Business investment research

small business investment research business investment research paper
Dr.ThorasRyder Profile Pic
Dr.ThorasRyder,Hong Kong,Researcher
Published Date:07-07-2017
Your Website URL(Optional)
Internationalisation of business investments in R&D EUR 25195 EN Research and Innovation EUROPEAN COMMISSION Directorate-General for Research and Innovation Directorate C — Research and Innovation Unit C.6 — Economic analysis and indicators E-mail: Contact: Matthieu Delescluse European Commission B-1049 BrusselsEUROPEAN COMMISSION Internationalisation of business investments in R&D and analysis of their economic impact Authors of the study Bernhard Dachs, Franziska Kampik, Thomas Scherngell, Georg Zahradnik AIT Austrian Institute of Technology Doris Hanzl-Weiss, Gabor Hunya, Neil Foster, Sandra Leitner, Robert Stehrer, Waltraut Urban The Vienna Institute for International Economic Studies - wiiw 2012 EUR 25195 EN Directorate General for Research and InnovationEUROPE DIRECT is a service to help you find answers to your questions about the European Union Freephone number (): 00 800 6 7 8 9 10 11 () Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed LEGAL NOTICE Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of the following information. The views expressed in this publication, as well as the information included in it, do not necessarily reflect the opinion or position of the European Commission and in no way commit the institution. More information on the European Union is available on the Internet ( Cataloguing data can be found at the end of this publication. Luxembourg: Publications Office of the European Union, 2012 ISBN 978-92-79-22835-3 ISSN 1831-9424 Doi 10.2777/6055 © European Union, 2012 Reproduction is authorised provided the source is acknowledged. Images cover: earth, © 2520287, 2011. Source:; bottom globe, © PaulPaladin 11389806, 2012. Source: Contents 1. INTRODUCTION 2 2. LITERATURE SURVEY 3 2.1. Drivers of R&D internationalisation 3 2.2. Impacts of foreign-owned R&D and innovation activities on host countries 7 2.3. Impacts of overseas R&D and innovation activities on home countries 10 3. ISSUES IN COLLECTING DATA ON THE INTERNATIONALISATION OF R&D 12 3.1. Definitions of R&D and innovation 12 3.2. Methodology, data sources and challenges 13 3.3. Experiences from the data collection 14 3.4. The most pressing data needs 15 4. CURRENT STATUS AND FUTURE DIRECTIONS OF R&D INTERNATIONALISATION 16 4.1. Inward BERD across countries and over time 16 4.2. The main countries of origin of inward BERD 20 4.3. Outward BERD across countries and over time 22 4.4. Inward BERD across sectors and industries 24 4.5. The relationship between the European Union and United States of America 28 4.6. New players in the internationalisation of R&D 31 5. THE CROSS-COUNTRY STRUCTURE OF R&D INTERNATIONALISATION 34 5.1. The network perspective 34 5.2. Some descriptive analyses from a social network perspective 34 5.3. The strength of inward BERD links between individual countries 38 6. R&D ACTIVITIES OF NON-EUROPEAN COMPANIES IN THE EUROPEAN RESEARCH AREA 40 6.1. Evidence from inward BERD data 40 6.2. The relationship between foreign direct investment and outward BERD 43 6.3. Evidence from the Community Innovation Survey 2008 43 7. DRIVERS OF R&D INTERNATIONALISATION 46 7.1. Internationalisation of production and R&D 46 7.2. The relationship between R&D intensities of domestic and foreign-owned firms 49 7.3. Host country determinants of R&D internationalisation 52 7.4. Host and home country determinants of R&D internationalisation 56 8. IMPACTS OF R&D INTERNATIONALISATION 60 8.1. The impact of inward BERD on domestic BERD and R&D intensity 60 8.2. The impact of inward BERD on domestic labour productivity and employment 62 8.3. The impact of inward BERD on domestic patenting activities 64 9. CLOSING REMARKS 66 10. ACKNOWLEDGEMENTS 67 11. REFERENCES 681. INTRODUCTION This report summarizes the outcomes of the project. It starts with a survey of the literature on business R&D inter- nationalisation in chapter two. Chapter three describes how In a seminal paper on research and development (R&D) in we collected data, points to various challenges and pitfalls large multinational enterprises, Pari Patel and Keith Pavitt in the interpretation of this data and discusses the most concluded in 1991 that the production of technology re- pressing data needs. mains “far from globalized”, but concentrated in the home countries of the enterprises (Patel and Pavitt 1991, p. 17). Chapter four investigates the current status and future di- In their words, research and development is “an important rections of R&D internationalisation, identifies the countries case of non-globalisation”. and sectors where the share of foreign-owned firms on R&D expenditure is particularly high, looks at R&D expenditure 20 years later, a vast amount of evidence draws a differ - of domestic firms abroad, and at the relations between ent picture of R&D internationalisation: Enterprises not only the United States and the European Union in the interna- produce and sell, but increasingly also develop goods and tionalisation of R&D. The chapter finishes with a view on services outside their home countries. Today, it seems to be new players from emerging economies. Chapter five looks the rule, not the exception, that large European firms have at R&D activities of non-European companies in the Euro- R&D activities at different locations inside and outside the pean Research Area. Chapter six analyses the structure of Single Market. In addition, firms from the United States and R&D internationalisation from a network perspective. The other non-European countries have considerably extended potential drivers of R&D internationalisation are studied in their R&D activities in the European Union, and new players chapter seven whereas Chapter eight looks at the impacts from emerging economies are entering the scene: India, of R&D of foreign-owned firms on domestic R&D perfor - the People’s Republic of China (referred to as China in the mance and other economic indicators. Chapter seven and report), and other locations have come into focus as host eight apply econometric methods as well as provide evi- countries for the R&D activities of European multinational dence from case studies. Finally, chapter nine summarizes enterprises (MNEs) in recent years. key messages from the study and draws conclusions. These developments form the background of the project “Internationalisation of business investments in R&D and analysis of their economic impact”. More specifically, the project pursues the following aims: • Collect data on R&D expenditures by foreign-owned firms; • Identify the most pressing needs for new data on R&D expenditure of foreign-owned firms; • Describe and analyse business R&D internationalisation at the national level, the sectoral level and in a cross-country perspective; • Analyse the drivers and impacts of the internationalisation of R&D on home and host countries. 