The ongoing debate:
The effect of multinationality on performance
Do firm size and home market size moderate the effect of multinationality on performance?
Student: Alexandra Knoef Student number: 11109963 Date: June 24th 2016
University of Amsterdam MSc Business Administration Track International Management Final Version Master Thesis
Supervisor: Dr. N. Pisani Second supervisor: Dr. I. Haxhi
State of originality
This document is written by Alexandra Knoef who declares to take full responsibility for the contents of this document. I declare that the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The multinationality – performance (M-P) relation received considerable attention in the academic field of International Business (IB). The findings of the nature of this relation remain however dissimilar from each other. The purpose of this study is to fill this inconsistency gap through the execution of an empirical study. Recent studies mainly focus on contextual factors influencing the M-P relationship. This study however combines a contextual factor (home market size) and an organizational factor (firm size), in order to verify their explanatory power on the M-P relationship. Three hypotheses are tested with the use of a scale- and scope measure of multinationality. No significance is detected for the baseline hypothesis that proposes a horizontal S-curve M-P relationship. The first moderator, firm size, has a positive moderating impact on the M-P relationship whereas the second moderator, home market size, negatively influences this relationship. Firm size shows a significant positive effect on the M-P relationship when analyzed with both, scale and scope, measures. Home market size shows a significant negative effect on the M-P relationship when analyzed with a scope measure. This implies that when the depth or breadth of multinationality increases, larger firms experience even higher performances. Moreover, this shows that the size of the home market attenuates the positive effect of the breadth of multinationality on performance. In conclusion, firm size and home market size both have a significant interaction effect on the M-P relationship.
Table of contents 1. Introduction ... 6 2. Literature Review... 9 2.1 Multinationality... 9 2.1.1 Scale measures ... 10 2.1.2 Scope measures ... 11 2.2 Performance ... 13 2.3 M-P relationship... 14
2.3.1 Scale measures used in M-P studies ... 20
2.3.2 Scope measures used in M-P studies ... 21
2.4 Empirical context considered ... 24
2.5 Research gap ... 25
3. Theoretical Framework ... 27
3.1 M-P relationship... 27
3.2 Firm size... 28
3.3 Home market size ... 31
4. Methods... 35
4.1 Sample and data collection ... 35
4.2 Variables ... 36
4.2.1 Dependent variable ... 36
4.2.2 Independent variable ... 36
4.2.3 Moderating variables ... 37
4.2.4 Control variables ... 38
4.3 Analysis and results ... 39
5. Discussion ... 48
5.1 Academic relevance ... 48
5.2 Managerial implications... 51
5.3 Limitations and suggestions for future research ... 52
6. Conclusion ... 54
Acknowledgement ... 56
References ... 57
List of figures
Figure 1. Main M-P relationships in previous studies ... 17
Figure 2. Firm size positively moderates M-P relationship ... 31
Figure 3. Home market size negatively moderates M-P relationship ... 33
Figure 4. Conceptual model ... 34
List of tables Table 1. Scale and scope measures (home and foreign region) to measure multinationality .. 12
Table 2. Multinationality measurement in recent related papers ... 18
Table 3. FG500 according to home country and home market size ... 41
Table 4. Means, Standard Deviations, Correlations ... 45
Table 5. Results of hierarchical regression analysis for scale of multinationality ... 46
1. Introduction
Geoffrey Garrett (2000, p. 946), professor of international studies and political science at Yale University, wrote in one of his famous papers “no matter how many different numbers are presented or how frequently one hears them, the growth of international economic activity in the past 30 years remains staggering”. There are many different reasons for a firm to reach beyond national borders as for example cost effectiveness. Hence, there could be no doubt about the truth of Garret’s quote. However, recent research questions whether going global is the most profitable strategy for a firm (Oh and Contractor, 2014; Oh et al., 2015; Rugman and Verbeke, 2004; Rugman and Oh, 2013). This research adds onto those works by studying the effect of multinationality on performance.
A firm is considered to be a multinational enterprise (MNE) when it operates in different countries (Dunning and Lundan, 2008). A MNE should engage in foreign direct investment (FDI) and own or control value added activities (Dunning and Lundan, 2008). Recent years, firms kept expanding internationally due to the lowering costs of moving goods and simultaneously removing of entry barriers by governments (Garrett, 2000). Various scholars confirm this phenomenon and show numerous advantages associated with foreign expansion (Dunning and Lundan, 2008; Garrett, 2000; Kogut, 1985). The degree of international activity denotes to the level of multinationality of a MNE. Multinationality refers to the degree to which business activities spread across national boundaries (Tseng et al., 2007).
There are many different ways to capture the degree of multinationality and performance, as there are various different measures. Previous research, therefore, indicates this conceptualization problem (Lu and Beamish, 2004). Partly due to this issue, there are inconsistent outcomes associated with the multinationality and performance (M-P) relationship. Previous studies find different relationships varying from positive linear (Delios
and Beamish, 2005), negative linear (Kumar, 1984), cumulative (Johanson and Vahlne, 1977), U-shaped (Contractor et al., 2007; Ruigrok and Wagner, 2003), inverted U-shaped (Chao and Kumar, 2010; Qian et al., 2010; Riahi-Belkaoui, 1998; Sullivan, 1994) to S-shaped (Benito-Osorio et al., 2016; Contractor et al., 2003; Lu and Beamish, 2004; Oh and Contractor, 2014; Oh et al., 2015; Rugman and Oh, 2010). This study adds to IB literature by filling this conflicting gap through the completion of an empirical study.
The importance of researching this relationship is forthright. First, if an optimal level of multinationality would exist, MNEs would be able to obtain their optimal performance (Contractor et al., 2003; Verbeke et al., 2009). Second, due to innumerable studies completed, large amounts of data on multinationality and performance are available and could lead to even better analysis in the future (Verbeke et al., 2009). Last, the initial internationalization barriers, identified later in this research, are an important aspect for companies’ strategists (Contractor et al., 2003). As consensus is not yet reached, the relationship between multinationality and performance needs deeper exploration.
Research should not only focus on this general relationship, but also needs to consider contingencies that affect this association (J. Li and Yue, 2008). Oh and Contractor (2014) suggest that future research should discover the effect of firm characteristics. This is important since it is possible that organizational characteristics are more powerful to influence the level of multinationality than contextual factors (Oh and Contractor, 2014). Furthermore, size-dependent factors have been neglected in M-P related research (Benito-Osorio et al., 2016). Accordingly, this research combines an organizational- and contextual interaction effect focusing on size-dependent factors. The research question this thesis aims to answer is: What is the moderating effect of firm size and home market size on the relationship between multinationality and performance?
The thesis is structured as follows. The next chapter includes the literature review regarding the concepts of multinationality and performance. Moreover, the recent findings concerning the M-P relationship are clarified. The two subchapters, the empirical context considered and the research gap, are the closing words in this chapter. Chapter three comprises the development of the three hypotheses. In the chapter thereafter, methods regarding the analyses are discussed such as the sample, variables, analysis and results. Chapter five includes the discussion of the results. This chapter furthermore explains the academic and managerial relevance and limitations of this study. The last chapter encloses concluding observations.
