Performance Differences between Portuguese Domestic Firms and Portuguese Multinational Firms

Performance Differences between Portuguese Domestic Firms and Portuguese Multinational Firms por João Nuno da Silva Santos Dissertação de Mestrado em Economia e Gestão Internacional Orientada por Ana Teresa Cunha de Pinho Tavares Lehmann 2014 i Biographic Note João Nuno da Silva Santos was born in Figueira da Foz in November 9th, 1988. He graduated in Economics from the University of Aveiro in 2011 and entered in MSc in International Business, Faculdade de Economia da Universidade do Porto, in 2012. Throughout his academic years he participated in student associations since his freshman year. Professionally, he worked in a summer internship in Caixa Geral de Depósitos (2010) and another one in Jumbo Hipermercados – Aveiro (2012), did also Erasmus in Czech Republic (2009/2010) and international volunteering in Ukraine / Crimea (2013) issued by AIESEC (in FEP; in Kiev; in Simferopol). ii Acknowledgements Since I am the product of all of my experiences and of everyone I’ve ever known, I’d like to use this section to make some acknowledgements to some people without whom this dissertation wouldn’t be possible. First to my professor and thesis supervisor Ph.D. Ana Teresa Tavares Lehman for the incredible professionalism and patience demonstrated ever since our very first meeting. Thank you for all the help and commitment! I’d like to thank Ph.D. Frederick Lehmann for spending his limited time helping me with STATA and for his noteworthy inputs on this dissertation. To all the professors and colleagues from MEGI from whom I’ve learnt so much and allowed me to have such nice experience in FEP. To all my friends from Aveiro, for the unconditional support and camaraderie. I couldn’t even describe the value you represent to me, and still you’d understand it. To all my friends from Figueira, I know I’m never around, but you are always with me. The most special acknowledgement goes to my entire family, especially my parents and my sister. You are the anchor that keeps me from drifting away. Thank you for investing in me more than you have. Finally, to my grandfather Frutuoso, grandmothers Esmeraldina and Fernanda and to my uncle José Carlos that sadly left us too early to see me concluding this chapter of my life. You were, are and will always be an important part of my life. iii Resumo Embora haja vários estudos que analisam o desempenho das empresas, esta dissertação difere desses estudos na medida em que vem dar um contributo para preencher uma lacuna na literatura que se foca na comparação entre empresas que se internacionalizam para outros países através do Investimento Direto Estrangeiro e empresas que se mantêm no país de origem. Mais ainda, esta dissertação distingue-se também por fazer esta comparação focando-se no panorama português num cenário empírico alargado. Pretende-se, portanto, fazer uma comparação ao nível do desempenho entre empresas portuguesas multinacionais e empresas portuguesas meramente domésticas de modo a averiguar se há, ou não, vantagens em internacionalizar e, caso hajam, em que medidas de desempenho as vantagens / diferenças são mais visíveis. O enquadramento teórico da dissertação baseia-se nas teorias do IDE e das empresas multinacionais (em suma, Negócios Internacionais), tais como o contributo de Hymer e o seu conceito de vantagem, o Paradigma Eclético, entre outros. O estudo empírico foi feito utilizando dados extraídos da base de dados SABI (Sistema de Análise de Balanços Ibéricos) e foram aplicados modelos de regressão linear (Pooled OLS) e também em cross-section, considerando tanto medidas de Rentabilidade (Profitability – ROS; ROA; ROE; Profit Margin) e de Produtividade (Gross Value Added per Employee). Os resultados sugerem que as Empresas Multinacionais têm melhor desempenho do que as Empresas Domésticas e que quanto maior for o envolvimento internacional, maior será a sua performance. Foi encontrada evidência de que estas diferenças de desempenho não são iguais entre Produtividade e Rentabilidade. A diferença de Produtividade é muito maior do que a diferença de Rentabilidade. Este estudo apresenta ainda algumas contribuições em termos de políticas a serem aplicadas. Palavras-chave: Negócios Internacionais; Investimento Direto Estrangeiro; Empresas Multinacionais; Empresas Domésticas; Performance; Lucro; Produtividade; Portugal. iv Abstract Although there are many studies concerning firms’ performance, this dissertation comes with the purpose of contributing to fill a gap in the research area that deals with the comparative performance analysis in terms of firms with outward Foreign Direct Investment (OFDI) vis-à-vis others that do not undertake outward FDI. Moreover, this dissertation focuses on Portugal, on widen-researched empirical setting, in this regard. Therefore, this dissertation aims to make a comparison, performancewise, between Portuguese multinational firms and Portuguese domestic firms in order to ascertain whether there are, or not, performance differences and, if there are, in which measures of performance these differences are more notable. The theoretical background for this dissertation is based on FDI and the multinational enterprise theories (i.e., International Business theories), such as Hymer’s concept of advantage, the Eclectic Framework, among others. The empirical part uses data extracted from SABI’s database and the methodology will include an econometric study with Pooled-OLS and Cross-Sectional OLS, considering some Profitability measures (ROS, ROA, ROE, Profit Margin) and Productivity measures (Gross Value Added per Employee). The results suggest that Multinational Enterprises have a better performance than Domestic Enterprises and that the more internationalized the firm is, the better it will perform. Evidence was found that these performance gaps differ across Productivity and Profitability measures. The Productivity gap between Multinational and Domestic Enterprises is bigger than the Profitability gap. This study also presents some contributions for policy making on the matter. Keywords: International Business; Foreign Direct Investment; Multinational Enterprises; Domestic Enterprises; Performance; Profitability; Productivity; Portugal. v Table of Contents Biographic Note ................................................................................................................ i Acknowledgements .........................................................................................................ii Resumo ........................................................................................................................... iii Abstract ........................................................................................................................... iv Table of Contents ............................................................................................................ v Index of Tables ..............................................................................................................vii Index of Acronyms ...................................................................................................... viii Introduction ..................................................................................................................... 1 Chapter 1. Theoretical Background .............................................................................. 3 1.1. The Concept of Advantage .................................................................................... 3 1.1.1. Hymer’s Contribution ..................................................................................... 3 1.1.2. Dunning’s Eclectic Paradigm ......................................................................... 4 1.1.3. Internalization Theory ..................................................................................... 5 1.1.4. Resource-based view of the Firm and Dynamic Capabilities ......................... 6 1.1.5. Network Theory .............................................................................................. 8 1.1.6. Final Remarks ................................................................................................. 9 Chapter 2. Literature Review ...................................................................................... 10 2.1. Firm Performance Measures ................................................................................ 10 2.1.1. Profitability ................................................................................................... 10 2.1.2. Productivity ................................................................................................... 19 2.1.3. Other Measures ............................................................................................. 21 2.2. The internationalization-performance relationship .............................................. 22 2.3. MNE vs. DE Performance ................................................................................... 23 2.3.1. MNEs have superior performance than DEs ................................................ 24 2.3.2. DEs have superior performance than MNEs ................................................ 24 2.3.3. Inconclusive cases ......................................................................................... 25 Chapter 3. Methodology ............................................................................................... 27 3.1. Data ...................................................................................................................... 27 3.2. Empirical Model .................................................................................................. 29 vi 3.3. Variables and Proxies .......................................................................................... 30 3.3.1. Dependent Variables ..................................................................................... 30 3.3.2. Independent Variables .................................................................................. 31 3.3.3. Descriptive Analysis and Correlations ......................................................... 34 3.4. Econometric Models and Empirical Results ........................................................ 37 3.4.1. Pooled-OLS .................................................................................................. 37 3.4.2. Cross-Section Model ..................................................................................... 41 Chapter 4. Conclusions and Policy Implications ....................................................... 58 4.1. Conclusions .......................................................................................................... 58 4.2. Policy Implications .............................................................................................. 59 4.3. Future research ..................................................................................................... 59 References ...................................................................................................................... 60 Annex I – List of Empirical Studies ............................................................................ 