22. LITERATURE SURVEY A first important regional or country level driver is income and market size. Empirical evidence indicates that the in- ternationalisation of R&D predominantly takes place be- This section summarizes the key findings of the literature tween high-income countries. Income is an important driver on the internationalisation of R&D focussing on three is- for various reasons; first, high income and high income sues: first, it reviews the drivers of the process - why firms growth attracts FDI (Ekholm and Midelfart 2004; Blonigen go abroad with R&D activities. Second, we discuss recent 2005; Jensen 2006). R&D investments often follow FDI, findings on the effects of the internationalisation of R&D on and overseas R&D activities are, in most cases, an exten- the host countries. Finally, it analyses the impacts of R&D sion of existing overseas production and marketing activi- internationalisation on the home countries of the firms. ties (Birkinshaw and Hood 1998; Birkinshaw et al. 1998; Archibugi and Iammarino 1999). Moreover, firms may find This survey has two important limitations: first we will not it easier to cover the cost of R&D in a country with a large include the literature on the internationalisation of R&D at market where they expect larger absolute revenues than in universities or public research centres. Second, the litera- a country with a small domestic market, even if wages are ture on foreign direct investment (FDI) and multinational considerably lower. enterprises is only covered if it is related to R&D. Interna- tionalisation refers to the internationalisation of business Another important attractor of foreign R&D is a skilled R&D through the remainder of this chapter, unless other- workforce and the quality of education systems (Thursby wise stated. and Thursby 2006; Kinkel and Maloca 2008; European Commission 2010). Skills shortage and a growing demand The oldest literature on the internationalisation of R&D for engineers and scientists in the home country is often dates back to the end of the 1960s and the beginning a motive for firms to go abroad with R&D. Ernst (2006), of the 1970s (e. g. Dunning 1958; Brash 1966; Safarian for example, relates the success of India and other Asian 1966). Only few articles and surveys emerged in the 1970s countries in attracting foreign R&D to their expanding pool (e. g. Creamer 1976; Ronstadt 1977; Lall 1979) and in the of graduates in science and technology. Hedge and Hicks 1980s (Behrman and Fischer 1980). Since the early 1990s (2008) demonstrate that the innovation activities of over- a growing body of empirical literature provides empirical seas US subsidiaries are strongly related to the scientific evidence that the internationalisation of R&D is gaining and engineering capabilities of the host countries. In the re- momentum (e. g. OECD 2005a; UNCTAD 2005; Hatzichrono- search of Lewin et al. (2009), an emerging shortage of high glou 2008; OECD 2008a; OECD 2008b; OECD 2008c; OECD skilled science and engineering talent partially explains the 2010). This literature has been accompanied by contribu- relocation of product development from the United States tions studying the motives and strategies of firms, among to other parts of the world, most notably Asian countries. them Cantwell (1989), Pearce (1989, 1992), or Bartlett and Kinkel and Maloca (2008) find that capacity bottlenecks are Ghoshal (1990). the most frequent reason why German firms move R&D to locations abroad. 2.1. Drivers of R&D Closely related to the quality of the education system are also knowledge spillovers between foreign-owned firms and internationalisation host country organisations (see section 2.2 for a detailed discussion). The search by foreign-owned firms for such The internationalisation of R&D is the product of a number spillovers is known as the asset-seeking motive (Dunning of actors at different levels. The most important actors are and Narula 1995), home-base augmenting strategy (Kue- firms which, for a variety of reasons, decide to do R&D at lo - mmerle 1999), or global R&D strategy (von Zedtwitz and cations abroad. The benefits and costs associated with the Gassmann 2002) in the literature. Spillovers as a determi- internationalisation of R&D vary between firms, industries, nant for R&D location decisions point to the importance of regions or countries. It is therefore important to distinguish the quality of university research as a driver of R&D inter- between these three levels. We will start with a discussion nationalisation at the country level (Belderbos et al. 2009; of the drivers at the regional and country level and then go Dachs and Pyka 2010). to the sectoral and firm level. Knowledge spillovers may be even more relevant at the re- The host country or the region shapes the internationali- gional than at the country level, because spillovers dimin- sation decisions of firms by providing different framework ish with distance between sender and receiver (Jaffe et al. conditions for R&D. Drivers at the regional or country level 1993; Breschi and Lissoni 2001). As a consequence, firms are also important from a policy perspective, because they which want to utilize such localized knowledge spillovers give room for policy intervention to increase the locational have to be present where they occur, and innovative activ- advantages of regions or countries. ity tends to cluster locally in industries with a high level of spillovers (Audretsch and Feldman 1996). This effect is related to institutional or technological conditions, such as 3tacitness of the knowledge base, but also to the existence provide a healthy business environment, political stabil- of specialized local or regional labour markets (see the sur- ity, good public infrastructure, reasonable tax rates, and a vey of Breschi and Lissoni 2001). stable legal system including the protection of intellectual property rights. This opinion is based on empirical studies Differences in labour cost between the home country and on the location decisions of MNE R&D activities (Cantwell locations abroad are one of the most important motives for and Mudambi 2000; Kumar 2001; Cantwell and Piscitello the internationalisation of production. Empirical evidence 2002; Thursby and Thursby 2006; Kinkel and Maloca 2008; that differences in the cost of R&D personnel are a major European Commission 2010), but also on the political prac- driver for the internationalisation of R&D, however, is weak; tice in Europe, which is characterized by non-discrimination survey results as well as econometric studies see only a of foreign-owned firms (Dachs et al. 2005; Guimón 2009). modest influence of cost advantages on R&D location de - cisions compared to other factors (Booz Allen Hamilton In addition, science, technology and innovation (STI) policy and INSEAD 2006; Thursby and Thursby 2006; Kinkel and measures can significantly shape locational advantages Maloca 2008; Belderbos et al. 2009; European Commis- and influence internationalisation decisions of firms in R&D. sion 2010). However, cost differences may gain importance This includes all measures to stimulate the creation, dif- when firms consider to locate R&D and innovation activities fusion and utilisation of new knowledge and technologies in emerging economies, or when firms have to choose be - (Steinmueller 