relationship in the following hypothesis. Considering the fact that SMEs are likely to have little experience in internationalization, I argue that a high level of exporting prior to FDI expansion will benefit this strategy due to a better knowledge of the international area that SMEs have already developed through its export activities. I therefore propose the following hypothesis:
H3: The relationship between FDI and SME performance is positive, especially if FDI is initiated by a high level of exports.
2.3.4. R&D and internationalization
Johanson and Vahlne (1990) mention the increased impact of R&D on internationalization, but they do not explain to what extent this relationship influences the international activities of companies (Szabó, 2002). Therefore, I want to research the influence of R&D on the internationalization process of SMEs.
cultures. Firms willing to expand its activities abroad, must therefore have some specific advantages over domestic firms to hold a strong competitive position. Dunning suggested three conditions, which all have to be fulfilled to guarantee successful foreign direct investment. These conditions have become known as the OLI – framework, which stands for Ownership, Location and Internalization respectively.
The ownership advantage of a firm could be a product or production process, which is specific to the firm and to which other firms do not have access. Often, ownership advantages are intangible, such as patents, trademarks or reputation. The ownership advantage reflects some valuable market power or cost advantage on the firm, which should outweigh the disadvantage of doing business abroad.
The location advantages are often the intrinsic argument to go abroad. These advantages make sure that FDI will be preferred over an export-strategy, because tariffs and transport costs can be foregone when firms rely on FDI. The use of cheap factor prices, such as cheap labor, is a good example of a location advantage. Other location arguments may be the access to critical resources and favorable governmental conditions.
A third condition that should be met for successful foreign expansion, is the internalization advantage, according to Dunning. This is the not the most obvious one of the three conditions. If the first two conditions are fulfilled, for example because a company possesses distinctive proprietary assets and the transport costs are high, it is still not certainly derived that a foreign subsidiary could be advantageous. A company could still license a foreign company to produce its goods. Internalization advantages are likely to occur whenever the benefits of an in-house organization of transactions are likely to outweigh the possibilities offered by an external market (Dunning, 1988).
A first potential discrepancy between the firm and the external market, is that firms may not want to reveal its process or product technology. Often knowledge capital has the characteristic of a public good: it is non-excludable. An exporter therefore could, once the information is acquired, copy the technology and produce it on its own. On the other hand, an exporter is not going to accept a contract, without knowing the exact content of the deal, so this requires information from the seller-side. In this way, no deal can be reached between a seller and an external party and therefore a fully owned foreign subsidiary might be preferred.
A second problem involves an informational asymmetry between the firm and the external market, particularly in case of innovative and complex products. Although the firm knows its product the best, it may not want to reveal the product quality entirely. In that respect, it is almost an impossible task to write contracts between the firm and the potential exporter. Especially in this case it might be more beneficial to internalize the market rather than explore the market at arm’s length (Ethier, 1986).
Another information problem exists when the foreign party has superior information about its market, for example if this person knows beforehand whether the product will be a success. A foreign company needs this information as well and therefore might rely on the experience of an agent. However, the agent will probably not reveal the truth, because if he reveals sales to be high, the firm might decide to set up a foreign plant. Thus, in this line of reasoning, the agent’s incentives might to reflect that sales are low, even in case that foreign demand is high. This also enhances the chance for the foreign firm to opt for FDI as an expansion strategy (Horstmann and Markussen, 1995).
A fourth characteristic of knowledge capital is that it can be easily transferred. This characteristic might be beneficial, but it also means that employees and also foreign exporters can easily learn this knowledge. This increases the chance that foreign exporters might learn the technology and defect to start their own business. It is argued that especially foreign exporters form a danger for the producing firm in this case, because they do not have such a reliable firm commitment as employees have (Ethier and Markussen, 1993). This argument again contributes to prefer FDI to exporting for domestic firms.
knowledge is clearly difficult and costly to transfer to an external party. Also in this case, FDI should be the dominant strategy for companies with foreign aspirations.