69 vii Index of Tables Table 1 - Literature Review on Performance Measured by Profitability Measures .................... 16 Table 2 - Literature Review on Performance Measured by Productivity Measures ................... 20 Table 3 - Internationalization-Performance Relationship Mainstream Perspectives .................. 23 Table 4 – Manufacturing subsectors and observation count ....................................................... 28 Table 5 - Variables ...................................................................................................................... 33 Table 6 - Descriptive Statistics ................................................................................................... 35 Table 7 - Correlation Matrix ....................................................................................................... 36 Table 8 - Pooled-OLS ................................................................................................................. 38 Table 9 - Pooled OLS, Only Manufacturing ............................................................................... 39 Table 10 - Pooled OLS, Manufacturing, Subsectors ................................................................... 40 Table 11 - LogGVAEMP Cross-Section ..................................................................................... 43 Table 12 - LogPM Cross-Section ................................................................................................ 44 Table 13 - LogROE Cross-Section ............................................................................................. 45 Table 14 – LogROA Cross-Section ............................................................................................ 46 Table 15 - LogROS Cross-Section .............................................................................................. 47 Table 16 - LogGVAEMP Cross-Section, Manufacturing ........................................................... 48 Table 17 - LogPM Cross-Section, Manufacturing ...................................................................... 49 Table 18 - LogROE Cross-Section, Manufacturing .................................................................... 50 Table 19 - LogROA Cross-Section, Manufacturing ................................................................... 51 Table 20 - LogROS Cross-Section, Manufacturing .................................................................... 52 Table 21 – LogGVAEMP Cross-Section, Manufacturing with Subsectors ................................ 53 Table 22 – LogPM Cross-Section, Manufacturing with Subsectors ........................................... 54 Table 23 - LogROE Cross-Section, Manufacturing with Subsectors ......................................... 55 Table 24 - LogROA Cross-Section, Manufacturing with Subsectors ......................................... 56 Table 25 - LogROS Cross-Section, Manufacturing with Subsectors .......................................... 57 viii Index of Acronyms CAPM: Capital Asset Pricing Model DCT: Dynamic Capabilities Theory DE: Domestic Enterprise EBIT: Earnings Before Interests and Taxes EU: European Union FDI: Foreign Direct Investment GDP: Gross Domestic Product GVA: Gross Value Added HFDI: Horizontal Foreign Direct Investment IJV: International Joint-Venture M&A: Mergers & Acquisitions MNE: Multinational Enterprise NACE: Statistical Classification of Economic Activities in the European Community OECD: Organization for Economic Co-operation and Development OLS: Ordinary Least Squares OPSAL: Ratio of Operating Cost to Sales R&D: Research & Development RBV: Resource-based View of the Firm ROA: Return on Assets ROE: Return on Equity ROIC: Return on Invested Capital ROS: Return on Sales SABI: Sistema de Análise de Balanços Ibéricos SMEs: Small and Medium Enterprises TFP: Total Factor Productivity VFDI: Vertical Foreign Direct Investment 7 advantage. Thus it must be unique/rare so that the firm can implement a strategy (based on that resource) that cannot be accompanied by the competition; Hard to imitate resources allow the firm to be innovative enabling itself to conceive new ground breaking strategies; Non-substitutable resources - however the difficulty to imitate a firm’s strategy or duplicate its competitive advantage, a competitor may be able to create a new different way to deal with this threat, developing its own resource/strategy (Barney, 1991). The RBV approach states that a firm’s specific advantage is based on its resources (tangible and intangible), ergo, the exploitation of firm-specific assets and also on the creation of new ones (Wernerfelt, 1984). In the process of overcoming their liability of foreignness when internationalizing, MNEs have to figure out the timing and method of entry (Peng, 2001). As they grow successfully global, these questions start to be answered based on the firm’s international experience which can be viewed as an intangible resource. This learning process is fastened when it’s done through Mergers & Acquisitions and Strategic Alliances with local partners which “facilitate local knowledge acquisition and strengthen firm performance” (Peng, 2001: 812), giving the MNE the edge on performance differences. Building up on the RBV, the Dynamic Capabilities theory (DCT) also sees the knowledge, know-how and managerial skills itself as an important resource to the firm’s competitiveness as it is valuable, rare and hard to imitate (and also because of its complexity) which makes it crucial to the competitive advantage creation. “Dynamic Capabilities” are the ability to adapt, change or integrate managerial skills, resources or competences in a dynamic and timely fashioned way (Teece & Pisano, 1994; Teece et al., 1997). In other words, they refer to the capabilities that will allow the firm to adapt itself to changing market circumstances. These capabilities can be divided into sub categories: Processes – the managerial and organizational routines and practices, integration, learning, reconfiguration and transformation; Positions – the firm’s technological property, customer and supplier relations, technological financial complementary and locational assets; and Paths – the strategic alternatives and opportunities (Teece & Pisano, 1994). 8 In contrast to the Internalization Theory, that uses market failure as an explanation for the internalization of technology in the internationalization process, the DCT explains the same process with the level of coding capacity and the difficulty to teach or transfer each capability or resource, as knowledge or technology (Kogut & Zander, 1993; Teece & Pisano, 1994). This theory asserts that “which is distinctive cannot be bought and sold short of buying the firm itself, or one or more of its subunits” (Teece & Pisano, 1994: 541). On a global environment and in some cases with hyper competition, the survival of the MNE would be sustained essentially on its ability to exercise difficult-to-imitate dynamic capabilities. This not only addresses the fast “innovation, adaptation and flexibility”, but also the “importance of proactive entrepreneurial behaviour” of the MNE (Augier & Teece, 2007: 185). This implies that the firm must have some capabilities that allow it to have a better response to market competitive demands. Because the MNE operates on a global market, and as constant adaptation to market circumstances is part of the managerial routine, they are able to obtain “superior [...] performance over multiple product life cycles.” (Augier & Teece, 2007:188). 1.1.5. Network Theory In Internalization Theory, the firm develops an intangible firm-specific advantage that gives the firm benefits on the in-house production rather than on market solutions. The Network Theory, contrarily to the “staged” and gradual internationalization theories (Johanson & Vahlne, 1977), states that these development activities depend on the relationships with other firms, that is, on the network position of the firm and its relationships with partners (Coviello & Munro, 1995). This theory features not only the firm but also the network created by itself and customers, suppliers, other business partners and their cooperative relationships (Hadley & Wilson, 2003) whether they’re industrial (formal) and social (informal) which helps the firm achieving higher growth rates (Coviello & Munro, 1995; 1997). On the other hand, a firm may become dependent on the network to create business opportunities, so the 9 network may facilitate or inhibit a firm’s internationalization forcing it to diversify outside the network (Coviello & Munro, 1995; Chetty & Holm, 2000). In the Network approach the MNE can “externalize some of its activities without losing control of its crucial intangible assets” (Johanson & Mattsson, 1988:308), eliminating the opportunistic behavior linked with other contractual entry modes. Based on this approach, the internationalization process can be done through (i) international extension – network created with local firms; (ii) penetration – development of an already existing network; and (iii) international integration – coordination of different networks (Johanson & Mattsson, 1988). Also, when considering the Network Theory one ought not to ignore the internal network that constitutes a MNE. The MNE itself, as it is spread out geographically, constitutes an integrated web of operations which “represent an important source of innovation” as they have the “ability to sense diverse market needs, technological trends, and competitive actions” (Bartlett & Ghoshal, 1998: 102). As suggested by Zanfei (2000), MNEs gain access to local knowledge and capabilities, through affiliates located abroad and their relations. In fact this might be one way to overcome the costs of foreignness. Given that a MNE is an international web of (internal and external) relations, i.e. network, they can reach to more sources of innovation that will allow them to perform more efficiently. Hence, networks may be beneficial to innovation and to the technological development of the firm. 1.1.6. Final Remarks We have asserted that a firm, in order to be successful in its internationalization process, must have some internal advantage(s) that would help them overcoming their liability of foreignness. Dunning’s Eclectic Paradigm gives us an important approach to why FDI occurs, explaining that the firm should have, once again, some ownership, location and internalization advantages simultaneously. The ownership advantages, that are central to this study, like a firm’s international experience and multinationality, reflect the ability to keep internationalizing successfully. The internalization theory provides us an explanation to why FDI occurs (that is also central to the Eclectic 10 Paradigm) and to when it is better to invest abroad through internal operations instead of contractual modes. So when FDI occurs, it’s because there are advantages in doing it, thus explaining the idea that there must be differences between those firms who internationalize through FDI and those who don’t. As to the RBV and Dynamic Capabilities, this approach states that in the firm’s resources and capabilities are the main competitive advantage. Through an internationalization and geographic spreading process, these resources tend to develop themselves as the firm grows its knowledge on the foreign market. The Network theory builds upon this as well. As the firm uses its internal and external networks to benefit from their knowledge on foreign markets, they are introduced to new networks. This will ultimately result in a great expansion not only in the business connections and partners, but also in a firm’s ability to benefit from it. Chapter 2. Literature Review 2.1. Firm Performance Measures Before jumping to the literature review of the relationship between internationalization and performance, we want to first show what kind of proxies were used in previous studies. In this literature review, we found that performance can be measured in many different ways – through profitability measures, productivity measures or others. 