2010). Examples of such measures are public tween two similar attractive locations (Booz Allen Hamilton subsidies for R&D performing firms or measures to foster and INSEAD 2006; Thursby and Thursby 2006; Cincera et co-operation between firms and universities. Science pol - al. 2009). icy also includes university education - the availability of skilled researchers is one of the most important location Previous research has also pointed out that geographical criteria for R&D. proximity between host and home country leads to higher levels of cross-border R&D investments (Guellec and van There is a consensus in the literature that special incentives Pottelsberghe de la Potterie 2001; Dachs and Pyka 2010). to foreign-owned firms are not an appropriate instrument This distance effect is often explained by additional co-or - to attract R&D of foreign-owned firms. This consensus is dination cost, the cost of transferring knowledge over dis- based on empirical studies on the location decisions of R&D tance, and a loss of economies of scale and scope when R&D activities of MNEs (Cantwell and Mudambi 2000; Kumar becomes more decentralized (von Zedtwitz and Gassmann 2001; Cantwell and Piscitello 2002; Thursby and Thursby 2002; Gersbach and Schmutzler 2006; Sanna-Randaccio 2006; Kinkel and Maloca 2008; European Commission and Veugelers 2007). In addition, the distance effect may 2010), but also on the political practice in Europe, which is also be explained by cultural, social and institutional fac- characterized by non-discrimination of foreign-owned firms tors. The international management literature stresses that (Guimón 2009). However, it is also clear from the literature foreign firms have to master additional institutional and that public support for R&D, including tax incentives, can cultural barriers in their host countries. This disadvantage create important additionalities and can help leverage R&D is known as the ‘liability of foreignness’ (Zaheer 1995; Eden efforts of firms, including foreign-owned firms. Hence, public and Miller 2004) or the ‘liability of outsidership’ (Johanson support may not be appropriate to attract R&D or foreign- and Vahlne 2009) in the literature. It may include a lack owned firms but helps to further augment R&D expenditure of market knowledge and understanding of customer de- of foreign-owned firms already located in a host country. mands, but also a lower degree of embeddedness in infor- mal networks in the host country. Disadvantages from the A second important level for the analysis of drivers is the liability of foreignness tend to decrease over time, but may industrial sector of the firm. The industry matters in two even exist in long-established affiliates with a local man - ways: on the one hand, there are large differences between agement and staff, because the subsidiary is still embedded sectors in terms of foreign direct investment, and sectors in intra-firm networks and have to stick to the rules, norms with high shares of inward FDI also tend to be technologi- and standards of the multinational group. cally intensive (Markusen 1995, p. 172). Hence, R&D inter- nationalisation can mainly be observed in R&D or knowl- Finally, public policy can considerably shape the attractive- edge-intensive sectors. ness of regions or countries for overseas R&D activities. Recent surveys of policy measures in the field of R&D inter - On the other hand, R&D processes differ considerably nationalisation include Dachs et al. (2005), UNCTAD (2005), across sectors. Firms in the same industry operate with a CREST (2007), OECD (2008a), TAFTIE (2009), Verbeek et al. (mostly) similar knowledge base and have to solve similar (2009) and Schwaag Serger and Wise (2010). problems in the innovation process. These intersectoral dif- ferences shape innovation behaviour of firms to a consider - There is a consensus in this literature that governments that able degree, resulting in vast differences between sectors in want to attract R&D of foreign multinational firms should many R&D and innovation indicators (Marsili 2001; Malerba focus on the economic fundamentals rather than grant spe- 2005; Castellacci 2007, Peneder 2010). We may assume cial incentives to foreign-owned firms. Governments should that the same factors also alter decisions to locate R&D 4abroad, leading to different degrees of internationalisation (Cohen et al. 2000; Cohen 2010). Firms in sectors with a low at the sectoral level. degree of appropriability, like many service sectors, may be reluctant to internationalize R&D because they have only A first important determinant at the industry level is the weak means to prevent involuntary knowledge spillovers. degree of tacitness of the knowledge base of a sector. Tac- itness results from the fact that cognitive capabilities and Forth, another potential source for inter-sectoral differences abstract concepts are not easy to articulate explicitly and to is the firm’s network of external relations with suppliers, transfer between people (Cowan et al. 2000). A knowledge clients, universities, public authorities etc (Marsili 2001; base which is highly tacit and bound to individuals may be Malerba 2002). Some industries, such as biotechnology an obstacle to internationalisation, because it makes knowl- or pharmaceuticals, have strong linkages to basic science. edge exchange over distance costly. Tacitness, however, Firms in these industries may find it useful to locate R&D may also be a driver for internationalisation, because firms close to excellent research universities. Firms in other sec- have to move to the place where this knowledge is available tors, such as the automotive of the electronics industry, when it cannot be transferred over distance. are closely connected to suppliers and customers through international production networks. Suppliers in these sec- Second, sectoral knowledge bases also differ in their degree tors may be forced to internationalise their R&D to have of cumulativeness, or, in other words, in the degree future development capabilities in proximity to key clients. The ex- innovation success depends on the knowledge which has istence of lead users or other potential co-operation part- been built up in the past (Marsili 2001). Cumulativeness ners may also pose a strong incentive to locate R&D in a is high in chemicals, pharmaceuticals, telecommunications particular country. and electronics, and low in mechanical engineering, food, clothing, or civil engineering (Malerba and Orsenigo 1996; Finally, the firm level is decisive for the explanation of over - Marsili 2001). A high degree of cumulativeness may require all patterns of R&D internationalisation. Internationalisation a high degree of specialisation in R&D, which gives advan- paths of two firms can be completely different - even if tages to centralized R&D. Cumulativeness may also pro- they are located in the same country and region and op- mote R&D centralisation when strong learning effects lead erate in the same industry - because firms differ in their to increasing returns to scale in R&D, or when the R&D pro- capabilities, characteristics, organisation and strategies. cess includes