A sixth problem in case of exporting arises, when the firm differentiates itself by means of product quality. Product quality is particularly important in the case of experience goods. Experience goods are goods that should first be experienced by customers, to discover the truth quality (Dibb et al., 2001). The problem comes up when the producer cannot extract all the rents from an exporter, because when it tries to do so, the exporter can offer a similar product of lower quality, while pretending that it has the same quality. This phenomenon often occurs in case of franchising, where each outlet manager has the incentive to free ride on the reputation of the whole (Caves and Murphy, 1976). In this case, it is again more beneficial to remain the product within the company and to sell and produce with a company-owned subsidiary.
Finally, the last characteristics of knowledge capital where firms must be aware of when doing transactions at arm’s length, is moral hazard. Moral hazard, is the situation where a person or organization does not bear the full adverse consequences of its actions (Brealey, Meyer and Marcus, 2001). For example, it may be the risk to an insurance company resulting from uncertainty about the honesty of the insured. For example, exporters may divert selling efforts to competing companies or simply shirk (Mathewson and Winter, 1985).
From these explanations of the characteristics of knowledge capital, it becomes clear that firms, relying on this type of capital, have an incentive to use internalization strategies in their foreign expansions.
However, there are some difficulties to apply the ownership paradigm of Dunning to smaller firms. This framework was initially intended to serve for MNEs. Therefore, one cannot automatically copy this framework to smaller firms. Smaller firms also possess ownerships advantages, but they are different compared to their larger counterparts. Ownership advantages for smaller firms are far more knowledge driven and competency-based (Storey, 1994). Typical distinctive competencies of SMEs are creativity and innovative thinking, opportunity recognition, risk taking ability and network and relationship building (Carson, Cronie, McGowan and Hill, 1995; Fillis and McAuley, 2000). Therefore, one could argue that internationalization through FDI is even more important for smaller firms relying on a large amount of R&D.
to sales (Morck and Yeung, 1991). Further, it was argued that firms relying on a high ratio of knowledge capital such as R&D, would prefer FDI as a foreign expansion mode, because in this way firms are creating an internal market for themselves. The distinct characteristics of firms relying on knowledge capital, such as incomplete contracting, non-excludability, asymmetric information and moral hazard makes FDI better suited as a foreign expansion strategy compared to an export strategy.
An argument for this is that it poses extremely high transaction costs on specifically exporting SMEs, to write contracts to protect their assets. An exporting SME should fear intelligence leakage and information sharing because the disclosure of core technology, know-how and innovations to exporters or distributors could undermine the competitive position of the firm.
SMEs relying on knowledge capital should therefore prefer direct FDI as a foreign policy compared to a more risky exporting approach. FDI protects companies better to asset appropriation and information leakage. Greater control over strategic resources would give a greater incentive to SMEs to transfer technology abroad, which would lead to an increased performance, considering the advantages of international expansion. I therefore argue that R&D will have a positive moderating effect on the relation between FDI and SME performance:
An organizing framework, relating all the hypotheses and its variables together can be found in figure 1.
Figure 1: Conceptual Model
Concept Measurement ROA Performance Dependent Variable Exports, FDI Internationalization Independent variables
Firm size Number of Employees
Control
Variables Leverage Debt/Total Equity
R&D investment Intangible Assets/Total Assets
The conceptual model reflects the relationship between the different variables, to be more precise between the endogenous and the exogenous variables. Performance is my endogenous variable. The exogenous variables consist of control variables and independent variables. The arrows illustrate the relation between the variables. Note that the R&D variable is a control variable in the main effects model, but an independent variable in the moderator model (model 5).