2.1.1. Profitability Concerning profitability, the most used proxies were the financial ratios such as Return on Sales (ROS), Return on Assets (ROA) and Return on Equity (ROE). We also found other profitability measures such as Gross Profit Margin (Elango, 2006), Tobin’s Q (Chari et al., 2007), EBIT (Chen & Hsu, 2010) that will also be presented. Concerning ROS, Grant (1987) when he studied British firms and their multinationality and performance, used different measures do assess the performance of 11 his sample. One of the measures used as proxy was ROS and the result was that overseas production to total sales ratio had positive impact on firms’ ROS and that multinationality has positive impact on performance on a thirteen-year period. Geringer et al. (1989) using ROS to assess the performance implications of diversification and internationalization strategies for US and European MNEs found that there is indeed a positive relationship between internationalization and firm performance but only until a certain point – inverted u-shaped relation which will be addressed later. Luo & Tan (1998) used also ROS and ROA on their comparison between MNEs and DEs performance, concerning the strategic choice (defensive; prospector; analyzer)1 when internationalizing, in an inward perspective of FDI and found that local firms adapt themselves to the entry of foreign firms. Foreign firms can keep their levels of performance, domestic firms improve their performance because they don’t take the internationalization risks and have to innovate in order not to keep competing. Lu & Beamish (2001) with a study focused on Japanese Small and Medium Enterprises (SMEs) found that exporting had a negative linear relationship with performance, mostly because of the appreciation of japanese Yen and that FDI had a nonlinear relationship with performance – u-shaped. Capar & Kotabe (2003) with a sample of 81 service firms found that international diversification has positive impact on performance after a certain level/weight of foreign operations on the firm’s total operations (multinationality). Contractor et al. (2003) found a three-staged relationship between performance and multinationality, using ROS as measure of performance for 103 firms. Qian et al. (2003) also studied the SMEs in a four-year period using ROS as measure and found that MNEs outperformed domestic firms and concluding that higher the international involvement is, the better they will perform. Brock et al. (2006) with a sample of law firms from the USA and UK found different patterns of effects in those countries but the overall effect, homogeneous to the entire sample, was that international diversification had positive impact on performance after a certain point of 1 The “Prospector” is a high risk strategy focuses on product development by “scanning, identifying, and capitalizing on (…) market opportunities” (Luo & Tan, 1998:24). It’s often connected to first mover strategy and firms adopting this strategy are always looking for market opportunities (Miles & Snow, 1978). The “Defensive” strategy consists on the opposite of the Prospector strategy as the firm maintains its secure and stable position in the market rather than advancing to new product development programs or searching for new opportunities (Miles & Snow, 1978). The “Analyzer” strategy is the place in between the previous strategies. Less risks than prospector strategy but also less commitment to a stable position. 12 foreign involvement. Besides ROS, they also used Profits per Equity Partner as performance measure. Chiao et al. (2006) also using ROS found that internationalization has positive impact in firm’s performance in both industries studied (Electronics and Textile) which also showed a nonlinear relationship. Coombs & Bierly (2006) in their study on technological capability and its implications on performance, using a sample of 201 manufacturing firms and combining it with many proxies, found that investing in technology development would increase shareholder return. Their study has a different object and purpose of our own in terms of multinationalityperformance relationship, but it does provide a different perspective on performance measurement. Contractor et al. (2007) analyzed the relationship between international expansion and performance, in India, using not only ROS but also ROA and ROE as performance measures. Their findings were consistent with the U-shaped relationship in the manufacturing firms’ sample. Return on Assets is another ratio used as performance measure. Buckley et al. (1984) used this measure (Net income to Assets ratio) as a proxy for profitability when analyzing the growth of firms between 1972 and 1977. As result they concluded that multinationality does not necessarily have an impact on performance. Grant (1987) also used this ratio but in a different version (“Pre-tax, pre-interest profits as percentage of Net Assets”, Grant, 1987:84) but also responded positively to a change in overseas production. Geringer et al. (1989) also used ROA which revealed the same results as ROS: the degree of internationalization has positive impact but only until a certain point, then it stops being productive and the performance starts to decay – inverted ushaped relationship. Luo & Tan (1998) whose study was already referred in the previous paragraph, used ROA as performance and arrived at the same results for both measures. Gomes & Ramaswamy (1999) also using ROA (among others measures) stated that multinationality brings positive performance impact but only until the optimal point. Lu & Beamish (2001), having both ROS and ROA highly correlated, confined the results to ROA having reach the conclusion that there is a nonlinear impact of multinationality on performance. Kotabe et al. (2002) on their study focused on R&D and its role concerning performance and multinationality, used ROA as proxy to measure financial performance and found that multinationality not only impacts positively on performance, but also that this impact is due to R&D and Advertising 13 expenditures and intensity. Contractor et al. (2003) also used ROA reaching the same results that were discussed in the previous paragraph as there was a certain degree of collinearity between ROS and ROA. Lu & Beamish (2004) analyzed the veracity of the s-curve hypothesis applied to Japanese firms. Using ROA, they found that there is consistency for their hypothesis and also that the higher the investment in technology and in advertising, the higher the profitability. Barbosa & Louri (2005) in their comparison between foreign-owned and domestic-owned firms in Greece and Portugal concerning the importance of ownership in performance, using ROA as measure, found that in Greece foreign-owned firms perform much better than domestic-owned. Coombs & Bierly (2006), as said before, used many different proxies when measuring the impact of technological capability on performance and to what concerns ROA their findings were that the ratio R&D Expenditures to Total Sales had negative impact on performance. Contractor et al. (2007) found that, for ROA, for manufacturing firms the higher Foreign Sales to Total Sales ratio, the lower the firms perform, as the opposite goes for services firms. So internationalization benefits more the Indian services firms than the manufacturing ones. Kimura & Kiyota (2007), using ROA and other measures, found that foreign-owned firms have superior static indicators and are able reach higher and faster levels of growth when comparing with domestic firms. Using different proxies, Adenaeuer & Heckelei (2011) analyzed the connection between FDI and performance of European agribusiness firms. Concerning ROA, they found no difference of performance. Return on Equity is the last of the most popular/used ratios. Grant (1987) also used this measure and results are not different than the previous ones (see two previous paragraphs). Chiang & Yu (2005) on their study focused on Taiwanese firms concluded that there is also a positive impact until a certain point. Coombs & Bierly (2006) state that although they use ROE as performance measure, it can be influenced by both changes in debt and equity as well as in the interest rate paid on debt making it difficult to understand if its behavior. Hsu (2006) also studied the s-curve hypothesis on 55 pharmaceutical companies in a four-year period. Using ROE as measure, the main finding was that companies benefit from internationalization activities but could not prove the existence of the s-curve for his sample. Contractor et al. (2007) found a proportional relationship between Foreign Sales to Total Sales ratio and ROE, being 14 that this relationship is nonlinear. Kimura & Kiyota (2007) and Adenaeuer & Heckelei (2011) used this measure as well and the results were presented in the previous paragraphs, as they don’t differ across measures. Other profitability measures were found in studies like the one carried out by Michel & Shaked (1986) which used market-base measures such as the Risk-adjusted Return using the Sharpe, Treynor and Jensen measures2 and applying the Capital Asset Pricing Model (CAPM) to obtain the betas. As result they found that although MNEs are bigger in terms of firm size, it does not explain the findings in which DEs present higher risks suggesting that the latter ones have better performance. Collins (1990) also used Sharpe, Treynor and Jensen measures to assess the Risk Return in a comparison between US firms that were active in domestic, developed and developing countries. He found that firms operating on domestic market and developed countries got higher returns and higher risk measures, hence higher performance. He also found that there were higher rates of return for FDI and that there was no benefit in diversifying to developing countries as they presented low risk and low return. Benvignati (1987) used a rate of profit calculated using Operating Income in each line of business minus the estimated capital costs divided by sales. The main finding of this study was that multinational firms have higher profits than the firms operating solely in domestic industries. Gomes & Ramaswamy (1999) in order to account for Operational Outcomes used “a ratio of operating cost to sales (OPSAL)” (Gomes & Ramaswamy, 1999:181) and also found a nonlinear relationship. Kotabe et al. (2002) besides using ROA, also used Sales to Operating Costs ratio to evaluate the Operational Outcome and the results were similar to the ones concerning ROA in the same study. Elango (2006) in order to assess the impact of internationalization on performance used Gross Profit Margin as profitability measure. He analyzed this relationship for 12 emerging markets and the main findings were a nonlinear relationship for manufacturing firms and a positive linear relationship for services firm. Some authors used Tobin’s Q as their proxy. Lu & Beamish (2004) studied the impact of international diversification on performance and also used ROA and the 2 Sharpe and Treynor measures “determine the premium of a security’s return per unit of risk” and Jensen “evaluates the difference between the security’s expected return and its actual return” (Michel & Shaked, 1986:93). 