2. Literature Review
This chapter introduces the important concepts multinationality and performance. Thereafter, the M-P relationship is illustrated, through the use of scale- and scope measures to separate the measurement methods of previous studies. Lastly, the empirical setting and research gap are clarified.
2.1 Multinationality
The relationship between multinationality and performance is studied comprehensively. However, the outcomes of these studies are inconsistent. One of the reasons is that the variables in this M-P relationship are measured differently in every research (Aggarwal et al., 2011; Annavarjula and Beldona, 2000; Hennart, 2011; Li, 2007). Within this chapter the different definitions and measures of multinationality are examined.
There are many ways to define multinationality. One of the earliest definitions of multinationality is by Dunning (1971, in Annavarjula and Beldona, 2000, p. 51) who defines a firm pursuing a multinational strategy as “a multinational enterprise that owns or controls facilities in more than one country”. Cantwell and Sanna-Randaccio (1992, in Annavarjula and Beldona, 2000, p. 51) define multinationality as “the value of international production carried out by affiliates in other countries relative to the value of domestic production of the parent company in this home country”. However, a more simplistic definition of multinationality refers to the degree to which business activities spread across national boundaries (Tseng et al., 2007). Sullivan (1994) has a similar definition referring to a firm’s expansion beyond national borders into foreign markets. Annavarjula and Beldona (2000) add the need of a firm to invest in assets and/or controlling activities, to that definition. The definition of Dunning seems to be the most widely accepted definition amongst scholars in IB
(Li, 2007). Therefore, this study refers to multinationality when a firm owns or controls facilities in more than one country.
As there are many definitions that focus on other elements of multinationality, there are also many ways to capture this variable. Aggarwal et al. (2011) reported 19 attributes that have been used to operationalize multinationality in the 393 studies conducted between 1987 and 2007. Therefore it is not surprising that outcomes are inconsistent or maybe even contradictory (Aggarwal et al., 2011). Since there is no agreed classification system for the degree of multinationality, scholars use many different ways of defining and measuring this concept. Until today, no standard way of measuring multinationality is agreed upon.
Hennart (2011, p. 141) however, mentions five main ways in literature to measure multinationality: “(1) a firm’s dependence on foreign sales; (2) its dependence on foreign production; (3) the dispersion of its foreign sales; (4) the number of foreign countries in which it is active; and (5) the diversity of the countries in which it is active”. These five measures can be divided into two scale measures, two scope measures and an entropy measure (Oh, 2009; Rugman and Oh, 2013). The entropy measure is used to quantify geographic diversification (Rugman and Oh, 2013). It reflects not only the dispersion of foreign markets, but also the importance of these markets (Rugman and Oh, 2013). Therefore, it shows the geographic diversification in depth and breadth (Oh et al., 2015). As the entropy measure is mainly used as an indicator of the relative importance of the subsidiaries to the firm, this measure is beyond the purpose of this thesis. The scale- and scope measures are discussed in more detail in the next sections and summarized in table 1.
2.1.1 Scale measures
Scale measures account for the depth of internationalization (Rugman and Oh, 2013). It relates to the level of multinationality (Verbeke and Brugman, 2009). This means that it
measures how dependent a firm is on foreign markets. When a firm has almost all its operations in foreign markets, it scores high on the scale measurement.
The two most common scale measures for multinationality are foreign sales to total sales (FSTS) and foreign assets to total assets (FATA). The FSTS represents the degree of multinationality due to downstream firm-specific advantages such as marketing activities. FATA on the other hand, accounts for the degree of multinationality due to the upstream firm-specific advantages such as sourcing (Oh, 2009). To explain the regional strategy of a MNE, the intra-regional sales-to-total sales (IRSTS) and intra-regional assets-to-total assets (IRATA) are respectively used to compare to the FSTS and FATA.
Most studies use FSTS as measure for multinationality, since these data are relatively easy to gather (Contractor et al., 2003; Contractor et al., 2007; Oh and Contractor, 2014; Ruigrok and Wagner, 2003). Rugman and Oh (2011) state that the FTST is an appropriate measure for multinationality. However, it is challenging to capture multinationality within a single measure (Sullivan, 1994). Despite the various critique points on the usage of this measure, recent studies still use this single measure (Oh and Contractor, 2014).
2.1.2 Scope measures
Scope measures account for the breadth of international expansion (Goerzen and Beamish, 2003). Where the scale measure ignores the dispersion of international activity, the scope measure computes in how many foreign markets a firm is active (Verbeke and Brugman, 2009). The scope measure is not widely used as a measure for multinationality in IB research. However, it adds an important dimension to scale measures.
The two common scope measures used to measure multinationality are the number of foreign countries in which a firm operates (NOFC) and number of foreign subsidiaries-to-total number of subsidiaries (FBTB) (Oh, 2009; Rugman and Oh, 2013). Correspondingly,
the measures used at the regional level are the number of home region countries in which a firm operates (NOIRC) and the number of subsidiaries in a home region to the number of total subsidiaries (IRBTB) (Oh, 2009).
Oh (2009) analyzes the international scale and scope of European MNEs. Oh (2009) uses five multinationality measures. These measures comprise two scale measures, two scope measures and one entropy measure. This method is comparable to the described measures of Hennart (2011). Within this study he concludes that scale and entropy measures are better in explaining multinationality than scope measures. The scope measures simply count countries or subsidiaries. These scope measures do not capture the strategic importance of the market or any geographic dispersion (Oh, 2009). Researchers might think that the MNE is more dispersed than they really are (Oh, 2009).
Table 1. Scale and scope measures (home and foreign region) to measure multinationality
Measures Foreign region Home region
Scale Foreign sales-to-total sales (FSTS) Foreign assets-to-total assets (FATA)
Intra-regional sales-to-total sales (IRSTS) Intra-regional assets-to-total assets (IRATA)
Scope Number of foreign countries in which a firm operates (NOFC) Number of foreign subsidiaries to total subsidiaries (FBTB)
Number of home region countries in which a firm operates (NOIRC)
Number of subsidiaries in a home region to total subsidiaries (IRBTB)
2.2 Performance
Performance refers to the execution or accomplishment of work, acts or feats. Within the M-P research performance is captured at two dimensions, which are financial- and operational performance (Ruigrok and Wagner, 2003).
Financial performance exists of accounting-based measures and market-based measures (Ruigrok and Wagner, 2003). The most commonly used measures to assess performance are accounting-based measures such as return on assets (ROA) and return on investments (ROI) (Annavarjula and Beldona, 2000). Some other accounting-based measures are return on sales (ROS) and return on equity (ROE) (Yang and Driffield, 2012). Relevant literature uses typically above-mentioned accounting-based financial indicators (Li, 2007; Yang and Driffield, 2012). Accounting-based indicators capture short-term performance and use a firm’s size as measurement (Yang and Driffield, 2012). Next to accounting-based indicators, market-based indicators are also used to capture financial performance (Li, 2007). Examples are Tobin’s Q and risk-adjusted return. Market-based indicators measure long-term performance and use the valuation of a firm (Yang and Driffield, 2012).