Overview 1: Regression Equations
(1) Model with only control variables:
Perfi = β0 + β1*LN(Firmsize)i + β2*(R&D)i + β3*(Leverage)i + ει [1]
(2) Model with exporting and control variables:
Perfi = β0 + β1*(Exports)i + β2*LN(Firmsize)i + β3*(R&D)i + β4*(Leverage)i + ει [2]
(3) Model with exporting, FDI (squared) and control variables:
Perfi = β0 + β1*(Exports)i + β2*(FDI)i + β3*(FDI2)i + β4*LN(Firmsize)i + β5*(R&D)i + β6*(Leverage)i + ει [3]
(4) Model with independent variables and moderator variable (FDI*EXP):
Perfi = β0 + β1*(Exports)i + β2*(FDI)i + β3*(FDI*EXP)i + β4*LN(Firmsize)i +
β5*(R&D)i + β6 *(Leverage)i + ει [4]
(5) Model with independent variables and moderator variable (FDI*R&D):
Perfi = β0 + β1*(Exports)i + β2*(FDI)i + β3*(FDI*R&D)i + β4*LN(Firmsize)i +
β5*(R&D)i + β6*(Leverage)i + ει [5]
All the models are derived from thorough theoretical arguments. The variables are not time-oriented, as only variables for one year (2005) have been taken into consideration. Therefore subscript i is considered to reflect the parameter for each individual firm. ει is the error term,
The interpretations of the coefficients in the regression equation of the main effects can be found in table 1.
Table 1: Interpretation of the regression coefficients: Main effect models1 Coefficient Interpretation of coefficients concerning models 1,2 and 3
β0
The intercept: the value of the function when it crosses the Y line: reflects the level of SME performance when exports are zero, when FDI is zero, when firm size is zero, when there is no leverage and when there is no R&D investment
β1 The change in the level of SME performance when the level of exports changes β2 The change in the level of SME performance when the level of FDI changes β3 The change in the level of SME performance when the level of FDI (squared) changes β4 The change in the level of SME performance when the firm size changes β5 The change in the level of SME performance when the level of R&D investment changes β6 The change in the level of SME performance when the leverage ratio changes
1 For the sake of simplicity, the interpretations of the coefficients for models 1, 2 and 3 have been
3. METHODOLOGY
3.1. Data source
In this research, only secondary data are analyzed, because time constraints make it too difficult to collect primary data. Moreover, it is quite hard to acquire reliable primary data for private companies. Therefore, all the data have been collected from a secondary data source: the Amadeus database of Bureau van Dijk.
3.2. Sample selection
The sample in my research consists of British SMEs that already have conducted a form of internationalization, which can be exporting, FDI or both. This way of sampling is called purposive judgment sampling, where sample members are selected to conform to a specific criterion (Coopers and Schindler, 2003). I therefore have a non-random sampling procedure, which can lead to selection bias2. The sample exists of British SMEs having less than 250 employees, according to the European and SBA definition. Only firms are selected, which have data available on the following variables: return on assets (ROA – hereafter), exports and FDI. This research focuses on SMEs that are already internationalized. That is, SMEs with data available on a certain form of internationalization: exports or FDI in my case.
The variables ROA and exports can be directly obtained from the Amadeus database. FDI is measured by the number of foreign subsidiaries that a SME possesses. FDI is only selected in case of a 10 percent ownership stake of the parent company. A 10 percent ownership stake is considered as the minimum requirement for a foreign enterprise to qualify as FDI (Larimo, 2002).
Further, data on R&D and leverage are included. The amount of R&D investment is calculated by dividing the amount of intangible assets by total assets and this ratio serves as proxy for R&D investment (Denekamp, 1995). However, it should be taken into consideration that intangible assets, apart from R&D, also consist of other variables such as goodwill and the brand name of a company. There may be a bias towards companies possessing large amounts of those variables. However, Denekamp (1995) argues that intangible assets can be a good proxy for R&D investment. Therefore, I consider intangible assets as a reliable proxy for R&D. The leverage variable is calculated by dividing total debt by total equity (Brealey et al., 2001).