15 results were that higher advertising expenditures and investment in technology lead to greater profitability. Chari et al. (2007) also analyzed the importance of technology investments on the relationship between international diversification and firm performance and found that multinationality impacts positively in a firms’ performance, especially if it’s a firm with high technology investment. Chen & Hsu (2010) using Earnings Before Interest and Taxes (EBIT) as proxy for performance measurement, analyzed a sample of Taiwanese firms in a 5 year period and found a nonlinear relationship between internationalization and performance and also between advertising expenditures and performance. They also found that R&D expenditures have positive impact on performance. Adenaeuer & Heckelei (2011), as said before, used many different measures and found that firms undertaking FDI had better Revenues, Profits before Taxes and Profit margins than the domestic counterparts. These results were the same for other proxies such as Return on Invested Capital (ROIC). Chang & Rhee (2011) also used ROIC to evaluate how important was the speed of the internationalization process in terms of firm performance. They found that rapid FDI has no main effect on firm performance unless the firm has high marketing capabilities or strong brand equity, and that fast international expansion is more favorable for industries facing intense global competition. Assaf et al (2012) using Total Costs logarithmized found that cost efficiency could increase performance if the internationalization process was undertaken majorly through Mergers & Acquisitions (M&A) and done in an early stage of internationalization process. So it is a remark on how fast the internationalization process should be. They also found a counterproductive home country effect, i.e., if home country GDP increases, there would be lesser gains from the internationalization process in terms of performance. The Table 1 sums the studies that used Profitability as measure and their results as well. 16 Table 1 - Literature Review on Performance Measured by Profitability Measures Authors Year Title Sample Measures Used Result Adenaeuer & Heckelei 2011 FDI and the performance of European Agribusiness Firms 1687 firms with plants only in EU-15; 314 firms with in and out of EU-15 Profitability & Productivity Advantage MNE Chen & Hsu 2010 Internationalization, Resource Allocation and Firm Performance 224 Taiwan Stock Exchange-listed electronics & IT firms (2000-2005) Profitability Advantage MNE Chari, Devaraj & David 2007 International diversification and firm performance: Role of Information Technology Investments 131 firms, 1997 Profitability Advantage MNE Contractor, Kumar & Kundu 2007 Nature of the Relationship between international expansion and performance: The case of emerging markets firms 269 Indian firms, 19972001. 142 manufacturing; 127 services Profitability Advantage MNE Kimura & Kiyota 2007 Foreign-owned vs. Domesticallyowned firms: economic performance in Japan 22.250 firms (21.716 DEs; 534 MNEs) between 1994 and 1998) Profitability & Productivity Advantage MNE Brock, Yaffe & Dembovsky 2006 International diversification and performance: a study of global law firms 76 US firms; 13 UK firms (2003) Profitability Advantage MNE Chiao, Yang & Yu 2006 Performance, Internationalization and Firm-Specific Advantages of SMEs in a NewlyIndustrialized Economy 1419 Taiwanese SMEs (1996). 818 electronics industry + 601 textile Profitability Advantage MNE Elango 2006 An Empirical Analysis of the InternationalizationPerformance Relationship Across Emerging Market Firms 719 firms from 12 emerging markets. 393 manufacturing, 326 services Profitability Advantage MNE 23 Table 3 - Internationalization-Performance Relationship Mainstream Perspectives Theory Graphic Shape Performance Measure Reference Linear Relationship Sales Growth; RONA; ROS; ROE Grant (1987) ROA Grant et al. (1988) ROS Tallman & Li (1996) Gross Profit Margin Elango (2006) U-shaped Relationship ROE; ROA; Pretax Operating Margin Mathur et al. (2001) ROS Capar & Kotabe (2003) ROA Ruigrok & Wagner (2003) ROA; ROE; ROS Contractor et al. (2007) EBIT Chen & Hsu (2010) Cost Efficiency Assaf et al. (2012) Inverted Ushaped Relationship ROS Brock et al. (2006) ROE Chiang & Yu (2005) ROS Chiao et al. (2006) Gross Profit Margin Elango (2006) ROS; ROA Geringer et al. (1989) ROA; Ratio of Operating Costs to Sales Gomes & Ramaswamy (1999) S-shaped Relationship ROS;ROA Contractor et al.(2003) ROE; ROA Thomas & Eden (2004) Source: Own elaboration based on Cardoso (2008). 2.3. MNE vs. DE Performance Concerning performance gaps between MNEs and DEs, and to the best of our knowledge, there are few studies that focus on the comparison between MNEs and DEs concerning OFDI. Nevertheless, there are those which compare these two in different situations (e.g. Imbriani et al., 2011; Hayakawa et al., 2012). The studies we found and analyzed come up with three outcomes: (a) MNEs have better performance than DEs; (b) DEs outperform MNEs and (c) some cases were inconclusive or with some complex 24 interpretation. Thus we divided this section by those three outcomes and presenting, lastly, a table summarizing our performance comparison literature review. 2.3.1. MNEs have superior performance than DEs From the studies included in Annex I, the majority of them presented the conclusion that MNEs outperform DEs. Some of these studies focused their research on Asian countries (Ramstetter, 1999; Lu & Beamish, 2001, 2004; Chiang & Yu, 2005; Chiao et al., 2006; Contractor et al., 2007; Kimura & Kiyota, 2007; Chen & Hsu, 2010; Hayakawa et al., 2012), other used European firms (Davies & Lyons, 1991; Capar & Kotabe, 2003; Girma et al., 2004; Anastassopoulos et al., 2007; Temouri et al., 2008; Imbriani et al., 2011) or US firms (Brewer, 1981; Benvignati, 1987; Grant, 1987; Lee & Kwok, 1988; Geringer et al., 1989; Qian et al., 2003; Brock et al., 2006). Besides Imbriani et al. (2011) which used a binary dependent variable (to be or not a multinational) with matching techniques as main methodology, almost all used Profitability measures as proxy for performance. Others used Productivity (Davies & Lyons, 1991; Ramstetter, 1999; Girma et al., 2004; Temouri et al., 2008; Hayakawa et al., 2013) or both (Anastassopoulos et al., 2007; Kimura & Kiyota, 2007; Adenaeuer & Heckelei, 2011). 2.3.2. DEs have superior performance than MNEs Concerning the studies which found the opposite of the previous ones, we have a total of three. Michel & Shaked (1986) used the Risk-adjusted Return to analyze the differences between MNEs and DEs and found that DEs have superior risk-adjusted performance, thus having higher total and systematic risk providing a higher return, for all their measures (Sharpe, Treynor and Jensen). Kim & Lyn (1990) used some financial and accounting-based ratios (Earnings per Share; ROE; Gross Profit Margin and Operating Profit Margin) to assess if foreignowned firms operating in US enjoyed advantages over US firms operating solely in their 25 domestic market. Although US firms spend less in R&D and in Advertising than foreign-owned firms, they tend to be more efficient than the latter. Mathur et al. (2001) with a sample of Canadian firms studied differences between local and multinational firms. Using ROE, ROA and Pretax Operating Margin they found that DEs have better performance for all the proxies in all of the sample time period. They also tested for internationalization effect on firm performance and found a nonlinear relation between these two similar to the u-shaped theory. 2.3.3. Inconclusive cases There were some cases we found whose final conclusion wasn’t as consistent as they should be or couldn’t be just labeled as only advantage for one side for having many conditions to its conclusions. Buckley et al. (1984) studied the growth and profitability of many US and non-US firms in the 1970s. Using sales growth rate and profitability (proxied by Net income to asset ratio), they’ve reached some inconsistent results being that for one of the sample’s years the results were insignificant for the full sample. So multinationality couldn’t even be accounted as contributor for the variance in growth and profitability of both sets of firms. Al-Obaidan & Scully (1995) used productivity measures to analyze the net benefits of multinationality. As result they found that (a) multinationality increases efficiency and reduces business risks and at the same time (b) it causes reduction in firm’s overall efficiency because of the costs incurring from the internationalization process. They state that the best strategy may be internalizing some markets so that the benefits could overcome costs (Internalization Theory). Luo & Tan (1998) compared MNEs and DEs in the Chinese electronic industry with profitability measures (ROS, ROA, Average Sales Growth). In this case, the result of their study was inconclusive because their comparison was based on the strategic behavior and philosophy of the firm (Defender; Analyzer and Prospector). They found that MNEs and DEs did not imitate their competitor hence adjusting to the market. MNEs follow mostly an Analyzer strategy as in the case of Local firms in order to 26 protect their market position adopted a Defender posture. Each thrived in their own way not having a significant difference in terms of performance. Barbosa & Louri (2005) also compared domestic and foreign firms but focusing the ownership influence on performance (ROA). For the Greek sample they actually found that foreign firms perform better than domestic but for the Portuguese one there were no significant differences. Given these different results we couldn’t just label this study as “Advantage MNE” as there was an inconclusive part in the study. 