The usage of only financial methods leads to shortcomings such as heterogeneity and possible managerial manipulation (Li, 2007). Operational performance measures, such as cost efficiency indicators, do not capture direct financial outcomes. These measures establish underlying success factors that eventually lead to financial outcomes (Ruigrok and Wagner, 2003).
Numerous studies use different measures, however researches almost never justify why they choose a certain measurement (Annavarjula and Beldona, 2000). On the other hand, there are also studies that use all the above-mentioned methods (ROI, ROS, ROA, ROE, growth and Tobin’s Q) to measure performance (Contractor et al., 2003).
2.3 M-P relationship
Previous scholars find different relationships when testing the effect of multinationality on performance (Figure 1). These outcomes vary from positive linear (Delios and Beamish, 2005), negative linear (Kumar, 1984), cumulative (Johanson and Vahlne, 1977), U-shaped (Contractor et al., 2007; Ruigrok and Wagner, 2003), inverted U-shaped (Chao and Kumar, 2010; Qian et al., 2010) to S-shaped (Benito-Osorio et al., 2016; Contractor et al., 2003; Lu and Beamish, 2004; Oh and Contractor, 2014; Oh et al., 2015; Rugman and Oh, 2010). The core perception in IB is that multinationality positively affects performance, as a firm is able to reach economies of scale and scope (Contractor et al., 2003; Hennart, 2011). However, a firm does not need to go overseas in order to exploit economies of scale, as long as the home market is large enough (Hennart, 2007). As not all scholars agree on the M-P relationship, this subject receives enormous attention (Li, 2007).
The first studies conducted in the 1970s focused mainly on the potential benefits of internationalization (Benito-Osorio et al., 2016). Such benefits were related to firm specific assets in new foreign markets (Benito-Osorio et al., 2016). The theory of FDI explains why firms would expand beyond national borders (Li, 2007). Two very influential theories are the eclectic paradigm of Dunning (1977; 1988) and the internalization theory of Buckley and Casson (1976). The internalization theory argues that firms internalize intermediate markets when they perceive to reduce costs and maximize profits (Buckley and Casson, 2009). Thereby, they aspire to overcome market imperfections by relying on their firm specific advantages (Buckley and Casson, 2009). Dunning’s eclectic paradigm on the other hand, proposes three advantages – Ownership, Location and Internalization – which influence the choice of FDI (Dunning, 1998). Besides firm advantages, which can be seen as the ownership aspect, Dunning suggests that location advantages also influence internationalization decisions (Dunning, 1998). Even though these theories made significant contributions to the
explaining of internationalization, some weaknesses limit the capacity of these theories to explain the M-P relationship (Li, 2007). The theories are rather static and are not able to provide insights over time (Li, 2007). Furthermore, they pay limited attention to the cost side of MNEs (Li, 2007). The next paragraph clarifies other theories that shed light on particular damaging sides of internationalization.
In the 1980’s scholars were more pessimistic about internationalization (Benito-Osorio et al., 2016). They considered the risks and costs of operating in foreign markets. The costs of expanding internationally are captured in theories from different disciplines (Ruigrok and Wagner, 2003). A recognized theory discussing the costs of doing business abroad is the liability of foreignness (LoF) (Hymer, 1976; Zaheer, 1995). The LoF involves all additional costs of a firm operating in a market overseas. A local firm would not suffer from these costs (Zaheer, 1995). Examples of these costs could be a lack of information or unfamiliarity with the local culture (Li, 2007). External costs of internationalization are for example financial- and political risks (Ruigrok and Wagner, 2003). Financial risks entail among others exchange rate fluctuations. When foreign governments unexpectedly change the business environment, political risk could be a cost of internationalization (Ruigrok and Wagner, 2003).
The first studies propose a positive linear- and thereafter a negative linear relationship between multinationality and performance. However, in the 1990’s researchers began to recognize non-linear relations (Benito-Osorio et al., 2016). Two models are proposed: the U-shaped relationship and the inverted U-U-shaped model (Chao and Kumar, 2010; Contractor et al., 2007; Ruigrok and Wagner, 2003). Ruigrok and Wagner (2003) find a U-shaped relationship. This suggests that MNEs that internationalize first experience a decline, which is followed by positive performance (Contractor et al., 2003). They argue that operating costs are not reduced by initial foreign expansion (Ruigrok and Wagner, 2003). Furthermore, benefits that come from cheap raw materials and low labor costs are only available for firms
that have a high degree of internationalization (Ruigrok and Wagner, 2003). Firms need time to learn and investigate in the most efficient ways of doing business (Ruigrok and Wagner, 2003). Therefore, there are many costs for firms that initiate foreign expansion. Once the reversal occurs and the firm experiences more benefits than costs, the performance grows exponentially (Ruigrok and Wagner, 2003). Ruigrok and Wagner (2003) find no threshold that various scholars propose. Chao and Kumar (2010) find an inverted U-shaped relationship between multinationality and performance. As they study the Fortune Global 500 (FG500), their sample consists only of large MNEs. These MNEs have various international operations and are therefore already internationally experienced. Hence, the costs of doing business abroad are substantially lower than for smaller firms still in the initiating phase of foreign expansion. However, the costs might exceed the benefits when the level of multinationality is too high.
Most recent studies, conducted in the 2000’s, try to address the evolving international behavior of MNEs (Benito-Osorio et al., 2016). Scholars attempt to capture the incremental benefits and costs varying across different stages (Benito-Osorio et al., 2016). These studies argue for a horizontal S-curve relationship (Benito-Osorio et al., 2016; Contractor et al., 2003; Lu and Beamish, 2004; Oh and Contractor, 2014; Riahi-Belkaoui, 1998; Sullivan, 1994). The three-stage theory shows a decline in the first stage, due to incremental costs of doing business abroad (Contractor et al., 2003). As a result of LoF, the costs will exceed the benefits (Oh and Contractor, 2014). Thereafter, MNEs receive more benefits in stage two, due to experience. The MNE keeps expanding internationally, which results in economies of scale and scope (Oh and Contractor, 2014). In the final stage, the exceeding international expansion reduces performance. The reason is that this expansion is likely to occur in distant markets (Oh and Contractor, 2014). Oh and Contractor (2014) even refine the horizontal S-curve theory by proposing a different regional coverage for each of the three stages. MNEs
pursuing the home region strategy, experience stage one and two of the S-curve (Oh and Contractor, 2014). A host region strategy encompasses the third stage of the S-curve whereas a global strategy covers all three stages (Oh and Contractor, 2014). Depending on which stage in the S-curve is examined, different relationships are found (Benito-Osorio et al., 2016). Oh and contractor (2014) conclude that this clarification helps to make sense of previous contradicting researches.