Finally, if all these criteria are applied, I am left with 352 British SMEs in my sample. It has to be mentioned that the time dimension for this study is the year 2005, which was the most recent year available at the time the investigation was carried out. It is beyond the scope of this study to examine trends and changes through out time. Therefore, this study can be regarded as cross-sectional.
3.3. The choice for the United Kingdom
This paper focuses on a British sample of SMEs. The United Kingdom (UK) as a nation and culture has several characteristics, which makes this country especially well suited for small companies to internationalize. In this section, I will come up with some arguments for British SMEs to expand abroad.
First, I would like to emphasize the comprehensive colonial history of the UK. After the end of the British Empire, the UK retains influence throughout the world because of the Commonwealth of Nation and the use of the English language. The English language has been spread to all corners of the world, particularly because of the contribution of the British Empire. The English language can clearly be considered as a global language. For this reason, many people speak English as a second language. This reduces internationalization barriers for British SMEs, compared to SMEs from other countries, which often encounter difficulties in adapting to different languages. The previous argument is confirmed by Spence and Crick (2006), who found that cultural difficulties appeared to be less problematic for British SMEs, compared to SMEs from other countries.
3.4. Research methodology
In this section, I will give a short overview about what kind of research methodology I applied in this study. In business strategy, the most common empirical approach to discover superior performance has been to regress a measure of performance on a specific variable (Hill, Griffiths and Judge, 2001).
According to Ruigrok and Wagner (2003), researchers who investigated the performance -internationalization link have employed different statistical techniques to data analysis. Some applied ANOVA techniques on averaged data (Daniels and Bracker, 1989; Geringer and Hebert, 1989; Sullivan, 1994), drawing performance comparisons over firms at different levels or ranges of internationalization. Others used classical multiple regression analysis on data for the variables under study (Hitt et al., 1997).
The different methodologies mentioned above are applied in business strategy, because multivariate techniques are preferred to simple bivariate tests, for example correlation coefficients. The main disadvantage of bivariate tests, when for example performance is analyzed, is that these kinds of tests omit variables, which better explain changes in the dependent variable. Bivariate techniques, therefore, are only useful when it is certain that there are no other variables influencing performance. Another disadvantage of bivariate tests is that these tests treat the variables symmetrically, rather than differentiating between the dependent and the independent variable.
Consequently, multivariate analysis will be used as methodology in this paper, because it is expected that a lot of different variables determine firm performance. The aim of multivariate regression is to predict the change in the dependent variable in response to changes in the independent variables. This is often achieved through the statistical tool of least squares (Hill et al., 2001).
3.5. Variables
In the subsequent paragraphs, I will explain the different variables, used in the regression analysis.
3.5.1. Dependent variable
should use different measures of organizational performance because of the differences in their research question”. These different fields of strategic management can be argued along three dimensions, namely theoretical, empirical and managerial.
In this thesis, business performance is employed in an empirical context. Ginsberg and Verkantraman (1985: 421) argue that: “empirically, most strategy research studies employ the construct of business performance to examine a variety of strategy content and process issues”. Business performance is a subset of the overall performance of organizational effectiveness. Verkantraman and Ramanujam (1985) argue that the narrowest conception of business performance centers on the use of simple outcome based on financial indicators that are assumed to reflect the fulfillment of the economic goals of the firm. This concept, referred to as financial performance, has been the dominant model in empirical strategy research (Hofer, 1983). In my study, I also use a financial operator as performance measure. The question remains what kind of financial operator is the most appropriate.
Contractor et al., (2003) discuss the previous literature on the link between performance and internationalization. They summarize a number of studies and they show how certain studies operationalize performance. Different authors researching the performance - internationalization relationship, use different indicators of performance. Return on assets is widely used by authors like Ruigrok and Wagner (2003), Geringer and Hebert (1989), Sullivan (1994), Ramaswamy (1995) and Hitt et al., (1997). I therefore consider ROA as a reliable indicator for business performance in my study.