economies of scope and effects from cross- The interplay of firm characteristics, firm motives and strat - fertilisation. Moreover, cumulativeness of the knowledge egies and the benefits and costs that arise from interna - base may also imply that R&D activities require a certain tionalisation, together with framework conditions from the minimum scale in order to be successful. country, egional and sectoral level, determines the degree of R&D internationalisation of firms (see Figure 1 below). Third, sectors also differ in terms of appropriability, the de - gree to which an innovation can be protected from imitation Figure 1: Determinants of R&D internationalisation at the firm level Firm motives and strategies Benefits and costs of Firm Characteristics internationalisation Source: own illustration 5Decisions on exports and FDI and R&D strategies of firms al. 1997; Verona 1999). These capabilities enable the firm are mutually connected. Joseph Schumpeter (1911) already to create new knowledge, but also absorb knowledge from regarded the opening of new markets as a type of innova- external sources. tion, together with product, process, organisational innova- tion and the conquest for a new source of supply. All come Besides firm characteristics, there is also a considerable in - from the same source: entrepreneurship. fluence of firm strategy and managerial intentionality on the internationalisation R&D. It is not enough that firms op - More recent contributions give evidence that R&D and in- erate in sectors with a high degree of internationalisation; novation intensity is positively related to FDI and exports. the head office of the MNE has to allow a higher degree of Theoretical as well as empirical research argues that firm decentralisation by changing firm organisation and giving a heterogeneity leads to self-selection in the internationali- higher degree of autonomy to the subsidiaries (Birkinshaw sation strategies of firms (Head and Ries 2003; Helpman et and Hood 1998; Birkinshaw et al. 1998; Zanfei 2000). al. 2004; Helpman 2006). Only the most productive firms expand their operations via FDI, while less productive firms This decision may be influenced by the costs of a decentral - choose to export or serve only domestic markets. FDI (and ized organisation of R&D (Gersbach and Schmutzler 2006; hence MNEs) exists because firms possess superior, firm- Sanna-Randaccio and Veugelers 2007). These costs first specific assets and exploit these assets at foreign mar - comprise the foregone benefits of R&D centralisation, in - kets via their subsidiaries (Dunning 1973; Markusen 1995; cluding economies of scale and scope from specialisation, Caves 1996 (1974); Markusen 2002). Dunning (1973; or a tighter control over core technologies of the firm. Sec - 1981) suggests that firms exploit these assets via FDI and ond, additional costs also arise from higher co-ordination not via exports or licensing because of ownership, location efforts and the cost of transferring knowledge within the and internalisation advantages associated with this mode MNE. Proximity also facilitates co-ordination of R&D and of exploitation. innovation activities with other parts of the firm, such as production and marketing (Ketokivi and Ali-Yrkkö 2009). In addition, there is also evidence for a positive relationship Third, a concentration of R&D activity in the home coun- between innovation and exports at the firm level (Green - try is also favoured by various linkages between the firm halgh and Taylor 1990; Lachenmaier and Wößmann 2006; and the host country innovation system. Patel and Pavitt Harris and Li 2009). Export experience or experience with (1999) or Narula (2002) point out that firms are strongly production in foreign markets is an important pre-requisite embedded in and dependent on their embeddedness in the for the internationalisation of R&D. The internationalisation home country innovation system. The ties that bind firms of R&D follows the internationalisation of other economic to their home country include relations with external actors activity, in particular production and sales; overseas R&D such as formal R&D co-operations with domestic universi- is in most cases an extension of existing overseas produc- ties, but also informal networks that grew from doing busi- tion and marketing activities (Birkinshaw and Hood 1998; ness together in the past. Informal networks between firms Birkinshaw et al. 1998; Archibugi and Iammarino 1999). may also evolve from joint training of staff at universities and labour mobility. Removing these linkages by moving We can therefore assume that there is a mutual relation- R&D abroad would incur considerable costs on the firms, ship between R&D and international activities. This rela- because they would need to re-install similar linkages with tionship also affects the internationalisation of R&D. Firm host country organisations. characteristics that drive internationalisation are also posi- tively related to the propensity to do R&D and R&D inten- The costs of R&D internationalisation have to be seen sity. Empirical analysis (Dogson and Rothwell 1994; Cohen alongside the benefits of R&D internationalisation. A fist 1995; Kleinknecht and Mohnen 2002; OECD 2009a; Cerrato benefit is that R&D can support overseas production. Prod - 2009) has examined the determinants of R&D and innova- ucts and technologies often have to be adapted to consum - tion in detail, so we will only give a very short overview of er preferences, regulation, or environmental conditions of this literature. foreign markets in order to facilitate their exploitation in these markets. These adaptations can be done more easily R&D and R&D intensity is, at first, associated with firm size. in proximity to potential clients in the host countries. MNEs There are different advantages and disadvantages of small therefore locate design, engineering and R&D units in main and large firms in the innovation process, leading to a U- foreign markets to support marketing and production fa- shaped relationship between size and R&D (Kleinknecht cilities abroad. There are various names for this motive in 1989; Cohen 1995). Regression analysis also finds a signifi - the literature, including asset-exploiting behaviour (Dunning cant and positive association between firm size and the in - and Narula 1995), competence-exploiting subsidiary man- ternationalisation of R&D or innovation activities (Arvanitis dates (Cantwell and Mudambi 2005), home-base exploiting and Hollenstein 2006; Kinkel and Maloca 2008; Schmiele strategies (Kuemmerle 1999), or market-driven interna- 2009). Innovativeness and R&D is also positively related to tionalisation of R&D (von Zedtwitz and Gassmann 2002). the internal knowledge and capabilities of the firm (Cohen and Levinthal 1989; Cohen and Levinthal 1990; Teece et 6A second benefit and important driver of R&D internation - by internal R&D, but also from a broad range of external alisation at the firm level is access to knowledge and the sources and actors. In this respect, asset-seeking can be creation of new knowledge abroad. This is known as the seen as a variant of ‘open innovation’ strategies