Figure 1. Main M-P relationships in previous studies Source: Benito-Osorio et al. (2016)
Table 2 summarizes the different multinationality measures and outcomes of recent studies, from 2000 until now. In order to elaborate on the different studies and outcomes, the scale- and scope measures are explored separately. The studies that use scale measures are discussed first, followed by the studies that use scope measures.
Table 2. Multinationality measurement in recent related papers
Authors and year Multinationality measures M-P
relationship Scale Scope Entropy
Goerzen and Beamish (2003) NOFC, FBTB ∑i Ec ln (1/Et) Ec = number of employees in a particular country c ln (1/Et) = the weight given to each country Positive and negative Ruigrok and Wagner (2003) FSTS U-shaped Contractor et al. (2003) FSTS, FETE, FOTO (foreign offices) S-curve Rugman and Verbeke (2004) Categori zation of regions based on sales Regionalization Lu and Beamish (2004) NOFC, FBTB S-curve Delios and Beamish (2005) FBTB Positive linear Contractor et al. (2007) FSTS U-shape Osegowitsch and Sammartino (2008) Categori zation of regions based on sales Regionalization Oh (2009) FSTS, FATA, IRSTS, IRATA NOFC, FBTB, NOIRC, IRBTB ENT= ∑ NBTBi log(1/NBTBi) NBTB = number of subsidiaries in country i/ total number of subsidiaries log(1/NBTBi) = weight given to each country
Qian et al. (2010) FSTS NOFC INTRAa= ∑i∈aPaialn(1/P aia)
INTRAa equals the sales or subsidiaries within the ath global market region and paia is defined as the proportion of sales or the number of subsidiaries in the ith country to the total sales or total subsidiaries of the ath global market region.
INTER =m∑ i=1 Piln(1/P i ) m is the number of regions in which a firm derives sales or has subsidiaries, and Pi is the proportion of the ith region to a firm’s total sales or total number of
subsidiaries in all regions.
Negative linear
Chao and Kumar (2010) NOFC, FBTB Inverted U-shaped Rugman and Oh (2010) FSTS, IRSTS S-curve Rugman and Oh (2013) FSTS, FATA, IRSTS, IRATA NOFC, FBTB, NOIRC, IRBTB 4 entropy measurements based on the formula: ∑ NBTBi log (1/NBTBi). ENTRS - uses sales as proxy for activity in each region; ENTRA- uses assets as proxy for activity in each region; ENTRC- uses number of countries as proxy for activity in each region; ENTRB- uses number of subsidiaries as a proxy for activity in each region.
Regionalization
Oh and Contractor (2014)
FSTS S-curve
Oh, Sohl and Rugman (2015)
Geographic diversification = n∑ i=1 Pi ln (1=Pi) Pi = proportion of sales in country i for a firm with operations in N different countries. S-curve Benito-Osorio et al. (2016) FSTS S-curve
2.3.1 Scale measures used in M-P studies
Most inquiries use scale measures to study multinationality. When operational performance is used to measure multinationality, the majority of these studies use the ratio foreign sales to total sales (Contractor et al., 2003; Li, 2007; Oh and Contractor, 2014; Verbeke et al., 2009; Yang and Driffield, 2012). This measure is used on a regular bases as it is objectively measured and the needed information is relatively easy to access. Nevertheless, this ratio only explains part of the multinationality construct (Li, 2007; Verbeke et al., 2009). This chapter probes studies that use scale measures.
Various studies use FSTS as their only measure for multinationality (Benito-Osorio et al., 2016; Contractor et al., 2007; Oh and Contractor, 2014; Ruigrok and Wagner, 2003). Several arguments ground their usage of this measure. Ruigrok and Wagner (2003) explain that data for their study was only attainable in FSTS form. Therefore they had no choice but to rely on the financial dimension of multinationality. Secondly, FSTS is the most used multinationality measure in previous studies (Benito-Osorio et al., 2016; Ruigrok and Wagner, 2003). However, Sullivan (1994) argues that a single item measure poorly estimates the construct it is supposed to measure. Thus, Sullivan (1994) suggests using an index of various variables as measurement. Since most scholars agreed that multinationality is a complex variable, such an index would not suffice (Ramaswamy et al., 1996). Ramaswamy et al. (1996) argue the need for a new measurement. Furthermore, some studies even show that multivariable measures are not superior to a single measure such as FSTS (Benito-Osorio et al., 2016). Rugman and Oh (2011 in Benito-Osorio et al., 2016) support this claim. After Rugman and Oh (2011) conducted their research using scale- and scope measures, they argue that FSTS alone is a suitable measure for multinationality. Even though all four studies use FSTS as a measure for multinationality, they establish different results. Ruigrok and Wagner
(2003) and Contractor et al. (2007) find a U-form whereas Oh and Contractor (2014) and Benito-Osorio et al. (2016) find a horizontal S-curve relationship.
Since no superior measurement has evolved, many studies remain to use FSTS as multinationality measure. However, some studies combine several measures to tackle robustness and sensitivity checks (Qian et al., 2010). Rugman and Oh (2010) use two scale measures; FSTS and IRSTS. Within their research they support the S-curve relationship. Furthermore, they find that IRSTS is a valuable measure supporting the S-curve fit (Rugman and Oh, 2010). Contractor et al. (2003) even combine three measures as suggested by Sullivan (1994). These measures are FSTS, FETE and FOTO (Contractor et al., 2003). Besides using only scale measures, studies also combine scale- and scope measure. Within the next chapter studies that incorporate scope measures are further discussed.
2.3.2 Scope measures used in M-P studies
A relatively small percentage of studies use scope measures to analyze the M-P relationship. If studies use a scope measure, it is almost always in combination with scale or entropy measures (Goerzen and Beamish, 2003; Qian et al., 2010; Rugman and Oh, 2013; Sullivan, 1994). Only a few studies exclusively use scope measures (Delios and Beamish, 2005; Lu and Beamish, 2004). Within this section well-known studies that use scope measures are discussed.
The paper of Rugman and Verbeke (2004) is one of the most cited articles on multinationality. They are predominantly known for categorizing the sales strategy of MNEs. Since scope measures reveal information on the distribution of foreign activity, Rugman and Verbeke’s approach is considered to be a scope measure. Scope measures might not be frequently used, however Rugman and Verbeke (2004) disengaged many scholars to explore
multinationality using a scope measure to prove their regionalization theory. Therefore, this influential paper is discussed in more detail below.
Fundamental IB studies assume that increased multinationality lead to higher performances (Contractor et al., 2003; Hennart, 2011). This is because the costs of moving goods decreased and removal of entry barriers by government globalization accelerated (Garrett, 2000). Globalization is the integration of markets in goods, services and capital and is changing the world rapidly (Garrett, 2000). Core IB studies list various advantages of globalization and therefore confirm this positive relationship (Dunning and Lundan, 2008; Garrett, 2000; Kogut, 1985). Examples of these advantages are greater international experience and access to cheaper resources (Contractor et al., 2003).