As a definition of ROA, I refer to De Meuse, Bergmann, Vanderheiden and Roraff (2004). They define ROA as a measure of profitability of a company in relation to the amount of dollars invested. According to them, ROA is an index of overall return on investments and indicates how efficiently dollars are utilized.
3.5.2. Independent variables
The second measure of internationalization is FDI, which reflects whether SMEs have foreign direct investments. I define FDI as the total number of foreign subsidiaries owned by the parent company. As discussed in a previous paragraph, the parent company should at least possess a 10 percent ownership stake in the foreign subsidiary to qualify the foreign subsidiary as FDI (Larimo, 2002).
3.5.3. Control variables
Like other studies researching the relationship between internationalization and performance, I will add control variables to my regression (Tallman and Li, 1996; Gomes and Ramaswamy, 1999). Internationalization alone is definitely not the only factor determining SME performance. The first control variable is firm size. The influence of size can be differently interpreted. It can be argued that small organizations may perform badly because they suffer from liabilities of smallness, including problems with recruiting employees and raising capital. On the other hand, when organizations grow, the possibilities of creating bureaucracy are increasing. Often, bureaucratic organizations have a larger chance to fail. In other words, firm size is included to control for economies and diseconomies of scale. Firm size is measured as a logarithmic function of the number of employees, to decrease the dispersion in this variable and to make the distribution of the data closer to normality.
As a second control variable, financial leverage is included in my model, which controls for the capital structure of the company. Financial leverage increases the risk of the company. However, because the costs of debt can be tax-deducted, savings can be made and this increases the value of the company (Brealey et al., 2001). Consequently, financial leverage is expected to have a significant influence on SME performance.
Third, a R&D variable is added, because it is researched that the R&D ratio has a large influence on SME performance (Lu and Beamish, 2001). I use a proxy for the R&D investment, namely the ratio of intangible assets divided by total assets (Denekamp, 1995). I assume that these intangible assets describe for a large part the amount of R&D investment.
statistically significant. A reason for this may be the fact that the sample size becomes too small, if my sample of 352 SMEs is divided in 5 different sectors. Therefore, I concluded that there is no point in analyzing the industries separately and I decided that the data just as well could be pooled into one sample.
3.6. Description of the interaction effects
In my thesis, the last two hypotheses contain interaction effects. Interaction effects are operationalized due to the use of moderator variables. A moderator variable is a second independent variable that is included in the regression equation, because it is believed to have a significant contributory or contingent effect on the originally stated independent variable - dependent variable relationship (Coopers and Schindler, 2003). This means that an interactive impact usually resembles a mathematical product moment, where the presence of one amplifies or reduces the direct impact of the other (Etemad, 2004).
Moderator variables are constructed by multiplication of different independent variables. The regression equation gets another meaning in this way, that is the meaning of the independent variables changes. In my regression equation, the moderator means that the relationship between the independent variable (FDI) and the dependent variable (performance of SMEs) changes, based on the level of exports and R&D, respectively. I add two moderators to my model, which are separately tested. The first moderator is the link between exports and FDI. Exports are multiplied by FDI, to discover whether FDI has a greater impact on performance when it is initiated by a high level of exports. In other words, the interaction between exports and FDI magnifies the process of internationalization. The second moderator is the link between FDI and the amount of R&D. FDI is multiplied by R&D, to discover whether FDI has a greater influence on performance when FDI follows a higher level of R&D.
Table 2: Interpretation of the regression coefficients: Interaction effects models3 Coefficient Interpretation of coefficients applied to model 4 and 5
β0
The intercept: the average value of Y when all predictors are zero. β0 reflects the average level of SME performance when FDI is at its average, with the average level of exports, the average level of R&D, with zero firm size and with zero leverage
β1 The change in the level of SME performance when the level of exports changes β2 The average difference in the level of SME performance for FDI when the level of exports and the level of R&D are at their average β3 (Model 4) The unit change in the average difference of FDI and SME performance when the level of exports changes
β3 (Model 5) The unit change in the average difference of FDI and SME performance when the level of R&D changes β4 The change in the level of SME performance when the firm size changes β5 The change in the level of SME performance when the level of R&D changes β6 The change in the level of SME performance when the leverage
ratio changes
3 For the sake of simplicity, the interpretations of the coefficients for models 4 and 5 have been summarized
in one table, because the way of interpreting the coefficients (except β3) does not change between the
4.