with a fo- asset-seeking motive (Dunning and Narula 1995), compe- cus on their geographical dimension. tence-creating subsidiary mandate (Cantwell and Mudam- bi 2005), home-base augmenting strategy (Kuemmerle There is evidence that asset-seeking strategies have be- 1999), or global R&D strategy (von Zedtwitz and Gassmann come more frequent in the recent years, although asset- 2002) in the literature. exploiting strategies still prevail (Narula and Zanfei 2005; Sachwald 2008). Moreover, some authors (for example Asset-seeking strategies are driven, on the one hand, by Criscuolo et al. 2005) stress the fact that the two motives the existence of superior local knowledge and favourable cannot be separated in a number of cases. Firms – inten- framework conditions for R&D in various host countries. tionally or unintentionally - often follow both strategies si - Some types of knowledge are tacit, bound to their local multaneously. Microsoft’s efforts to adapt their products to context, and transferable over distance only at high costs the Chinese language resulted in new knowledge that could (Cowan et al. 2000; Breschi and Lissoni 2001). This knowl- also be used in other contexts (Gassmann and Han 2004). edge may be found at universities and other research or- ganisations, in clusters, or be available from clients, suppli- ers or competitors. Various authors describe foreign-owned 2.2. Impacts of foreign-owned subsidiaries as ‘surveillance outposts’ or ‘antennas‘ (Florida R&D and innovation activities on 1997; Almeida 1999) that extensively monitor and assimi- late knowledge from local sources. host countries On the other hand, asset-seeking strategies may also be The technological and economic characteristics of countries driven by factors related to the nature of various tech- provide different locational advantages and disadvantages nologies and changing firm strategies. Narula and Zanfei for foreign-owned firms to set up R&D activities. In turn, the (2005) for example, suggest that the increasing complexity R&D activities of foreign-owned firms may also influence of products is a driver of the internationalisation of R&D. the innovation systems of their host and home countries to Rising technological complexity increases the knowledge a considerable degree. requirements of firms and forces them to search for new knowledge abroad. A similar argument is brought forward The literature has identified various potential challenges by Chesbrough (2003a; 2003b). He points out that many and opportunities for host and home countries from the innovative firms have shifted to an ‘open innovation’ model internationalisation of R&D and innovation (see Table 1). where they exploit ideas and knowledge not only provided Table 1: Potential opportunities and challenges for national innovation systems from the internationalisation of R&D and innovation Opportunities Challenges & Risks  Increases in aggregate R&D and innovation  Loss of control over domestic innovation expenditure capacity and commercialisation  Knowledge diffusion to the host economy  Less strategic research, less radical innovations, more adapting  Demand for skilled personnel  Separation of R&D and production  Structural change and agglomeration effects  Competition with domestically owned firms for resources (‘Crowding out’)  Improved overall R&D efficiency  Loss of jobs due to relocation  Reverse technology transfer  ‘Hollowing out’ of domestic R&D and innovation activities  Market expansion effects  Technology leakage and involuntary  Exploitation of foreign knowledge at home knowledge diffusion Source: Adapted from Sheehan (2004), UNCTAD (2005), Veugelers (2005). 7 Home Country Host country Most of these effects are rooted in the following three im - should try to attract inward investment. Empirical evidence portant facts about multinational firms: on the size and the effects of these spillovers, however, is mixed. Meta-studies (Görg and Greenaway 2004; Mayer First, economic theory states that multinational firms pos - and Sinani 2009; Havránek and Iršová 2010) as well as sess some types of valuable intangible assets – technolo- literature surveys show no clear relationship between for- gies, knowledge, trademarks, business practices, etc. – that eign presence and the performance of domestically owned allow them to enter foreign markets and exploit these as- firms. Görg and Strobl (2001) for example indicate that the sets internationally. These assets may be – voluntarily or number of studies that identify positive spillovers roughly involuntarily - transferred to the host economy through equals those identifying no effects or even negative conse - various channels, generating considerable benefits for do - quences from the presence of foreign-owned enterprises. In mestic organisations. the majority of cases considered by Görg and Greenaway (2004), no significant effect of MNE presence on domestic Second, the size of R&D expenditure of multinational firms firm productivity is observed. Reinhilde Veugelers (2005, is considerable, even if we compare it with aggregate R&D p 37) finds that it is “fair to conclude that the results on expenditure of countries (OECD 2010, p. 121). Foreign-con- positive spillovers on host economies are not strong and trolled R&D expenditure exceeds public funding of R&D in robust”. the business sector in a number of countries (OECD 2009b, p. 120f). Hence, a new R&D venture of a foreign-owned firm Empirical evidence is clearer at the aggregate level. New may affect aggregate R&D activity in the host country. contributions by Keller and Yeaple (2009) and by Coe, Helpman and Hoffmaister (2009) reveal substantial spillo - Finally, the literature has pointed out that multinational ver effects from foreign R&D stocks and the presence of firms differ in their R&D strategies from firms that operate foreign-owned firms at the sectoral level. Moreover, Coe et only in one country (Narula and Zanfei 2005; Dunning and al (2009) show the importance of institutional factors and Lundan 2009). MNEs have various choices in the location thus institutional distance for the degree of R&D spillovers. and organisation of R&D which mono-national firms do not have. They can, for example, move R&D activities between A main reason for this vagueness of the results, besides countries. This has implications for policy. measurement and estimation issues, is the fact that spillo- vers from foreign-owned firms to the local economy are We first discuss impacts from the perspective of the host bound to specific industry and economy-wide conditions country. The presence of foreign-owned firms can help to to occur. There have to be specific factors or conditions in increase aggregate R&D and innovation expenditure. R&D place for the host countries to benefit from activities of expenditure by MNEs constitutes a considerable share of foreign-owned firms. These factors include a certain level gross R&D expenditure in many countries and often exceeds of absorptive capacity (Cohen and Levinthal 1989; 1990; public funds for R&D (Lonmo and Anderson 2003; Costa Cantner and Pyka 1998) of domestic organisations; weak and Filippov 2008). Empirical evidence