Rugman and Verbeke (2004) are the first to notably question the degree of globalization of MNEs. As they distinguish between intra-regional and extra-regional sales, they use a scope measure to categorize the FG500 companies’ strategies. They categorize the FG500 in four regional groups according to where the company generated its sales. MNEs can follow a regional strategy when maintaining organizational operations within their region. Next to a home region focus, Rugman and Verbeke (2004) categorize companies to be host region-, bi-region- or globally focused. When pursuing foreign activities worldwide, a MNE pursues a global strategy. After categorizing all operations of the 500 MNEs, they conclude that around 80% of the MNEs operate within their home region (Rugman and Verbeke, 2004). Consequently, they find that only a small percentage of the 500 largest companies in the world are truly global. Subsequently, they conclude that most MNEs pursue a regional strategy (Rugman and Verbeke, 2004).
Osegowitsch and Sammartino (2008) question the data gathering method of Rugman and Verbeke (2004). Within their paper they retest the data using longitudinal data. Instead of only using data from 2001, Osegowitsch and Sammartino (2008) compare that dataset to the
dataset from 1991. Thereby they lower the host region threshold from 20% to 10%. Consequently, they find a larger share of firms with a bi-regional or global strategy than reported by Rugman and Verbeke (2004). Nevertheless, more than 50% of the firms are home region based. Therefore they report that the regionalization claim of MNEs is true but overstated (Osegowitsch and Sammartino, 2008).
Other studies similarly use scope measures to test the M-P relationship (Chao and Kumar, 2010; Delios and Beamish, 2005; Goerzen and Beamish, 2003; Lu and Beamish, 2004; Qian et al., 2010; Rugman and Oh, 2013; Sullivan, 1994). Within the sample of most cited studies on internationalization (Table 2), Delios and Beamish (2005), Chao and Kumar (2010) and Lu and Beamish (2004) are the three studies that exclusively use scope measures. Delios and Beamish (2005) measure multinationality using only one scope measure, which is the number of subsidiaries a firm has in a country (FBTB). They conclude that firms opting for a regional strategy have lower performance than firms with a global strategy (Delios and Beamish, 2005). Chao and Kumar (2010) and Lu and Beamish (2004) use the two most used scope measures, specifically NOFC and FBTB. Chao and Kumar (2010) study the moderating role of institutional distance on the M-P relationship. Based on the FG500 they find that the M-P relationship has an inverted U-shaped curve (Chao and Kumar, 2010). Lu and Beamish (2004) investigate almost 1500 Japanese firms over 12 years. They conclude a horizontal S-shaped relationship between multinationality and performance.
More studies use scope measures in combination with scale and entropy measures. Goerzen and Beamish (2003) use the same two scope measures as Lu and Beamish (2004). Goerzen and Beamish (2003) however also use an entropy measure. They find a combination of a positive and negative relation. They propose that two dimensions of multinationality have a different effect on performance. International asset dispersion is positively related to performance whereas country environment diversity is negatively associated (Goerzen and
Beamish, 2003). In order to achieve a robust and sensitive measure Qian et al. (2010) use both, a scale- and scope measure, respectively sales-based and subsidiary-based. They find that regional expansion engenders higher performance (Qian et al., 2010). Sullivan (1994) and Rugman and Oh (2013) both use a combination of five measures including two scale measures, two scope measures and an entropy measure. Sullivan (1994) imposes a relationship that is consistent with the IB theories of FDI. Therefore they assume a period of “convergence, decline, reorientation and convergence, which is an organizational evolutionary process” (Li, 2007, p. 122). Within the study of Rugman and Oh (2013) they show that region and industry factors explain most of a firm’s international activity. One of the reasons is that most firms have more than half of their sales and assets within their home region (Rugman and Oh, 2013). They support Rugman and Verbeke’s (2004) regionalization theory and find that MNEs are mostly home region based according to their sales, assets and subsidiary locations (Rugman and Oh, 2013).
As pointed out in this chapter, scholars disagree on the suitable measures for multinationality. Therefore, consensus is not yet established on the M-P relationship (Li, 2007). Furthermore, most studies generalize findings of specific regions. In order to relate the papers, the empirical context of previous inquiries is considered in the next subchapter.
2.4 Empirical context considered
Yang and Driffield (2012) analyze 54 studies that focus on the M-P relationship. The studies range from 1974 until 2008. The majority of these studies base their research on US data (Yang and Driffield, 2012). Yang and Driffield (2012) even conclude that US-based studies include more than 70% of all studies completed on the M-P relationship.
Results of these studies may be biased due to their US-based sample. This has two reasons. First of all, the US is a large economy. Therefore, US companies experience many
exclusive domestic advantages. These advantages might be as high as other advantages gained through international expansion (Yang and Driffield, 2012). Non-US firms are less likely to acquire such economies of scale and scope within their domestic market. Hence, internationalization might be more valuable for non-US companies. Secondly, non-US firms might internationalize for different reasons than US companies. Asia uses internationalization as a way to learn from foreign competitors (Yang and Driffield, 2012). This is less likely for US firms. These factors are important to take into account when studying the M-P relationship. Therefore, the effect of the home region is considered in this study. In the theoretical framework, the home market size moderator is further explained.
2.5 Research gap
Size-dependent factors have received relatively less attention in M-P studies (Benito-Osorio et al., 2016). Therefore, Benito-Osorio et al. (2016) suggest the need for further research on size-dependent factors. In order to contribute to IB research, both moderators used in this study are size-dependent factors.
Li and Yue (2008) recommend that the M-P relationship needs investigation at a lower level of aggregation. This demands that future studies should focus on the contingencies that affect the M-P relationship. Scholars who study the M-P relationship mainly focus on contextual factors that influence this relationship, such as institutions (Benito-Osorio et al., 2016). Consequently, organizational factors received comparatively less attention (Oh and Contractor, 2014). It is however important to research this effect as firm characteristics could be more powerful than contextual factors in influencing firms to choose a certain level of multinationality (Oh and Contractor, 2014). Therefore, this study analyzes the interaction effect of the organizational factor firm size.
Besides the above-mentioned firm characteristic, this study also analyzes the effect of a contextual factor. In this way, this study is able to verify the explanatory power of an organizational factor and a contextual factor. Moreover, Rugman and Oh (2013) mention that M-P research should be conducted within a certain context. The impact of different regions, countries and industries is important for future research to consider (Rugman and Oh, 2013). Home region effects are therefore critical in this debate (Rugman and Oh, 2013). In order to consider a contextual factor in combination with a size-dependent factor, the effect of home market size on the M-P relationship is researched.
As this study used a contextual- and organizational factor, the resource-based view (RBV) and institutional theory are combined. The firm characteristic is generated from the RBV whereas the contextual characteristic derives from the institutional theory. These factors together offer a more comprehensive understanding of the M-P relationship since an internal- and external factor are combined.