EMPIRICAL FINDINGS
4.1. Descriptive statistics4
As discussed, I use the multiple regression model for my analysis. For this model to work properly, several tests have to be conducted. Tests regarding heteroskedasticity, autocorrelation and multicollinearity will be executed. The descriptive statistics of the study can be found in table 3. Several tests will be performed to look at the descriptive statistics.
Table 3: Descriptive statistics and bivariate correlations of data on 352 British SMEs
Variable Mean S.D. 1 2 3 4 5 6 7 1 ROA 4,660 14,739 1 2 R&D 11,364 19,720 -0,233** 1 3 Leverage 23,221 17,525 0,142** -0,223** 1 4 Firm Size 4,729 0,739 0,102 0,145** 0,104 1 5 Exports 48,063 32,437 -0,180** 0,116* -0,177** -0,091 1 6 FDI 2,168 30,063 -0,064 0,248** -0,181** 0,067 0,146** 1 7 FDI (squared) 14,054 71,227 0,049 0,091 -0,132* 0,039 0,044 0,701** 1
*Correlation is significant at the .05 level (two tailed) ** Correlation is significant at the .01 level (two tailed)
First, I will look at the outliers, by inspecting the standard deviations of the different
variables. The standard deviations are more or less the same for all variables, only the natural logarithm for firm size (measured by the number of employees) has a low standard deviation, which was expected, because the natural logarithm reduces the dispersion in this variable. It is always hard to determine which are outliers in the data and consequently should be removed. As a common approach, I decided to remove any outliers that would exceed 1000%. As I did not have any outliers of this size, I did not remove any observation.
The second inspection is to look at normality of the variables. The Jarque Bera test for normality can be used to discover whether the variables are normally distributed. The null hypothesis of this test assumes normality, so it follows that rejection of the null corresponds with a non-normal distribution. The Jarque Bera test consists of both the measures of skewness and kurtosis. Skewness refers to the symmetry of values around the mean value and kurtosis refers to the peakedness of the distribution. None of my variables are actually normally distributed. The presence of non-normal variables can lower the parametric tests’ validity. However, for sample
sizes that are sufficiently large, the violation of the normality assumption has virtually no consequences (Brooks, 2002). I assume that my sample size of 352 observations is sufficiently large.
Endogeneity problems deal with reversed causality. For example, if performance, which is my dependent variable, influences internationalization in a systematic way. A common way to prevent endogeneity problems is to think which independent variable will be influenced by the dependent variable. I can imagine that good performance will lead to expanding investments abroad, in the form of exports but especially FDI, because this requires more investment. As a solution therefore I will run some reversed regressions:
Perfi = β0 + β1*(Exports)i + β2*(FDI)i + β 3*LN(Firmsize)i + β4*(R&D)i + β5*(Leverage)i + ει [6]
Exportsi = β0 + β1*(Perf)i + β2*(FDI)i + β 3*LN(Firmsize)i + β4*(R&D)i + β5*(Leverage)i + ει [7]
Now the dependent variable is the level of exports and SME performance is exogenous. If exporting is significant, possible endogeneity problems could arise. However, this is not the case in this regression, nor when FDI as a dependent variable is tested on SME performance.
Another problem in regression analysis is heteroskedasticity. This means that the errors exhibit varying variances. Heteroskedasticity does not influence the estimated coefficients, but it influences the standard errors of the coefficients and in this way it impacts the t-statistics. According to Hill et al., (2001), heteroskedasticity is often encountered when cross-sectional data (data on a number of economic units in a given point in time) are analyzed. As that is the case in my research, I should watch out for this problem. An overview of the heteroskedasticity tests can be found in table 4.