suggests that small instruments of foreign-owned firms to protect proprietary countries benefit most in relative terms because they usu - knowledge, which is mostly sector-specific; and the propen - ally exhibit higher degrees of internationalisation in FDI sity of the transfer channel or type of interaction between than large countries. foreign-owned firms and domestic organisations. MNE subsidiaries – in contrast to domestically owned firms Foreign-controlled R&D activity in a particular country may - can also access financial means of their parent enterprise also help to enhance the level and quality of human re- abroad; expansion of R&D activity is therefore not limited sources. R&D activities of foreign-owned firms may create by a lack of internal resources or incomplete credit markets additional demand for researchers and give incentives to in the host country. Moreover, the threat of market entry by governments to improve higher education systems. MNEs R&D intensive MNEs may also spur R&D activities of do- are attractive employers, because they can offer inter - mestically owned firms (Aghion et al. 2009). national career perspectives and pay higher wages than domestically owned enterprises (Lipsey 2002; Bailey and A second, more indirect benefit for the host country is the Driffield 2007). Moreover, jobs created by foreign-owned diffusion of information and knowledge (referred to as firms appear to be more persistent than jobs generated in knowledge diffusion in the text) to host country organisa - domestically owned plants (Görg and Strobl 2003). There tions. Potential receivers of this knowledge are domestic may, however, also a challenge from the presence of for- firms, universities, or research centres. eign-owned firms when domestic and foreign-owned firms compete for skilled personnel. The literature gives considerable attention to knowledge diffusion and spillovers by foreign-owned firms (see, for ex - Finally, foreign-owned firms can also contribute to struc - ample, the surveys by Keller (2004 and 2010) or Mayer and tural change towards a higher share of technology-inten- Sinani (2009)). According to Blomström and Kokko (2003), sive firms and the emergence of clusters. Structural change spillovers are the strongest argument as to why countries is related in two ways to the presence of foreign-owned 8firms. On the one hand, foreign-owned firms operate pre - in terms of novelty and originality. R&D of foreign-owned dominantly in technology-intensive industries. Market en- firms may be associated with a higher degree of adaptation trance and subsequent growth of the foreign-owned firm and less basic, strategic research, leading to fewer radical will therefore move the industrial structure of a country innovations than in the case of domestic ownership. towards higher technology intensity. There is also evidence that FDI contributes to the shift in labour demand towards Empirical evidence that supports these concerns, however, skilled labour in the host country (Blonigen and Slaughter is thin. Internationalisation certainly leads to a shift of con - 2001; Driffield et al. 2009). trol from domestic headquarters to organisations abroad. However, we also have to consider that domestic ownership On the other hand, MNE subsidiaries trigger structural does not necessarily mean that enterprises act in best na- change because their demand for inputs favours the tional interest. Moreover, domestic policy does not neces- growth of domestic technology-intensive suppliers. This sarily have a higher ability to influence R&D decisions when demand may lead to the emergence of clusters and other enterprises are domestically owned (Dunning and Lundan agglomerations at the regional or local level in the host 2008, p. 249 ff). country (Young et al. 1994; Bellandi 2001; Pavlínek 2004). Agglomeration effects may be further intensified by the de - The question if foreign ownership is associated with a gree foreign-owned firms are embedded into their local en - downsizing of R&D activity has been evaluated both for vironment. Foreign-owned subsidiaries in clusters are often take-overs as well as for all foreign-owned and domesti- strongly embedded locally, but also internationally-oriented cally owned firms. In the case of take-overs, there are both and can therefore act as bridges for knowledge transfer be- examples of downsizing as well as examples of expansion, tween domestic organisations and abroad (Birkinshaw and depending on the complementarity between acquiring and Hood 2000; Lorenzen and Mahnke 2002). acquired firms and other factors (Cassiman et al. 2005; UNCTAD 2005). Studies that compare innovation input and We now turn to potential challenges for host countries that output of domestically owned and foreign-owned firms find emerge from the presence of foreign-owned firms. One no negative effect of foreign ownership after controlling for interesting aspect of the literature on spillovers from FDI firm characteristics such as size, sector, export intensity etc. is the considerable number of studies that report negative effects of the foreign presence on domestic firms (see, for R&D internationalisation may also lead to a separation of example, Aitken and Harrison 1999; Konings 2001; Castel- R&D and production. Multinational firms do not necessar - lani and Zanfei 2002; Damijan et al. 2003). An example is ily do research, development and production at the same a recent study by Wang (2010) investigating the determi- place. They have various choices in locating their activi- nants of R&D investment at the national level for 26 OECD ties, which may lead to a separation of innovation, R&D countries from 1996-2006. Wang (2010) finds that foreign and production (Pearce 2004; Pearce and Papanastassiou technology inflows through trade and FDI had a robust and 2009). MNEs may find it useful to develop products in one negative impact on domestic R&D. One explanation put for- country and manufacture those products in another country ward for this negative impact is that increased competition where conditions for production seem more favourable. As in product and factor markets can have a negative impact a consequence, policy measures to foster R&D and prod- on a domestic firm’s productivity (Aitken and Harrison 1999; uct development may yield only few jobs and give only a Konings 2001). These negative effects are predominantly weak stimulus to growth, when foreign-owned firms decide found in developing or transition economies. A stronger to produce abroad. competition from foreign-owned firms may also reduce R&D activities of domestically owned firms, because it de - To our knowledge, no empirical study so far has thoroughly creases expectations of future demand. examined the effects from the separation of R&D and pro - duction. It is, however, plausible that this leaking-out effect Concerns that foreign presence may lead to a downgrad- may be stronger in small countries and in countries with a ing of domestic R&D are nurtured by more general doubts high share of foreign-controlled R&D, and weaker or even against MNE activities (see Barba Navaretti and Venables reverse when foreign-owned firms have