Lastly, the horizontal S-curve relation between multinationality and performance has been a relative recent finding. Therefore, research has not paid much attention to the different effects of benefits or costs within the three stages of horizontal S-curve (Benito-Osorio et al., 2016). This study replenishes this gap and clearly shows the positive and negative effects of the two moderators in all three stages of the M-P relation.
3. Theoretical Framework
This chapter elaborates on the theory explained in the literature review. Within this theoretical framework three hypotheses are proposed. First, the relationship between multinationality and performance is hypothesized. Thereafter, the effect of the first moderator, firm size, is argued. Finally, the influence of the second moderator, home market size, on the M-P relationship is discussed. The conceptual framework of all three hypotheses is shown in figure 4.
3.1 M-P relationship
As indicated in the literature review, most recent research proposes a horizontal S-curve regarding the M-P relationship (Benito-Osorio et al., 2016; Oh and Contractor, 2014; Oh et al., 2015; Rugman and Oh, 2010). The first findings on the M-P relationship were however linear. These outcomes were mostly positive linear as studies suggest that multinationality has a positive effect on performance. Thereafter, curvilinear relationships received more attention, since scholars identified costs corresponding to multinationality. As both the U-shaped and inverted U-U-shaped have adherents, scholars are in anticipation for the seamless explanation. Contractor et al. (2003) reason that some studies only capture part of the S-shaped function, which results in a U-S-shaped or inverted U-S-shaped adherence. A U-S-shaped relationship proposes that a firm starts with high performance, which declines to a low point after which it ascends again. In contrast, an inverted U-shaped relationship shows a positive arising movement towards maximum performance, after which this declines again. Depending on what part of the S-curve is examined, studies find positive or negative linear, U-shaped or inverted U-shaped relationships (Contractor et al., 2003).
Chao and Kumar (2010) conclude that large firms only experience stage two and three of the horizontal S-curve. They argue that all firms from their FG500 sample initiated foreign
experience in their past. Therefore, these large firms are already passed stage one. Conversely, even the largest companies could be regionally focused (Osegowitsch and Sammartino, 2008; Rugman and Verbeke, 2004). These firms might not have any international operations at all. For that reason all levels of multinationality could be present in the sample. Oh and Contractor (2014) even emphasize that the three-stage paradigm is a self-evident proposition. Besides being intuitively obvious, the proposition is also empirically verifiable. However, it is only verifiable for a sample in which firms are present with global coverage within all regions, both home and foreign. Benito et al. (2016) complement on that, as they especially find strong support for the three-stage model when the sample consists of large firms. The used sample of the FG500 consists of large firms with global coverage. The horizontal S-curve relationship shows strong empirical validation in recent previous studies (Benito-Osorio et al., 2016; Contractor et al., 2003; Oh and Contractor, 2014). Concluding, the baseline hypotheses of this study are as follows.
Hypothesis 1a: The relationship between the SCALE of multinationality and performance is a horizontal S-curve relationship.
Hypothesis 1b: The relationship between the SCOPE of multinationality and performance is a horizontal S-curve relationship.
3.2 Firm size
Literature acknowledges that firm size influences performance (Lee et al., 2014). Three main factors influence the effect of firm size on performance. These factors are risk bearing, experience and firm resources. Accordingly, these three factors are discussed below, related to their effect on the three stages of the S-curve relationship as hypothesized in hypothesis 1.
During stage one of the horizontal S-curve, multinationality increases whereas performance decreases. In this stage firms encounter costs due to the increase of multinationality. Firm size positively moderates this relationship. This means that the larger the firm size, the more the costs diminish in this first stage (Figure 2). One of the reasons is that small and large firms bear risk differently. Larger firms are usually more diversified and therefore able to bear risk better (Artikis et al., 2007). To be more precise, larger firms are more capable to incur risks associated with internationalizing (Javalgi et al., 1998). They have access to learning channels and stronger bargaining power in host country institutional environments (Kirca et al., 2012). Besides their privileged position with host country institutions, large firms have more power to retain new comers and competitors to enter the market (Kirca et al., 2012). Artikis et al. (2007) reinforce preceding statement and claim that larger firms have a lower default change. For that reason they are able to acquire loans from the banks more easily (Artikis et al., 2007). Consequently, larger organizations are able to reduce transaction costs associated with long-term debt and arrange lower interest rates (Artikis et al., 2007). Since the risks in operating internationally are lower for larger firms, their performance will be higher. Larger firms are better able than smaller firms to decrease the risks involved with an increase in the level of multinationality.
In stage two, the performance increases concurrently with the level of multinationality. The benefits outweigh the costs as firms have overcome the initial stage but have not reached too high risks either. This increase in performance is stronger for larger firms. Two reasons are the foundation of this relation. First, firms with higher levels of multinationality have more experience in foreign countries. Larger firms are more adequate to use this experience to their advantage due to the available knowledge and streamlined processes. Consequently, larger firms reinforce the positive effects of an increase in multinationality on performance. Second, firm resources differ between small and large firms
(Benito-Osorio et al., 2016). Due to the increase in the level of multinationality, larger firms are more capable to relocate their production to lower costs countries (Benito-Osorio et al., 2016). These firms are able to fully exploit investment opportunities and seek needed resources as they have more accessible financial instruments (Benito-Osorio et al., 2016). In contrast, smaller firms are often not in the possession of key resources (Benito-Osorio et al., 2016). Therefore, the larger the firm the more they are able to acquire economies of scale and scope. Higher levels of multinationality increase the available opportunities for firms, which larger firms are able to embrace due to their larger knowledge-pool and resources. These two processes reinforce each other, which results in a positive moderating effect (Figure 2).
In the last stage, performance declines again. As multinationality is at its highest level, risks increase again. However, similar to the first stage larger firms are better equipped than smaller firms to cope with these risks. Due to all the above-mentioned factors, such as experience and knowledge, larger firms are able to diminish the performance decline. Consequently, the decline will be less steep for larger firms. In conclusion, the S-curve relationship moves upwards and to the right due to the positive moderation of firm size (Figure 2). The following hypotheses are probed.
Hypothesis 2a: Firm size positively moderates the relationship between the SCALE of multinationality and performance.
Hypothesis 2b: Firm size positively moderates the relationship between the SCOPE of multinationality and performance.
Figure 2. Firm size positively moderates M-P relationship
3.3 Home market size
Besides the effect of firm size, another size influence has been neglected in M-P relationship studies (Hennart, 2007). This factor is home market size. The home market equals the market of the home country. Frequently, the home country is where the firm’s headquarter is positioned. It is the country with which it is identified culturally, normatively, operationally or by its founding (McGahan and Victer, 2010). The home- and host region are crucial factors influencing the international strategy of a firm (Li and Yue, 2008). Due to differences in home regions, firms pursue diverse international strategies. Various factors of the home region are influential, for example institutions and culture. These factors however, have been intensively researched in this relationship. Nonetheless, the home market size has received less attention.