27 Chapter 3. Methodology 3.1. Data This study uses firm-level data extracted from the SABI (Sistema de Análise de Balanços Ibéricos) – Bureau van Dijk’s database (last update on January 6th, 2014), which contains financial and corporate information on 500 thousand and 2 million firms located in Portugal and Spain, respectively. The main purpose of this dissertation is to assess whether there are differences in performance between Portuguese DEs and Portuguese MNEs, and, if so, in what performance variables are those differences more relevant. The present analysis implies that we needed two samples to work with (Portuguese DE and Portuguese MNEs). Due double counting problems, we needed to extract a third one, containing foreign-owned firms that would allow us to purge the duplicates on the first two samples. Initially, all samples were extracted for the period between 2002 and 2012, but later we found that this period did not granted us the data quality we needed. Thusly we shortened it to a 5 year period between 2008 and 2012. The criteria set for the extraction was as follows. In order not to get our database biased by the size of the much smaller enterprises, we set the minimum of employees for each firm to be 10 employees. This number is based on the European Union definition of a Micro Companies. Since we also need to guarantee the multinational status and the domestic status for our samples and to differentiate DEs from domestic MNEs from foreign MNEs, we used a criterion based on OECD benchmark definition for Foreign Direct Investment – “10% or more of the voting power of an enterprise resident in one economy by an investor resident in another” (OECD, 2008: 48, 49). So for the Sample 1, that is the DE sample, we set that (1) the owner has to be Portuguese and owning at least 90% and (2) companies must have 10 or more employees, thus excluding Micro-Companies (European Union-based nomenclature). The Sample 2, relative to the Portuguese MNEs was extracted with the following criteria: (1) Portuguese firms with foreign subsidiaries owned by at least 10%; and (2) also must have 10 or more employees. Since we need to eliminate the duplicated observations 28 from both samples, we extracted a third sample (Sample 3) of firms operating in Portugal owned by a foreign shareholder by at least 10% and also with at least 10 employees. The procedure for eliminating the double counting problem was: (a) eliminate the firms from Sample 3 contemplated in Sample 2 (MNE sample) which gave us a final MNE sample of 536 firms (from the initial 624). Then (b) we eliminated the remaining firms of both Portuguese and Foreign MNEs that exist also in the DEs sample [taking Sample 2 and 3 repeated firms from Sample 1] which ended with a total of 44290 firms (from the initial 45115). We also wanted to guarantee continuous data for each company throughout the entire period, so we’ve only kept the firms that were able to give us data in each of the 5 years of the period for some central variables. This left us with 18.941 DEs (6171 of which are Manufacturers) and 408 MNEs (192 of which are Manufacturers). Synthesizing, by manufacturing subsectors: Table 4 – Manufacturing subsectors and observation count Sector Sector Description DE MNE Total Sector_10 Manufacture of food products 4400 75 4475 Sector_11 Manufacture of beverages 330 35 365 Sector_12 Manufacture of tobacco products 10 5 15 Sector_13 Manufacture of textiles 1925 50 1975 Sector_14 Manufacture of wearing apparel 4405 65 4470 Sector_15 Manufacture of leather and related products 2755 40 2795 Sector_16 Manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials 1480 40 1520 Sector_17 Manufacture of paper and paper products 395 45 440 Sector_18 Printing and reproduction of recorded media 1100 20 1120 Sector_19 Manufacture of coke and refined petroleum products 0 5 5 Sector_20 Manufacture of chemicals and chemical products 485 45 530 Sector_21 Manufacture of basic pharmaceutical products and pharmaceutical preparations 100 25 125 Sector_22 Manufacture of rubber and plastic products 1160 55 1215 Sector_23 Manufacture of other non-metallic mineral products 2090 45 2135 Sector_24 Manufacture of basic metals 235 25 260 Sector_25 Manufacture of fabricated metal products except machinery and equipment 4760 175 4935 Sector_26 Manufacture of computer, electronic and optical products 110 10 120 29 Sector_27 Manufacture of electrical equipment 445 45 490 Sector_28 Manufacture of machinery and equipment n.e.c. 1050 70 1120 Sector_29 Manufacture of motor vehicles, trailers and semitrailers 470 40 510 Sector_30 Manufacture of other transport equipment 125 0 125 Sector_31 Manufature of furniture 1810 20 1830 Sector_32 Other manufacturing 535 10 545 Sector_33 Repair and installation of machinery and equipment 680 15 695 Total Manufacturing obs. 30855 960 31815 Total Manufacturing firms 6171 192 6363 Source: Own elaboration based on NACE Codes Rev. 23. 3.2. Empirical Model Many previous studies that also focused on evaluating the performance of MNEs versus DEs used similar variables to those that we are about to use (Gomes & Ramaswamy, 1999; Contractor et al., 2003; Elango, 2006; Contractor et al., 2007) and applied Pooled Cross Section and Time Series regressions as their model. In this dissertation we’ll present results in both panel data and cross-sectional analyses. Firstly we use Pooled-OLS (Ordinary Least Squares) regression estimation which is a “pooled linear regression” that “assumes a constant intercept and slopes regardless of group and time period” (Park, 2011: 19). Combining the advantages of time-series with cross-sectional, Pooled-OLS will allow us to assess the performance differences of two sets of firms, concerning a set of dependent variables, over a period of time. Secondly, because of the 2008 crisis this time period could give us some atypical results, so we decided to use a cross-sectional model that would allow us to analyze individually each year and draw some conclusions. Furthermore, each of these two estimation procedures will be used in order to investigate the effect of multinationality on performance using (a) the full database; (b) only manufacturing firms and (c) controlling for manufacturing subsectors. Our main functional forms to be used in each regression are: 3 Statistical Classification of Economic Activities in the European Community (EUROSTAT, 2008). 30 For Pooled-OLS: (3.1) Performancei = β0 + β1 DNSUBi1 + β2MANUFACTi2 + β3LOGAGEi3 + β4 LOGAGE2i4 + β5LOGNEMPi5 + β6LOGNEMP2i6 + β7SECTOR_10i7 + … + β30SECTOR_33i30 + εi Where “i” represents a firm, with “i” = 1 to 96.745 firms; DNSUB = Dummy of Number of Subsidiaries; MANUFACT = Dummy for Manufacturing firms; SECTOR = Dummy for Manufacturing Sub-sectors, 24 sectors = 24 – 1 = 23 dummies, from Sector_10 to Sector_33. For Cross-Sectional OLS: (3.2) Performancei = β0 + β1 DOMi1 + β2MANUFACTi2 + β3LOGAGEi3 + β4 LOGAGE2i4 + β5LOGNEMPi5 + β6LOGNEMP2i6 + β7SECTOR_10i7 + … + β30SECTOR_33i30 + εi Where DOM = Dummy for MNEs (=1 if MNE; =0 if DE). 3.3. Variables and Proxies 3.3.1. Dependent Variables For the performance variables’ selection we followed the empirical literature review presented on the previous section, which divided such performance measures into two major categories: Profitability and Productivity. As Productivity measure, we’ll use Gross Value Added (GVA) (Davies & Lyons, 1991) divided Number of Employees Girma et al., 2004) yielding GVA per Employee (GVApEmpl). In order to measure 31 performance in terms of Profitability we will use (1) ROS (Grant, 1987; Grant et al., 1988; Geringer et al., 1989; Luo & Tan, 1998; Capar & Cotabe, 2003; Contractor et al., 2003; Qian et al., 2003; Brock et al., 2006; Chiao, et al., 2006; Coombs & Bierly, 2006; Contractor et al., 2007), (2) ROA (Grant, 1987; Grant et al., 1998; Geringer et al., 1989; Luo & Tan, 1998; Gomes & Ramaswamy, 1999; Lu & Beamish, 2001; Mathur et al., 2001; Contractor et al, 2003; Ruigrok & Wagner, 2003; Lu & Beamish, 2004; Barbosa & Louri, 2005; Coombs & Bierly, 2006; Kimura & Kiyota, 2007; Contractor et al., 2007; Adenaeuer & Heckelei, 2011), (3) ROE (Grant, 1987; Grant et al, 1988; Kim & Lyn, 1990; Mathur et al., 2001; Chiang & Yu, 2005; Hsu, 2006; Coombs & Bierly, 2006; Kimura & Kiyota, 2007; Contractor et al., 2007; Adenaeuer & Heckelei, 2011), (4) Profit Margin (Kim & Lyn, 1990; Elango, 2006; Adenaeuer & Heckelei, 2011). The operationalisation of this and other variables will be presented in Table 4. 3.3.2. Independent Variables Main explanatory variable: Although the literature suggests Foreign Sales to Total Sales ratio a popular proxy to use for a “Multinationality” variable, we have to take into account the reality of our data. We have firms that operate solely on national territory, thus, having no Foreign Sales at all. Since our purpose is to make a comparison between MNEs and DEs, we introduce dnsub as main explanatory variable. This is an interaction variable composed by a dummy (MNE = 1; DE = 0) and by the Number of Subsidiaries each firm has. Thus, this variable assumes value =0 if it’s a DE and ≠0 if it’s a MNE. Not only it will allow us to understand the importance of being multinational but also give us a sense on how the degree of internationalization affects a firm’s performance. We expect MNEs to have a superior performance, so the expected signal is expected to be positive. 32 Multinational Dummy We’ll also use a dummy variable which assumes value of 1 if it’s a MNE and 0 if it’s a DE. This dummy is one of the components of the previously presented variable and it will prove itself useful for our cross-sectional models since there is no need to analyze the temporal development of the Multinational status, based on the number of subsidiaries. We expect a positive impact of this dummy on performance. Manufacturing For firm sector-specific considerations, we’ll include a dummy of the NACE Rev. 2 (EUROSTAT, 2008) concerning the sector of the firms on our samples (Manufact) assuming 1 when it’s a Manufacturing firm and 0 when it’s not. Age Age is also a measure that has been used by some authors (Qian et al., 1003; Barbosa & Louri, 2005; Contractor et al., 2007). If Experience is a hard-to-imitate and non-substitutable asset (Resource-based View; Chapter 1.1.4.), and if it’s also a result of continuous improvement, then Log of Number of Years (logage) is a very much correct proxy of this intangible asset, leading us to expect a positive effect. In order to assess the behavior of the curve of this variable, we introduce logage2 as the Squared Log of Number of Years. This will allow us to test a quadratic function and draw conclusions on the nonlinear relationship between Age and Performance. Size One of the explanatory variables is Size (Buckley et al., 1984; Grant, 1987; Lee & Kwok, 1988; Gomes & Ramaswamy, 1999; Lu & Beamish, 2001; Capar & Kotabe, 2003; Contractor et al., 2003; Qian, et al., 2003, Barbosa & Louri, 2005; Chiao et al., 2006; Hsu, 2006; Anastassopoulos et al., 2007; Contractor et al., 2007). Although some authors use Log of Total Sales as proxy for Firm Size, we will use the Log Number of Employees gives us a more stable/less volatile measure (henceforward lognemp). MNEs are, generally, larger than DEs. 39 Considering our Productivity variable, this new model shows us stronger value in our main dummy. So, for manufacturing firms, the increase in the number of subsidiaries abroad means even higher Productivity. The Profitability measures show the same result as the coefficient for logPM increased more than three times; eight times for ROE; more than six times for ROA, and so on. Regarding Age, if