Table 4: Tests for Heteroskedasticity
Model 1 Model 2 Model 3 Model 4 Model 5
White’s Heteroskedasticity Test
F-statistic 6,221 4,215 2,829 2,555 3,000
Probability 0,000 0,000 0,000 0,000 0,000
I perform White’s heteroskedasticity test (Hill et al., 2001) to discover the presence of heteroskedasticity. The null hypothesis of this test is no heteroskedasticity. According to the F-test of all the models, I have to reject the null hypothesis and conclude that I have heteroskedasticity of some unknown form in my models. In order to control for heteroskedasticity I use White’s heteroskedasticity consistent estimation method (Hill et al., 2001) in each regression.
Autocorrelation is the case when the covariance between any pair of random errors is not equal to zero. However, for cross-sectional data, the randomness of the sample implies that the error terms are uncorrelated (Hill et al., 2001). Since I use cross sectional data, I assume that autocorrelation is not a problem in my regression analysis.
4.2. Results of the regression
The results of the regression analysis are summarized in table 5.
Table 5: Results of the Multiple Regression Analysis of data on 352 British SMEs
Independent Variable
Model 1 Model 2 Model 3 Model 4 Model 5
Coefficients Coefficients Coefficients Coefficients Coefficients
1 Intercept -0,414 3,133 2,499 -2,279 -0,164
(-0,046) (0,358) (0,291) (-0,449) (-0,019)
2 Firm Size (Log) 0,971 0,746 0,777 1,065 0,739
(-0,539) (-0,421) (0,449) (1,016) (0,415) 3 Leverage 0,099 0,085 0,094 0,082 0,081 (2,378)** (2,122)** (2,362)** (1,803)* (2,059)** 4 R&D -0,159 -0,155 -0,156 -0,135 -0,144 (-3,624)*** (-3,480)*** (-3,515)*** (-3,280)** (-3,166)** 5 Export -0,046 -0,045 -0,033 -0,036 (-1,900)* (-1,875)* (-1,354) (-1,462) 6 FDI -0,187 -1,830 0,167 (0,510)* (-2,769)** (0,431) 7 FDI (squared) 0,019 (1,809)* 8 FDI*Exports 0,015 (-2,769)* 9 FDI*R&D 0,001 (1,216) R-square 0,071 0,081 0,089 0,109 0,102 Adjusted R-square 0,063 0,070 0,075 0,091 0,084 F 8,836*** 7,599*** 6,726*** 6,039*** 5,611***
Upper Number in cell is parameter estimate, numbers between brackets are t-statistics Probabilities: *p<0.10, **p<.05, ***p<.001 (all two tailed tests)
I test my four hypotheses with five regressions, which are described in overview 1. The overall significance of the models is tested with the F-test. The F-test tests two hypotheses. The first hypothesis tests whether one of the parameters of the multiple regression model is equal to zero, the alternative hypothesis is that one or more of the parameters is not equal to zero. The F-statistics of the different models indicate that all models are highly significant.
adjusted R2 should be considered, because I do not have a lot of independent variables, which explain my model.The adjusted R2 are rather low for the different models. However, the adjusted R2 indicates a reasonable explanatory power for this kind of research (Lu and Beamish, 2001). Similar studies researching internationalization and SME performance (Lu and Beamish, 2001; Majocchi and Zucchella, 2003) did not find higher values for the overall explanatory power of their models, while they were adding more explanatory variables. A reason for the low adjusted R2 can be explained with the determinants of performance. Exporting and FDI are certainly not the only determinants of firm performance. There are much more factors which influence firm performance, for which only some have been controlled for in my analysis.