In order to compute the level of multinationality, it is unsatisfactory to only look at the foreign sales to total sales or global coverage (Hennart, 2007). Multinationality does not only depend on the ratio FSTS or GSTS, as the amount of sales also depends on the home market size (Hennart, 2007). Furthermore, M-P scholars have overlooked the fact that in order to achieve the optimal scale of economies, firms do not essentially need to go beyond
borders. Whenever firms are located in large home markets they may be equally profitable, or even more profitable, in comparison to highly international firms in small domestic markets, without foreign sales (Hennart, 2007). Large domestic markets bring competitive advantages in sectors in which economies of scale or knowledge are generated (Li and Yue, 2008).
Brock et al. (2006) study the moderating effect of the country of origin on the M-P relationship. Within this study they compare US and UK legal firms. For the general M-P relationship they find an inverted U-shaped relationship. However, when splitting the sample in US versus UK firms, the outcome is contradictory. US firms still have an inverted U-shaped relationship, whereas UK firms show a U-U-shaped relationship. Yang and Driffield (2012) confirm the difference in the M-P relationship between US and non-US firms. They show that performance, as a result of multinationality, is higher for US firms. Still, non-US firms show a U-shaped relationship. This means that they have higher results but also incur higher costs at the early stages of internationalization (Yang and Driffield, 2012). The reason for this result is that non-US firms are constrained to their limited home market size. For that reason they might face shortages in resources and are required to go abroad more quickly. Due to the size of the market, US firms are able to expand their business within their country. In contrast to non-US firms, US firms as a result have more experience in expanding their business before going abroad. However, they lack experience in expanding to foreign countries with dissimilar institutions. Firms from smaller countries, such as European firms, do not have this opportunity. They need to immediately go abroad if they want to expand their business. This increases the costs of doing business abroad.
Thus, the size of the home market influences the costs incurred and experience gained before internationalization. US firms experience incremental learning within their home market. The Uppsala model is applicable in this case, which is envisioned by the inverted U-shaped model. Firms within small home markets expand abroad more quickly and therefore
do not experience incremental learning. Contrarily, they experience costs due to LoF. When the home market is large enough, firms do not need to expand abroad to achieve competitive advantages and efficiencies (Hennart, 2007).
Above-mentioned literature emphasizes the importance of the home market size. In order to hypothesize about its effect on the M-P relationship, the moderation effect is scrutinized. The larger the home market, the less attractive it is for a company to engage in business abroad. These companies engage preferably in their domestic market since the risks are kept at a lower level. Furthermore, when the home market size increases, the geographical distance between the domestic market and foreign markets increases similarly. Due to the increase in distance, the costs of doing business abroad grow (Li, 2007). Therefore, an increase in multinationality results in a steeper performance decline when the home market is large. This indicates that in stage one the performance declines even more whereas in stage two the home market size contributes to a lower increase in performance (Figure 3).
Since firms within larger home markets prefer not to engage outside of their domestic market, they are not familiar with different foreign opportunities. Due to this inexperience, the firm encounters a steeper performance decline when they increase foreign activity from stage two to three. The risks for a company with a small home market are lower since the company is more experienced to foreign activity. These companies gained experience and are therefore in the possession of applicable knowledge and tools. Due to the higher experience and foreign involvement, these firms are able to slow down performance decline. However, at the highest level of multinationality, firms with a larger home market size are not experienced to deal with multiple foreign activities at once. This results in a steeper performance decline for firms from larger home markets than for companies from smaller home markets. In conclusion, the home market size negatively moderates the M-P relationship (Figure 3). Below the associated hypotheses are formulated.
Hypothesis 3a: Home market size negatively moderates the relationship between the SCALE of multinationality and performance.
Hypothesis 3b: Home market size negatively moderates the relationship between the SCOPE of multinationality and performance.
Figure 3. Home market size negatively moderates M-P relationship
Figure 4. Conceptual model
Performance
Firm size H2a: Scale H2b: Scope
Home market size H3a: Scale H3b: Scope Multinationality
H1a: Scale H1b: Scope
4. Methods
This chapter illustrates the used methods and analyses. The first section elaborates on the sample and data collection of this study. Thereafter all variables are clarified. Lastly, the analyses and results are discussed.
4.1 Sample and data collection
This study conducts a quantitative research to measure the effect of multinationality on performance. The sample of this research is the Fortune Global 500 listed companies in 2015 and is based on the financial numbers of fiscal year 2014. Accordingly, the level of analysis is the firm. This list ranks the companies with the largest operating revenue in the world. Many scholars use this list when they research multinationality (Osegowitsch and Sammartino, 2008; Rugman and Oh, 2013). They argue that most of these 500 largest companies in the world are MNEs as they engage in foreign activities (Rugman and Oh, 2013). Explicitly, these 500 companies account for over 90% of world’s stock of FDI and over half the world’s trade (Rugman, 2000 in Rugman and Oh, 2013). As this study aims to investigate the international behavior of MNEs worldwide, the FG500 serves as the suitable sample for this study.
Various data are needed in order to conduct this study. To collect all data on the FG500 companies, two sources were used. Most data was available through the Bureau van Dijk (BvD) Orbis database. Orbis offered detailed financial- and historical information on companies located worldwide. This is an advantage over other databases as some databases limit their data to for instance US companies (Compustat) or European companies (Amadeus). However, above-mentioned data were not available for every company. Figures that were not available through Orbis were retrieved through the companies’ annual reports and financial statements of the fiscal year 2014.
In order to test the relationship between multinationality and performance several independent-, dependent-, moderator- and control variables were needed. If the required information was not available, the company was removed from the dataset. This left 188 companies for the scale analysis and 163 companies for the scope analysis.
4.2 Variables
4.2.1 Dependent variable
The dependent variable in this research was the performance of the firm. Return on assets (ROA) was used to measure the performance. The ROA is the ratio of earnings before taxes to total assets and is one of the most commonly used measures for performance (Benito-Osorio et al., 2016; Lu and Beamish, 2004; Rugman and Oh, 2010).
4.2.2 Independent variable
The independent variable was the level of multinationality. Multinationality is reliant on the depth and breadth of internationalization. Thus, the scale and scope of internationalization of the firm determine the level of multinationality. Therefore, a scale and a scope measure were both used to account for multinationality. The scale measure was FSTS as this is the most commonly used method (Benito-Osorio et al., 2016; Li, 2007; Oh and Contractor, 2014). The scope measure was global sales to total sales (GSTS). Rugman and Verbeke (2004) use this measure to compare the sales in the home- versus the host region. This measure shows where the sales of a company take place, which could be in the home-, host-, bi- or global region. For the scope measure, this same approach is used except for one difference. Instead of only considering North America, the EU and Asia, this research also considers South and Central America, Oceania and other regions. Within this study these two specific measures will be referred to as the scale measure and the scope measure.