anything, it became clearer that the more experienced a firm gets, the more Productive it is. Although the same doesn’t happen with Profitability, this corroborates the Resource-based theory which states that as the firm grows bigger and more experienced, it becomes more efficient, hence, more Productive. The results in these two sections are consistent with what was previously stated and found in the literature, that there is a positive relationship between internationalization and performance, whether it is measured by Profitability (Grant; 1987; Geringer et al., 1989; Qian et al., 2003) or Productivity (Davies & Lyons, 1991; Girma et al., 2004) Table 9 - Pooled OLS, Only Manufacturing Only Manufact. Pooled OLS loggvaemp logPM logroe logroa logros dnsub 0,0509*** 0,0441*** 0,0225*** 0,0272*** 0,0145** (0,0056) (0,0061) (0,0040) (0,0044) (0,0062) logage 0,0816*** -0,1106*** -0,6511*** -0,3083*** -0,2574*** (0,0044) (0,0115) (0,0138) (0,0139) (0,0171) lognemp 0,1375*** 0,0539*** 0,0912*** 0,1150*** 0,0722*** (0,0047) (0,0106) (0,0125) (0,0125) (0,0138) cons 2,1822*** 0,9334*** 3,2946*** 0,9915*** 2,1980*** (0,0181) (0,0427) (0,0504) (0,0507) (0,0621) Used Subsector Dummies No No No No No R-sq. 0,0854 0,0077 0,0784 0,0214 0,0123 F-test 692,64*** 57,36*** 754,62*** 193,14*** 81,37*** No. Obs. 31739 24725 26150 25333 22748 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 40 Pooled-OLS, Only Manufacturing, with Subsectors By controlling for manufacturing sub-sector effect, we intend to understand whether or not the results presented in the previous section may be biased by the type of sector. The functional form to be used is: (3.5) Performancei = β0 + β1 DNSUBi1 + β2LOGAGEi2 + β3LOGNEMPi3 + β4SECTOR_10i4 + … + β27SECTOR_33i27 + εi When placing more information to the model, this is how it reacted: all of the Rsquared improved, all models are still globally significant. Controlling the sample for all sectors we can see that the degree of internationalization is still positive and significant. As Table 10 shows, Size, for Productivity measure, increased its coefficient and now for each 1% increase in Size, the Productivity is predicted to increase by 0,1624%. LogPM, LogROE and LogROA still have positive and significant results for dnsub as before. LogROS lost its significance concerning dnsub, but overall there is still evidence that both Profitability and Productivity of MNEs still outperform the ones of DEs, corroborating the previous estimations and the literature review. Table 10 - Pooled OLS, Manufacturing, Subsectors Only Manufact. Pooled OLS loggvaemp logPM logroe logroa logros dnsub 0,0357*** 0,0369*** 0,0274*** 0,0282*** 0,0057 (0,0039) (0,0058) (0,0040) (0,0044) (0,0074) logage 0,0080** -0,1279*** -0,6082*** -0,2786*** -0,2773*** (0,0039) (0,0117) (0,0140) (0,0141) (0,0167) lognemp 0,1624*** 0,0530*** 0,0491*** 0,0779*** 0,0887*** (0,0040) (0,0107) (0,0129) (0,0129) (0,0138) cons 2,5329*** 1,1791*** 3,2971*** 1,0718*** 2,7670*** (0,0165) (0,0454) (0,0542) (0,0545) (0,0662) Used Subsector Dummies Yes(a) Yes(a) Yes(a) Yes(a) Yes(a) R-sq. 0,3318 0,0349 0,1019 0,0489 0,0729 41 F-test 680,98*** 37,28*** 126,44*** 55,80*** 59,60*** No. Obs. 31739 24725 26150 25333 22748 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% (a) Sectors 19 and 30 were omitted for having observations only for one type of firm, either DE or MNE, respectively. Source: Own elaboration based on STATA outputs. 3.4.2. Cross-Section Model As said before, in this section we will conduct the analysis of each dependent variable by year to see if the time period on which we are working with has effect on the estimations. This will also allow us to have a better understanding of how they’ve evolved from year to year. Because it is cross-sectional estimation, we decided that it would be better to use a normal dummy as main explanatory variable. Also, in preliminary regressions, dnsub wasn’t allowing the model to be both individual and global significant. Moreover, in this section we will apply the squared logarithms of Age and Size in order to ascertain whether there are or not curvilinear relationships for these variables. Again, for purposes of global and individual significance, these variables weren’t applied to every single model that we are about to present. We will follow the structure of the previous section, presenting the results for the full database, then only manufacture and lastly manufacturing controlling for subsector. The estimation results of cross-sectional analysis for the full database are presented in Appendix I, Tables 11 to 15. Our Productivity measure keeps showing signs of better performance from MNEs. An interest detail is the signal between both normal Size and Age and their squared term. This variable is telling us that GVA per Employee will improve if both Size and Age increase but only until a certain point. Here we have evidence that Size and Age have an inverted u-shaped relationship with GVA per Employee. Profit Margin, on the other hand, has the opposite of our Productivity measure. First, the squared logarithm of age had to be cut off for multicollinearity and individual 42 significance problems. Second, Size was only manageable to be significant in the year 2010. We must keep in mind that this five year period corresponds to a period of crisis and its aftermath. Perhaps this was the year when firms were already recovering from crisis, after some massive layoff, and were now under the minimum efficient scale, in need of hiring. LogROE’s model was globally significant but our main explanatory variable is only significant for 2010. Again, this might be a very atypical year to analyze and, again, we find evidence for a u-shaped curve for the variable Size. This negative impact from lognemp on the Profitability corroborates the negative relationship found by Capar & Kotabe (2003). For each year, Age has a negative impact on Profitability, suggesting a negative slope in a linear relationship. The same goes for Manufacturing dummy as it states that for Manufacturing firms there isn’t necessarily positive evidence that they have better performance. Like GVA per Employee and Profit Margin, ROA provided us an estimation output with both global and individual significance for all models, except 2012. We still find evidence of a u-shaped relationship from Profitability and Size. MNEs had better performance in each year, and Age is still a negative factor. The regression of the Log of ROS shows us totally different values. If we focus on the year 2009, given that is our only year with dom significant, we get the multinational status as a negative contributing factor. Mathur et al., (2001) also concluded that multinationality was a contributor to lower performance, but when analyzing for a nonlinear relationship, they found evidence for a u-shaped relationship. Here we aren’t able to assess whether there is a turning point for multinationality to be a positive factor. Maybe future investigations will be able to find more conclusive findings in this particular matter. 43 Table 11 - LogGVAEMP Cross-Section OLS - Log GVA per Employee 2008 2009 2010 2011 2012 Dom 0,8638*** 0,8280*** 0,8289*** 0,7891*** 0,7768*** (0,0616) (0,0622) (0,0645) (0,0629) (0,0649) Manufact -0,2319*** -0,2452*** -0,2169*** -0,2082*** -0,1753*** (0,0093) (0,0091) (0,0090) (0,0092) (0,0100) logage 0,1702*** 0,1642*** 0,1982*** 0,2622*** 0,3341*** (0,0223) (0,0242) (0,0338) (0,0453) (0,0593) logage2 -0,0140*** -0,0142*** -0,0200*** -0,0303*** -0,0434*** (0,0047) (0,0049) (0,0064) (0,0082) (0,0103) lognemp 0,1990*** 0,2015*** 0,2111*** 0,3180*** 0,4476*** (0,0365) (0,0356) (0,0341) (0,0342) (0,0350) lognemp2 -0,0190*** -0,0183*** -0,0176*** -0,0284*** -0,0408*** (0,0049) (0,0048) (0,0452) (0,0045) (0,0046) Cons 2,3204*** 2,3167*** 2,0272*** 1,8358*** 1,3798*** (0,0692) (0,0699) (0,0751) (0,0865) (0,1061) R-sq. 0,0879 0,0875 0,0842 0,0820 0,0755 F-test 208,57*** 216,01*** 191,19*** 202,85*** 185,61*** No. Obs. 19106 19345 19345 19345 19345 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 44 Table 12 - LogPM Cross-Section OLS - Log Profit Margin 2008 2009 2010 2011 2012 dom 0,5353*** 0,4607*** 0,6435*** 0,6263*** 0,6393*** (0,0870) (0,0888) (0,0833) (0,0875) (0,0916) Manufact -0,1151*** -0,0660*** -0,0836*** -0,0807*** -0,0660*** (0,0220) (0,0217) (0,0211) (0,0225) (0,0239) logage -0,0327** -0,0299** -0,0411*** -0,0442*** -0,0597*** (0,0130) (0,0131) (0,0145) (0,0170) (0,0195) logage2 - - - - - lognemp -0,1259* -0,0299* -0,1814*** -0,0010 0,0203** (0,0718) (0,0680) (0,0626) (0,0754) (0,0802) lognemp2 0,0090 0,0117 0,0255*** 0,0082 -0,0116 (0,0091) (0,0086) (0,0077) (0,0094) (0,0101) cons 1,4228*** 1,3990*** 1,311*** 0,7560*** 0,2938** (0,1344) (0,1303) (0,1248) (0,1490) (0,1588) R-sq. 0,0068 0,0039 0,0086 0,0087 0,0122 F-test 16,91*** 9,87*** 23,96*** 22,17*** 31,40*** No. Obs. 14759 14856 16316 15187 13861 The numbers in parentheses are the Robust Std. Error corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 45 Table 13 - LogROE Cross-Section OLS - Log ROE 2008 2009 2010 2011 2012 dom 0,0470 0,0250 0,1437** 0,1244 0,0946 (0,0784) (0,0746) (0,0709) (0,0788) (0,0875) Manufact -0,2651*** -0,2227*** -0,1739*** -0,0834*** -0,1098*** (0,0253) (0,0249) (0,0246) (0,0283) (0,0318) logage -0,5125*** -0,5239*** -0,5243*** -0,5391*** -0,5299*** (0,0145) (0,0148) (0,0164) (0,0205) (0,0256) logage2 - - - - - lognemp -0,2705*** -0,0675 -0,2017*** -0,0863 0,1403* (0,0674) (0,0652) (0,0635) (0,0740) (0,0804) lognemp2 0,0405*** 0,0204*** 0,0416*** 0,0323*** 0,0106 (0,0082) (0,0078) (0,0075) (0,0087) (0,0094) cons 3,9717*** 3,5875*** 3,6583*** 3,1940*** 2,5528*** (0,1309) (0,1297) (0,1301) (0,1545) (0,1703) R-sq. 0,0879 0,0832 0,0693 0,0506 0,0396 F-test 296,08*** 284,00*** 243,02*** 166,78*** 118,17*** No. Obs. 16182 16309 16374 15414 14403 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 46 Table 14 – LogROA Cross-Section OLS - Log ROA 2008 2009 2010 2011 2012 dom 0,1778** 0,1842** 0,2907*** 0,2786*** 0,1885** (0,0798) (0,0762) (0,0755) (0,0808) (0,0875) Manufact -0,2525*** -0,2410*** -0,1716*** -0,1086*** -0,1167*** (0,0259) (0,0251) (0,0246) (0,0271) (0,0299) logage -0,0169*** -0,1990*** -0,2130*** -0,2211*** -0,2684*** (0,0155) (0,0151) (0,0166) (0,0200) (0,0239) logage2 - - - - - lognemp -0,2877*** -0,2192*** -0,3139*** -0,1676** 0,3347*** (0,0713) (0,0713) (0,0668) (0,0819) (0,0779) lognemp2 0,0365*** 0,0308*** 0,0470*** 0,0358*** -0,0114 (0,0086) (0,0087) (0,0080) (0,0100) (0,0093) cons 1,8628*** 1,8278*** 1,8603*** 1,2707*** 0,1103 (0,1390) (0,1398) (0,1356) (0,1648) (0,1643) R-sq. 0,0174 0,0195 0,0184 0,0149 0,0245 F-test 53,15*** 61,91*** 61,67*** 45,69*** 71,85*** No. Obs. 15774 15938 16104 14878 13458 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 47 Table 15 - LogROS Cross-Section OLS - Log ROS 2008 2009 2010 2011 2012 dom -0,1924 -0,3535** 0,0675 0,0266 0,1349 (0,1368) (0,1391) (0,1299) (0,1379) (0,1491) Manufact -0,7640*** -0,7581*** -0,7858*** -0,8854*** -0,8230*** (0,0400) (0,0401) (0,0381) (0,0400) (0,0432) logage -0,2659*** -0,3011*** -0,3062*** -0,3129*** -0,3696*** (0,0273) (0,0290) (0,0301) (0,0343) (0,0381) logage2 - - - - - lognemp 0,2843*** 0,2899*** 0,3185*** 0,3737*** 