a high degree of 2004; Jensen 2006; Forsgren 2008 for a summary of autonomy and strong mandates in their enterprise groups, this discussion). One concern is that decisions on R&D of because these firms may try to concentrate not only R&D, foreign-owned firms may not be taken by the subsidiaries but also production at their location of maximise influence. themselves, but by corporate headquarters abroad. Others The effect may level out when studied at the EU instead of fear that MNEs are ‘footloose’, because they mainly pursue the national level. economic activities that can be easily transferred between countries; foreign-owned enterprises act in ways that are Foreign-owned firms may also compete with domestical - not in accordance with the national interest; they show ly owned firms for resources. Foreign-owned subsidiaries rent-seeking behaviour in selecting locations and try to are attractive employers for researchers and other R&D undermine national labour standards. Another fear is that staff. This demand for skilled personnel is beneficial for the foreign ownership may change the nature of business R&D host country in the short run when there is unemployment 9among scientists, engineers and technicians and alterna- ties abroad can therefore strengthen the growth of the par- tive employment opportunities (for example at universities) ent company in the home country (Rammer and Schmiele are scarce (Marin and Sasidharanb 2010). Additional de- 2008). The degree these benefits occur depends on the mand by foreign-owned firms, however, may have nega - motives for overseas R&D activities of domestic firms, as tive consequences for the host country when the supply for well as on the degree of complementarity between over- research personnel is inelastic and foreign-owned firms and seas and home activities (Arvanitis and Hollenstein 2009), domestic organisations compete for qualified staff (Figini and the absorptive capacities and other firm characteristics and Görg 1999; Driffield and Taylor 2000). of the parent company (Schmiele 2009). Moreover, there seems to be a positive relationship between internation- In the long run, the effects of the demand by foreign-owned alisation and the returns from R&D at home (Añón Higón subsidiaries on the labour market for R&D staff look more and Manjón Antolín 2009; Criscuolo and Martin 2009) which positive. Stronger demand for high-skilled labour due to may further increase the benefits for the home country. market entry of foreign-owned firms and structural change may foster academic training and increase the number Todo and Shimiztuani (2005) find for Japan that the scope of graduates in science and technology in the long run. A for positive reverse technology transfer effects on the pro - higher skill intensity in the economy, in turn, may foster lo- ductivity of firms in the home country is large when for - cational advantages and further increase the attractiveness eign-owned affiliates undertake R&D that tap into advanced of the country for inward investment. Barry (2004) illus- knowledge abroad. Adaptive R&D however was found to im- trates such a ‘virtuous circle’ for the case of Ireland. prove productivity in the host country, but did not contribute to enhanced productivity in the home country. Griffith et al. (2004) find that R&D by UK firms in the USA have resulted 2.3. Impacts of overseas R&D in benefits from reverse technology with the effects being larger in the case of R&D units set up to source technology. and innovation activities on home Results for Sweden, however (Fors 1997; Braconier et al. countries 2002) indicate that there have not been significant spillo - vers to the home country, possibly because much R&D has The internationalisation of R&D has implications for the been of the adaptive type. AlAzzawi (2004) find that out - home country of the multinational firm (Dunning and Lun - ward-FDI-induced R&D had a positive impact on the home dan 2009). Before we briefly discuss the literature on home country’s level of innovation activity in both developed and country effects, two remarks are important. First, as Kokko newly industrialized countries, but productivity benefits (2006) points out, the decision to engage in foreign activi- were found for newly industrialized countries only. ties is typically a voluntary decision and it thus can be as- sumed that overseas activities benefit the MNE. More spe - Potential challenges or costs from the internationalisation cifically, it can be expected that the MNE will grow larger of R&D for the home country may arise when firms replace than what would have been possible if it had remained a domestic R&D and innovation activities with similar activi- purely national firm. ties abroad. This may lead to a hollowing out of domestic innovation capacity, a loss of jobs in R&D, and a downward Second, as discussed above, a main reason for firms to go pressure on wages of R&D personnel in the home country. abroad with R&D activities is to get access to knowledge not available in the home country. Hence, a first main ben - Despite public discussions on the offshoring of R&D and efit for the home countries is the transfer of knowledge possible consequences for home country innovation sys- 1 resulting from overseas R&D activities which brings new tems , empirical results that confirm negative effects from knowledge into the home country. Various studies pro- overseas R&D on the home country are rare. A reason for vide evidence for such reverse knowledge transfers (Fors this is the fact that the most R&D activities of domestic 1997; AlAzzawi 2004; Feinberg and Gupta 2004; Rabbiosi firms is still located in the home country (see section 4.3). 2005; Todo and Shimizutani 2005; Ambos and Schlegelm- We also know that tight bounds between a firm’s knowledge ilch 2006; Piscitello and Rabbiosi 2006; Narula and Michel base and its surrounding innovation system exist, including 2009; Rabbiosi 2009). institutions and universities (Patel and Pavitt 1999; Narula 2003). Reverse knowledge transfer can increase the overall tech- nological capacity, help to develop new products and foster growth and employment in the home country. R&D activi- 1 An example is the June 2010 issue of the Journal of Technology Transfer which discusses production offshoring and its effects on US manufacturing R&D in detail. 10These linkages make the offshoring of R&D difficult, and so far. This finding is also supported by empirical evidence most overseas R&D activity complements domestic activ- that points to the complementarities between domestic ity. Data on R&D expenditure or patent inventions give no and foreign production and sales activities, at least in the indication for a substitutive relationship between foreign- long term (Lipsey 2002; Barba Navaretti and Falzoni 2004). based and home-based R&D activities. This data suggests Here, a main argument is that the home economy benefits that countries which increased their overseas R&D activities in the long run because internationalisation creates a new saw also considerable gains in domestic R&D activities in division of labour within the firm, where home country units the past (Dachs et al. 2010). A ‘hollowing out’ (Criscuolo specialize in innovation, R&D and other headquarter activi- and Patel 2003) of domestic R&D has not been observed ties, which generate a greater value added to the economy. 