Model 1 only consists of the control variables. In this way the incremental value of adding independent variables to the other models can be discovered. There are three control variables: firm size, R&D and leverage. For firm size the natural logarithm has been calculated to reduce the dispersion in the observations. Firm size is not significant, however. The two other control variables are significant. Leverage has a positive impact on SME performance (p<.05). This result seems to confirm the finance theory, which argues that SMEs having larger amounts of debts are subject to greater interest rate volatility and therefore must produce higher returns in order to balance the risk of the financial structure (Brealey et al., 2001).
The R&D ratio is highly significant (p<.001) but has a negative influence on performance. This result is actually quite logical, if one keeps in mind that my sample only applies to one year (2005). SMEs that have invested a lot in R&D will first see their performance declining because of the investment in R&D. Positive returns due to this R&D investment will accrue later, but my sample does not include for this effect, because it only covers one year. This argument contradicts with previous studies on SMEs (Majocchi and Zucchella, 2003). It should be remarked that later on, in model 5 to be precise, the R&D ratio does not qualify as control variable any longer, but is inserted as independent variable.
Model 2 adds exports to the regression. This model tests the first hypothesis, which states that exporting is positively related to firm performance. In contrary to what was hypothesized in the first hypothesis, exports have a significant (p<0.1) but negative relation to firm performance.
performance. As can be noted from the regression output table, this relationship is negative and significant (p<.10). The squared term of FDI tests whether the relation between FDI and SME performance is non-linear. The positive and significant result (p<.10) of this variable confirms, together with the negative FDI variable, my second hypothesis and provides evidence for the existence of the liability of foreignness. This means that SME performance declines at the early stages of FDI up to a threshold and then increases at an accelerating rate, when more resources are invested in FDI.
Models 4 and 5 are the moderator models and therefore deserve special attention. Some adaptations were applied to the independent variables to make the interaction effects more interpretable. These adaptations concerned the independent variables, which served as input for the moderators. These independent variables were mean centered and this was discussed in a former paragraph.
The control variables firm size, leverage and R&D stayed the same in sign and significance compared to the former models. Model 4 adds the moderator between FDI and exports. I hypothesized a positive relation between these variables and this hypothesis was confirmed by empirical investigation. The moderator was statistically significant (p<.10), which proves that the moderator effect was present and has a positive sign. This means that the relationship between FDI and performance is stronger in case FDI is initiated by a higher level of exports.
5.
DISCUSSION
The main objective of this thesis was to research different avenues of internationalization and the implication of these different internationalization paths on SME performance. To be more precise, the following research questions were answered: What is the relationship between a SME’s performance and its export intensity? What kind of relationship exists between the level of FDI and SME performance? Does internationalization due to FDI enhance SME performance if FDI is initiated by a high level of exports? Will a greater investment in R&D lead to better SME performance, if FDI is considered as a form of foreign expansion?
Considering the first hypothesis, I did not find any empirical evidence that exports would be positively related to SME performance (model 2). Although it was argued that several economic benefits, such as increased economies of scale and scope and the opportunity to gain valuable international experience could be gained due to a higher exporting intensity, my empirical tests do not provide any support for these arguments.
Shoham (1998) revealed that many studies, exploring the performance effects of exporting were inconsistent in their findings, mainly because these studies differed in their operational definitions of exporting. If I compare my results to similar studies, with the same operational definitions of exporting, the results remain mixed. Lu and Beamish (2001) also find a negative relation between exports and performance for their Japanese sample, but according to them, this can be explained by the appreciation of the yen. Majocchi and Zucchella (2003) do not find any evidence for a relation between exporting and performance, which can be attributed to a sample made up of only exporters according to them. Although I deal with a similar sample, I find a negative relationship between these variables. I argue that because exporting is often the first mode of internationalization, learning and experience costs may be relatively high. SMEs have to pay costs for their mistakes that are made in case of initial foreign expansion. For example, it can be argued that reliable contacts with foreign customers have to be built and this may take time. In this way, a negative relation can be explained.