0,3999*** (0,0329) (0,0317) (0,0289) (0,0310) (0,0317) lognemp2 - - - - - cons 2,3724*** 2,4617*** 2,2957*** 2,1103*** 2,1855*** (0,1150) (0,1193) (0,1172) (0,1314) (0,1427) R-sq. 0,0422 0,0439 0,0487 0,0572 0,0569 F-test 120,05*** 124,17*** 144,70*** 162,78*** 136,27*** No. Obs. 11341 11028 11905 11263 10154 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. Cross-section, Only Manufacturing In Appendix II (Tables 16 to 20) are the regression outputs for the Manufacturing sample. For Log of GVA per Employee we observe a small drop in the coefficients of dom, but they are still positive and significant, confirming that multinationality does impact positively on Productivity. Age has a significant improvement as it increases Productivity even more for manufacturing firms, when comparing with the total sample. The same happens for Size, where besides an increase when comparing with the full dataset, there is also an increase from year to year. The Profitability measures show, again, a different reality from Productivity. Here both Age and Size are negative contributors for the performance, showing 48 evidence for Profit Margin. The other Profitability variables still have Age showing negative impact, but Size is positive and significant. The dom dummy is positive and significant for every variable. ROS wasn’t able to produce any significant evidence, so we won’t be drawing any interpretation. Table 16 - LogGVAEMP Cross-Section, Manufacturing OLS - Log GVA per Employee 2008 2009 2010 2011 2012 dom 0,5368*** 0,5017*** 0,5129*** 0,4892*** 0,4089*** (0,0525) (0,0525) (0,0518) (0,0540) (0,0626) logage 0,0949*** 0,0924*** 0,0880*** 0,0832*** 0,0869*** (0,0086) (0,0089) (0,0099) (0,0109) (0,0124) logage2 - - - - - lognemp 0,0998*** 0,0998*** 0,1134*** 0,1345*** 0,1588*** (0,0100) (0,0104) (0,0099) (0,0099) (0,0101) lognemp2 - - - - - cons 2,2968*** 2,2862*** 2,2593*** 2,1651*** 2,0313*** (0,0363) (0,0377) (0,0386) (0,0414) (0,0459) Used Subsector Dummies No No No No No R-sq. 0,0948 0,0842 0,0923 0,0958 0,0904 F-test 150,11*** 138,10*** 146,25*** 153,88*** 161,15*** No. Obs. 6295 6361 6361 6361 6361 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 55 Table 23 - LogROE Cross-Section, Manufacturing with Subsectors OLS - Log ROE 2008 2009 2010 2011 2012 dom 0,1418 0,0094 0,3166*** 0,2403** 0,1207 (0,1269) (0,1232) (0,1098) (0,1119) (0,1240) logage -0,8218*** -1,1290*** -1,1669*** -1,6752*** -2,1589*** (0,0982) (0,1135) (0,1609) (0,2167) (0,2759) logage2 0,0600*** 0,1119*** 0,1101*** 0,1933*** 0,2714*** (0,0208) (0,0228) (0,0300) (0,0388) (0,0479) lognemp -0,0653** 0,0007 0,0207 0,0816*** 0,1177*** (0,0298) (0,0304) (0,0278) (0,0307) (0,0335) lognemp2 - - - - - cons 4,0313*** 4,1608*** 4,0962*** 4,4011*** 4,8468*** (0,1503) (0,1719) (0,2356) (0,3202) (0,4141) Used Subsector Dummies Yes Yes Yes Yes Yes R-sq. 0,1233 0,1306 0,1081 0,0990 0,0869 F-test 29,14*** 31,72*** 26,80*** 23,28*** 20,50*** No. Obs. 5353 5268 5427 5124 4978 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 56 Table 24 - LogROA Cross-Section, Manufacturing with Subsectors OLS - Log ROA 2008 2009 2010 2011 2012 dom 0,1705 0,0970 0,3657*** 0,3511*** 0,1256 (0,1243) (0,1260) (0,1151) (0,1132) (0,1228) logage -0,1867*** -0,2399*** -0,2797*** -0,2856*** -0,2943*** (0,0287) (0,0273) (0,0314) (0,0347) (0,0407) logage2 - - - - - lognemp -0,0303 0,0091 0,0380 0,1215*** 0,2216*** (0,0301) (0,0300) (0,0284) (0,0299) (0,0323) lognemp2 - - - - - cons 1,4078*** 1,3923*** 1,2717*** 0,7594*** 0,2704*** (0,1152) (0,1150) (0,01186) (0,1318) (0,1500) Used Subsector Dummies Yes Yes Yes Yes Yes R-sq. 0,0500 0,0566 0,0518 0,0620 0,0637 F-test 10,70*** 13,16*** 12,82*** 15,45*** 14,59*** No. Obs. 5228 5140 5345 4939 4681 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 57 Table 25 - LogROS Cross-Section, Manufacturing with Subsectors OLS - Log ROS 2008 2009 2010 2011 2012 dom -0,1392 -0,3620** -0,0016 0,0520 -0,1056 (0,1139) (0,1408) (0,1049) (0,1093) (0,1112) logage -0,1974*** -0,2686*** -0,2703*** -0,3193*** -0,2867*** (0,0322) (0,0363) (0,0375) (0,0415) (0,0451) logage2 - - - - - lognemp 0,0132 0,0597** 0,1101*** 0,1547*** 0,1496*** (0,0308) (0,0325) (0,0302) (0,0324) (0,0327) lognemp2 - - - - - cons 2,9050*** 2,9310*** 2,7167*** 2,5286*** 2,5152*** (0,1360) (0,1476) (0,1467) (0,1567) (0,1772) Used Subsector Dummies Yes Yes Yes Yes Yes R-sq. 0,0698 0,0730 0,0820 0,0831 0,0746 F-test 11,60*** 11,10*** 15,34*** 14,41*** 13,25*** No. Obs. 4662 4455 4813 4566 4252 The numbers in parentheses are the Robust Std. Err. Corrected for Heteroskedasticity Significance level: ***p<1% ; **p<5% ; *p<10% Source: Own elaboration based on STATA outputs. 58 Chapter 4. Conclusions and Policy Implications 4.1. Conclusions The main focus of this dissertation was to test performance differences between two types of firms (MNEs and DEs) and whether these differences, if any, differ across a set of variables. There are many studies that treat this matter but only focusing on the relationship between the two components (Multinationality and Performance), as this dissertation presents a different perspective, aiming for the comparison between two sets of firms and using different proxies to verify the performance difference across different types of variables. Also, there isn’t any other study, to the best of our knowledge, that deals with performance comparison, focusing on OFDI and testing for differences between Manufacturing and non-Manufacturing firms. Being a Multinational Enterprise means higher performance, as shown before, with the only exception for when it comes to ROS analysis. But even that was only a single year observation showing different results. ROS in all Pooled models showed positive influence of Multinationality on Profitability. The findings also suggest that MNEs have much better Productivity than DEs. It’s a performance gap even higher than all of Profitability measures. This can be explained by the economies of scale that a MNE is able to get. The Eclectic Paradigm also predicts this outcome when it states that a MNE has the advantage in common governance, transport infrastructures, a wider and more efficient supplying network. This means that becoming a MNE compensates all the serious business-related risks that a firm undertakes when going global – although not every firm is fit to become multinational. The variables used in our models present evidence that for Productivity, Age is a positive contributor and for Profitability is a negative contributor. Size, however, presents a unanimous and positive influence on Performance. As a firm gets larger, the better it performs. This might have something to do with the Multinational’s ability to diversify its value chain optimizing it throughout its area of activity. 59 4.2. Policy Implications Becoming a MNE, as said before, isn’t a simple process. And not every firm is able to do it. Portugal has some credit lines that help firms starting to internationalize through exports, but there’s more than that. Invoking the both the Eclectic Paradigm, Resource-based view and the Network Theory, becoming Multinational requires a certain level of knowledge, ownership and connections without which the process will likely fail. AICEP is doing helping firms in terms of knowledge of external markets, local connections abroad, etc, but the ownership part is still difficult to overcome. For domestic firms it’s impossible to suddenly obtain these ownership advantages. The entry to a new market by a DE should be accompanied by both Governments from country of origin and country of destination and the Portuguese Government should assist the DE in order for it to get the ownership advantages needed. For instance, the government should deal with a local firm that also wishes to internationalize, permitting facilitating its entry in return of assistance to the Portuguese MNE-to-be. In some way this is already happening, but not as proactively as it needs. The Foreign Affairs Ministry should open a platform of applications for DEs who wish to internationalize and become global but lack of something crucial for the process to begin. Concisely, there should be more policies promoting internationalization through outward FDI. 4.3. Future research Further researches can be made from this one. For starters, this study could be applied to a variety of countries – European Union, Euro-Zone, NAFTA. vs. EU and so on – creating a cross-country analysis and whether, or not, MNEs that outperform DEs are from countries that export more or less, etc. It should be interesting to expand the set of independent variables of each firm. We weren’t able to do so because of the quality of our database, but with some previous work on that and getting a complete and balanced database, it could be possible. Using a different time period can also be positive. We used this time period because it had the most data quality but we must keep in mind that this was a period of economic struggle. 60 References Adenaeuer, L. and Heckelei, T. (2011), “Foreign Direct Investment and the Performance of European Agribusiness Firms”, Journal of Agricultural Economics, Vol. 62, No. 3, pp. 639-654. Al-Obaidan, A. and Scully, G. (1995), “The theory and measurement of the net benefits of multinationality: the case of the international petroleum industry”, Applied Economics, Vol. 27, No. 2, pp. 231-238. Annastassopoulos, G., Filippaios, F. and Phillips, P. (2007), “An ‘eclectic’ investigation of tourism multinationals’ activities: Evidence from Hotels and Hospitality Sector in Greece”, GreeSE - Hellenic Observatory Papers on Greece and Southeast Europe, Paper no. 8, London School of Economics. Assaf, A., Josiassen, A., Ratchford, B. and Barros, C. (2012), “Internationalization and Performance of Retail Firms: A Bayesian Dynamic Model”, Journal of Retailing, Vol. 88, No. 2, pp. 191-205. Augier, M. and Teece, D. J. (2007), “Dynamic Capabilities and Multinational Enterprise: Penrosean Insights and Omissions”, Management International Review, Vol 47, No. 2, pp. 175-192. Barbosa, N and Louri, H. (2005), “Corporate Performance: Does Ownership Matter? A Comparison of Foreignand Domestic-Owned Firms in Greece and Portugal”, Review of Industrial Organization, Vol. 25, No. 1, pp. 73-102. Barney, J. (1991), “Firm Resources and Sustained Competitive Advantage”, Journal of Management, Vol. 17, No. 1, pp. 99-120. Bartlett, C. A. and Ghoshal, S. (1998), Managing Across Borders: The Transnational Solution, 2nd Edition, Boston, Massachusetts: Harvard Business School Press. Bellak, C. (2004a), “How Domestic and Foreign Firms Differ and Why Does it Matter?”, Journal of Economic Surveys, Vol. 18, No. 4, pp. 483-514. 