Using either a scale- or scope measure, leads to discrepancy in the ability to explain the internationalization strategy of a firm. When combining a scale- and scope measure, the coverage of multinationality is more reliable (Hennart, 2011). Furthermore, the scale variable explains for example that 80% of a firm’s sales are not domestically attained but through foreign sales. However, this could still mean that these foreign sales are acquired in the same region or even within the adjacent country. This implies that the company is not global but regionally oriented. Therefore, the scale measure accounts for whether a company is regionally or globally oriented. In conclusion, it is important to include a scale- and scope measure as these constructs reinforce each other in explaining the level of multinationality.
4.2.3 Moderating variables
In this research, two moderating variables were used. These moderating variables were firm size and home market size.
The size of the firm was measured by the total turnover in millions of US (Qian et al., 2010). A logarithm was used in order to further analyze this effect (Benito-Osorio et al., 2016; Qian et al., 2010).
The home market size was measured according to the Domestic Market Size Index of 2013. This index was specified in The Global Competitiveness Report of 2014-2015, published by the World Economic Forum (Schwab, 2015). This report is the most complete valuation of the national competitiveness worldwide (Schwab, 2015). The Domestic Market Size Index was computed as a log of the sum of gross domestic product plus value imports of goods and services, minus the value of exports of goods and services. The outcome was normalized on a one-to-seven scale (Appendix 2).
4.2.4 Control variables
Consistent with control variables used in former studies, this research used the following four control variables: firm age, state-ownership, gearing and industry.
The first control variable was the age of the firm. Older firms are the most dominant players in the market. Moreover, they are more active in foreign activity and therefore gain more experience (Capasso et al., 2015). Due to their experience, they generate higher performances. However, older firms are less able to change their way of doing things and consequently no longer fit in the new environment that could also cause performance decline (Capasso et al., 2015). Since the age of the firm influences a firm’s performance, this construct is included as a control variable. This variable was measured as the number of years after the incorporation date until 2016 (Benito-Osorio et al., 2016). A logarithm of this number was used.
The second control variable was state-ownership. State-owned firms have different motives for expanding abroad than private-owned firms (Amighini et al., 2013). Where state-owned firms expand abroad due to political reasons, private-state-owned firms give preference to economic objectives (Amighini et al., 2013). Thus, the state-ownership was included as a control factor. This variable was converted into a dummy variable. A state owned company was labeled with a value of 1 contrasting to privately held companies that were coded 0.
The third control variable was gearing. This variable shows the level of a firm’s debt in relation to its equity. It also refers to the amount of money that is funded by lenders and which part comes from shareholders. The higher this percentage, the higher the debt of a company. A firm with high debt is considered to be more risky. Firms with higher risks are also more vulnerable as they cannot rely on a safety net if times turn bad. Therefore, they might need to decrease business in order to pay their debt, which could lead to performance decline.
The fourth variable that was controlled for was the industry in which the firm operates. External industry factors can influence the performance of a firm (De Jong and van Houten, 2014). The industry of the FG500 companies was classified in five categories. This was categorized according to their primary industry which is presented by the SIC codes retrieved from ORBIS (Qian et al., 2010). The industries were labeled as followed: 1. Mining, utilities and construction, 2. Automotive, machinery and transport, 3. Professional and information services, 4. Wholesale and retail, 5. Manufacturing and other services. These five categories were coded into four dummy variables. For each category, the company at hand received a value of 1 if it operated in that industry and a value of 0 if it fitted in one of the other industries.
4.3 Analysis and results
This section presents the descriptive statistics and regression models of the dependent-, independent-, moderator- and control variables. Table 4 shows the descriptive statistics and the correlations of all variables used. It is important to detect any correlations with a value above 0.7 as this indicates the existence of problematic variables (Pallant, 2011). All variables showed values lower than 0.7 except for the scale and scope variables. These two variables were strongly correlated (r=.78). As firms with a high degree of internationalization often have a global orientation, this was expected. This caused no problem as the two variables were used in separate models.
Besides looking at the correlations of the different variables, it is also important to check for the less noticeable forms of multicollinearity. Variance Inflation Factors (VIF) and Tolerance levels are mostly used to test the more imperceptible forms of multicollinearity. Tolerance is a measure that shows how much of the variable of the indicated independent is not explained through the other independent variables (Pallant, 2011). VIF is the inverse of
Tolerance. It would be problematic if the VIF values are above 10 and Tolerance is lower than 0.1 (Pallant, 2011). The variables all had a VIF value between 1.1 and 3.0. Correspondingly, the Tolerance level was far above 0.1 for all variables. Concluding, there was no existence of multicollinearity within this study.
The dependent variable, ROA, showed a mean of 5.10. This number illustrates the efficiency of managers as it indicates how much return is generated using its assets. The scale and scope variables gave insights into the level of multinationality, the independent variable. Within the dataset, 248 companies had available data on the scale measure. The mean of the scale measure was 0.51. This indicates that the FG500 is quite international as on average 51% of their total sales is in foreign countries. Data for the scope measure was available for 205 firms. The scope measure showed a mean of 0.43, which indicates that on average 43% of the sales is generated outside the home region. The first moderator, firm size, showed a mean of 62,424 billion US dollars in revenue. Since this sample included the largest firms worldwide, this high average was not out of the ordinary. The home market size was the second moderator and showed a mean of 6.12. Since the scale was between one and seven, this mean was rather high. Almost half of the 500 firms were located in the US or China (Table 3). Since these are the two largest markets in the world, this mean was explainable.
In order to test the effect of multinationality on performance, a hierarchical regression analysis was used. The regression analysis consisted of two separate analyses. The first computed the scale of multinationality (Table 5) whereas the second used the scope of multinationality (Table 6) as independent variable. Both tables consist of 6 models. The models are described below.
Table 3. FG500 according to home country and home market size
Country Frequency In percentages (%) Domestic Market Size Index
United States 128 25.6 7 China 97 19.4 6.8 Japan 54 10.8 6.1 France 31 6.2 5.6 Germany 28 5.6 5.8 Great Britain 28 5.6 5.7 Korea 17 3.4 5.4 The Netherlands 14 2.8 4.7 Switzerland 12 2.4 4.3 Canada 11 2.2 5.4 Others* 80 16 n.a. Total 500 100
Countries that appear less than 10 times in the FG500 list
The first step in the regression analysis was to test the effect of the control variables on the dependent variable. The results are shown in model 1 of both table 5 and 6. Thereafter, the independent variable was added with which the S-curve was tested in model 2, 3 and 4. Model 2 adds the linear term of multinationality, model 3 includes the squared term and model 4 uses the cubic term. All four models were computed with the linear regression function. An Ordinary Least Squares (OLS) regression was used since the dependent variable was a continuous value. Interaction terms were used in order to test the hypothesized moderations. Accordingly, multiplying the independent variables by the moderator created the interaction terms. The firm size, defined by hypothesis 2, is shown in model 5. Model 6 illustrates the moderation effect of home market size, which denotes hypotheses 3.
The most important values to consider are the significance value, the unstandardized b-coefficient and the R-squared value. The significance value specifies whether the results are reliable and therefore indicates whether the hypotheses can be supported or rejected. The b-coefficient shows the change in the dependent variable as a result of the independent