61 Bellak, C. (2004b), “How Performance Gaps between Domestic and Foreign Firms Matter for Economic Policy?”, Transnational Corporations, Vol. 13, No.2, pp.2955. Benvignati, A. (1987), “Domestic Profit Advantages of Multinational Firms”, The Journal of Business, Vol. 60, No. 3, pp. 449-461. Brewer, H. (1981), “Investor Benefits from Corporate International Diversification”, The Journal of Financial and Quantitative Analysis, Vol. 16, No. 1, pp. 113-126. Brock, D., Yaffe, T., Dembovsky, M. (2006). “International Diversification and Performance: A Study of Global Law Firms”, Journal of International Management, Vol. 12, No. 4, pp. 473-489. Buckley, P. J. and Casson, M. C. (1976), “A long-run theory of the multinational enterprise”, in P. J. Buckley and M. C. Casson (eds), The Future of the Multinational Enterprise, London: Macmillan, pp. 32-65. Buckley, P. J., Dunning, J. and Pearce, R. (1984), “An Analysis of the Growth and Profitability of the World’s Largest Firms 1972 to 1977”, Kyklos, Vol. 37, No. 1, pp. 3-26. Capar, N. and Kotabe, M. (2003), “The Relationship between International Diversification and Performance in Service Firms”, Journal of International Business Studies, Vol. 34, No. 4, pp. 345-355. Cardoso, V (2008), “Foreign Multinational and Domestic Companies in Portugal: Are There Significant Performance Gaps?”; Master Thesis, Faculdade de Economia do Porto. Chang, S. and Rhee, J. (2011), “Rapid FDI expansion and firm performance”, Journal of International Business Studies, Vol. 42, pp. 979-994. Chari, M., Devaraj, S. and David, P. (2007), “International Diversification and Firm Performance: Role of Information Technology Investments”, Journal of World Business, Vol. 42, No. 2, pp. 184-197. Chen, H. and Hsu, C. (2010), “Internationalization, Resource Allocation and Firm Performance”, Industrial Marketing Management, Vol. 38, No. 7, pp. 1103-1110. 62 Chetty, S. and Holm, D. (2000), “Internationalisation of small to medium-sized manufacturing firms: a network approach”, International Business Review, Vol. 9, No. 1, pp. 77-93. Chiang, Y. and Yu, T. (2005), “The Relationship between Multinationality and the Performance of Taiwan Firms”, Journal of American Academy of Business, Cambridge, Vol. 6, No. 1, pp. 130-134. Chiao, Y., Yang, K. and Yu, C. (2006), “Performance, Internationalization, and FirmSpecific Advantages of SMEs in a Newly-Industrialized Economy”, Small Business Economics, Vol. 26, No. 5, pp. 475-492. Coase, R. H. (1937), “The Nature of the firm”, Economica, New Series, Vol. 4, No. 16, pp. 386-405. Collins, J. (1990), “A Market Performance Comparison of US Firms Active in Domestic, Developed and Developing Countries”, Journal of International Business Studies, Vol. 21, No. 2, pp. 271-287. Contractor, F., Kumar, V. and Kundu, S. (2007), “Nature of the Relationship between International Expansion and Performance: The Case of Emerging Market Firms”, Journal of World Business, Vol. 42, No. 4, pp. 401-417. Contractor, F., Kundu, S. and Hsu, C. (2003), “A Three-Stage Theory of International Expansion: The Link between Multinationality and Performance in the Service Sector”, Journal of International Business Studies, Vol. 34, No.1, pp. 5-18. Coombs, J. and Bierly. P. (2006), “Measuring Technological Capability and Performance”, R&D Management, Vol. 36, No. 4, pp. 421-438. Coviello, N. and Munro, H. (1995), “Growing the entrepreneurial firm: Networking for market development”, European Journal of Marketing, Vol. 29, No. 7, pp. 49-61. Coviello, N. and Munro, H. (1997), “Network Relationships and the Internationalisation Process of Small Sotfware Firms”, International Business Review, Vol. 6, No. 4, pp. 361-386. Davidson, R. and MacKinnon, J. G. (1993), Estimation and Inference in Econometrics, New York City, New York: Oxford University Press. 63 Davies, S. and Lyons, B. (1991), “Characterising Relative Performance: The Productivity Advantage of Foreign Owned Firms in the UK”, Oxford Economic Papers, New Series, Vol. 43, No. 4, pp. 584-595. Dunning, J. (1977), “Trade, Location of Economic Activity and the Multinational Enterprise: A search for na Eclectic Approach” – in P. J. Buckley and P. N. Ghauri (eds.) The Internalization of the Firm. London: Thomson Learning Europe, pp. 61-80. Dunning, J. (1980), “Explaining changing patterns of international production: in defense of the eclectic theory”, Oxford Bulletin of Economics and Statistics, Vol. 41, No. 1, pp. 269-295. Dunning, J. (1993), Multinational Enterprises and the Global Economy, Wokingham, UK: Addison-Wesley. Dunning, J. and Lundan, S. (2008), Multinational Enterprises and the Global Economy, 2nd Edition, Cheltenham, UK: Edward Elgar. Elango, B. (2006), “An Empirical Analysis of the International-Performance Relationship Across Emerging Market Firms”, The Multinational Business Review, Vol. 14, No. 1, pp. 21-44. EUROSTAT (2008), “NACE Rev.2 – Statistical classification of economic activities in the European Community”, Methodologies and Working Papers, Luxembourg: Office for Official Publications of the European Communities. Geringer, J. and Hebert, L. (1991), “Measuring Performance of International Joint Ventures”, Journal of International Business Studies, Vol. 22, No. 2, pp. 249-263. Geringer, J., Beamish, P. and DaCosta, R. (1989), “Diversification Strategy and Internationalization: Implications for MNE Performance”, Strategic Management Journal, Vol. 10, No. 2, pp. 109-119. Girma, S., Görg, H. and Strobl, E. (2004), “Exports, international investment, and plant performance: evidence from a non-parametric test”, Economics Letters, Vol. 83, No. 3, pp. 317-324. 64 Gomes, L. and Ramaswamy, K. (1999), “An Empirical Examination of the Form of the Relationship between Multinationality and Performance”, Journal of International Business Studies, Vol. 30, No. 1, pp. 173-187. Grant, R. (1987), “Multinationality and Performance among British Manufacturing Companies”, Journal of International Business Studies, Vol. 18, No. 3, pp. 79-89. Grant, R., Jammine, A. and Thomas, H. (1988), “Diversity, Diversification, and Profitability among British Manufacturing Companies”, The Academy of Management Journal, Vol. 31, No. 4, pp. 771-801. Hadley, R. and Wilson, H. (2003), “The network model of internationalization and experiential knowledge”, International Business Review, Vol. 12, No. 6, pp. 687717. Hayakawa, K., Matsuura, T., Motohashi, K., and Obashi, A. (2012), “Two-dimensional analysis of the impact of outward FDI on performance at home: Evidence from Japanese manufacturing firms”, Japan and the World Economy, Vol. 27, pp. 2533. Hennart, J.-F. (1977), A Theory of Foreign Direct Investment, PhD Thesis, Maryland: University of Maryland. Hsu, C. (2006), “Internationalization and Performance: The S-curve Hypothesis and Product Diversity Effect”, The Multinational Business Review, Vol. 14, No. 2, pp. 29-46. Ietto-Gillies, G. (2002), “Hymer, the nation-state and the determinants of multinational corporations’ activities”, Contributions to Political Economy, Vol. 21, No. 1, pp.43-45. Ietto-Gillies, G. (2005), Transnational Corporations and International Production: Concepts, Theories and Effects, Cheltenham: Edward Elgar. Imbriani, C., Pittiglio, R. and Reganati, F. (2011), “Outward Foreign Direct Investment and Domestic Performance: the Italian Manufacturing and Services Sectors”, Atlantic Economic Journal, Vol. 39, No. 4, pp. 369-381. 71 Chiang & Yu 2005 The Relationship between Multinationality and the Performance of Taiwan firms 119 Taiwanese Companies, 1998-2002 = 595 observations Profitability Advantage MNE Girma, Gorg & Strobl 2004 Exports, international investment, and plant performance: evidence from a non-parametric test Manufacturing plants in Republic of Ireland with more than 10 employees; year 2000. Productivity Advantage MNE Lu & Beamish 2004 International Diversification and Firm performance: the s-curve hypothesis 1489 Japanese firms, 1987 to 1997 Profitability Advantage MNE Capar & Kotabe 2003 The Relationship between International Diversification and Performance in Service Firms 81 German service firms (1997 - 1999) Profitability Advantage MNE Contractor, Kundu & Hsu 2003 A three-stage theory of international expansion: the link between multinationality and performance in the service sector 11 service industries, 103 firms Profitability Advantage MNE Qian, Yang & Wang 2003 Does multinationality affect profit performance? Na empirical study of SMEs 271 US firms (1993-1997) Profitability Advantage MNE Kotabe, Srinivasan & Aulakh 2002 Multinationality and Firm Performance: The Moderating Role of R&D and Marketing Capabilities 49 US firms from 12 different industries; 1987-1993 Profitability Advantage MNE Lu & Beamish 2001 The internationalization and Performance of SMEs 164 Japanese SMEs from19 different industries, as defined by the Nikkei stock market; 1986-1997 Profitability Advantage MNE 72 Mathur, Singh & Gleason 2001 The evidence from Canadian firms on multinational diversification and performance 427 (1997) Canadian firms; 1992–1994 and 1997 Profitability Advantage DE Gomes & Ramaswamy 1999 An empirical examination of the form of the relationship between multinationality and performance 95 firms (28 chemicals; 14 drugs and pharmaceuticals; 24 computers and computer office; 29 electrical products) Profitability Advantage MNE Ramstetter 1999 Comparisons of Foreign Multinationals and Local Firms in Asian Manufacturing Over Time firms from Hong-King; Malaysia; Indonesia; Singapore; Taiwan; 1970 to 1996 Productivity Advantage MNE Luo & Tan 1998 A comparison of multinational and domestic firms in na emerging market: a strategic choice perspective 60 state Chinese firms (electronics industry); 51 MNE subunits Profitability Inconclusive Al-Obaidan & Scully 1995 The Theory and measurement of the net benefits of multinationality: the case of the international petroleum industry 44 oil companies, 1976-82; 25 different countries Productivity Inconclusive Davies & Lyons 1991 Characterising relative performance: The productivity advantage of foreign owned firms in the UK (1971-1987) Productivity Advantage MNE Collins 1990 A market performance comparison of US firms active in Domestic, Developed and developing countries 133 firms (51 domestic, 44 in developed countries, 38 in developing countries), from January 1976 to June 1985 Profitability Advantage MNE Kim & Lyn 1990 FDI theories and the performance of foreign multinationals operating in the US 54 firms from different countries; 1980-1984 Profitability Advantage DE 73 Geringer, Beamish & DaCosta 1989 Diversification strategy and internationalization implications for MNE performance 200 MNE (US and European), 1982 - 1983 Profitability Advantage MNE Lee & Kwok 1988 Multinational Corporations vs Domestic Corporations: International Environmental Factors 834 firms (421 domestic; 413 MNE-10%; 231 Mne-25%); US and non-US firms Profitability Advantage MNE Benvignati 1987 Domestic Profit Advantages of Multinational Firms 2635 lines of business of 457 US manufacturing firms, 1975 Profitability Advantage MNE Grant 1987 Multinationality and performance among British manufacturing companies 304 large firms, 1972-1984 Profitability Advantage MNE Michel & Shaked 1986 Multinational Corporations vs Domestic Corporations: financial performance and characteristics 58 US-based MNEs; 43 DMCs; 1973 - 1982; Profitability Advantage DE Buckley, Dunning & Pearce 1984 An analysis of the growth and profitability of world's largest firms between 1972 and 1977 535 firms in 1972, 866 firms in 1977; US and non-US firms. Profitability Inconclusive Brewer 1981 Investor Benefits from corporate international diversification 151 US-based MNEs; 137 US Nationals. Jan 1963-Dec 1975 Profitability Advantage MNE Source: Own elaboration.