113. ISSUES IN in fact, R&D expenditure accounts for around half of in- novation expenditure. The share is higher in countries with COLLECTING DATA ON THE high average R&D intensity, such as Finland, Austria, France or Sweden, where R&D expenditure accounts for more than INTERNATIONALISATION OF 60% of total innovation expenditure. R&D But not all R&D performed in a country is innovation activ- ity, because a considerable part of total R&D is performed by universities which do not introduce new products or pro- This chapter summarizes the results of the data collection cesses to the market. Moreover, some activities which are process. After presenting some basic definitions of R&D and not R&D may be innovation activity; examples are design innovation in the context of internationalisation of R&D a activities, staff training activities related to market intro - summary of the available data is provided and still exist- duction or production preparations. In a number of service ing data gaps and other important pitfalls that have to be industries, these activities comprise the bulk of innovation taken into account when analysing data on foreign-owned expenditure. R&D are discussed. The focus of this project is on business R&D. Following the OECD Frascati Manual (OECD 2002, p. 54), the business 3.1. Definitions of R&D and enterprise sector includes ‘all firms, organisations and in - stitutions whose primary activity is the market production innovation of goods or services for sale to the general public at an economically significant price’ and ‘the private non-profit For the analysis of the internationalisation of business R&D institutions mainly serving them’. This definition excludes a first important step is to clearly define business R&D and the government sector, the private non-profit sector and distinguish R&D from other aspects of the innovation. higher education, no matter if privately or publicly funded. Public enterprises are included if they are mainly engaged The OECD Frascati Manual defines R&D as ‘creative work in market production. undertaken on a systematic basis in order to increase the stock of knowledge, including knowledge of man, culture Data on R&D expenditure is usually collected separately for and society, and the use of this stock of knowledge to de- intramural R&D and extramural R&D. The OECD Frascati vise new applications’ (OECD 2002, p. 30). Compared to in- Manual (OECD 2002, p. 21) defines intramural R&D expen - novation, the term R&D rather refers to scientific discovery ditures as ‘all expenditures for R&D performed within a sta- and knowledge creation than to the economic application tistical unit or sector of the economy during a specific pe - of new knowledge. In practice, however, many firms may riod, whatever the source of funds’. In contrast, extramural find it difficult to distinguish between innovation and R&D R&D expenditures are defined as ‘the sums a unit, organi - activities. Both terms are overlapping: R&D financed and sation or sector reports having paid or committed them- performed by enterprises is always an innovation activity; selves to pay to another unit, organisation or sector for the Table 2: Main definitions Business Enterprise R&D Total business enterprise research and development (BERD) by Total BERD domestically owned firms and foreign-owned affiliates performed in the reporting country Inward BERD BERD by foreign-owned affiliates in the reporting country Domestic BERD BERD by domestically owned firms in the reporting country Outward BERD BERD of domestically owned firms outside of the reporting country Sector and industry classification Sector A NACE 1.1 section (mainly two-digit level) The data aggregated into high-, medium-high-, medium- low-, low- Industry technology manufacturing sectors, knowledge-intensive and less knowledge-intensive services 12performance of R&D during a specific period’ (OECD 2002, Most major non-OECD countries, including Brazil, China, p. 21). Generally, R&D expenditures include both, intramural India and Russia, follow the guidelines for the collection and extramural R&D expenditures and follow the definition of R&D data provided by the OECD Frascati Manual. The of R&D as stated above to distinguish them from other in- Frascati Manual, however, offers only little guidance for col - novative activities like design or staff training (OECD 2002). lecting data on R&D internationalisation. Hence, the availa- bility of data and the level of aggregation vary significantly This project frequently uses some key terms and abbrevia- across these countries (OECD 2010). tions for different types of R&D which are summarized in Table 2. An important distinction is between business R&D Despite increasing efforts to collect data and harmonize expenditure of domestically owned firms (domestic BERD) survey methodologies, some open methodological issues and business R&D expenditure of foreign-owned firms (in - remain that have to be kept in mind when interpreting the ward BERD) in a particular country. Outward BERD refers to data (Colecchia 2005; Colecchia 2006; Colecchia 2007; all business R&D expenditure of domestically owned firms Cozza 2010; OECD 2010): performed outside their home country. A first critical issue is the correct identification of the ‘ul - timate controller’ and the ultimate country of ownership. 3.2. Methodology, data sources Multi-level ownership structures of multinational corpora- tions make is sometimes difficult to identify the ultimate and challenges controller of a firm. The increasing interest in the internationalisation of eco- A second critical issue are accounting practices. R&D inter- nomic activity during the 1980s and 1990s brought forward nationalisation takes place to a considerable degree within various initiatives to collect data on the internationalisation large multinational firms with a presence in many coun - of R&D, to the largest part organized by the OECD (Godin tries. Thus, data on R&D expenditure of affiliates in different 2004). These efforts have been intensified after the year countries may be distorted by internal transfer pricing, non- 2000, pushed forward by the OECD Working Party of Na- priced transfers of R&D personnel and possibly problems tional Experts on Science and Technology Indicators (NESTI), in dividing flows of funds for R&D within the group across and the OECD Working Party on Innovation and Technol- borders. ogy Policy (TIP) (OECD 1998; Colecchia 2005; OECD 2005a; Colecchia 2006; Colecchia 2007; OECD 2008a). Another challenge for statistics on foreign-owned R&D is institutional separation. Responsibilities for collecting data Another impetus for the collection of data on the interna- on R&D expenditure and on the activities of foreign-owned tionalisation of R&D came from the European Union. Regu- firms (FDI or FATS surveys) are divided between statistical lation (EC) No 716/2007 requires EU member states to col- agencies and the central banks in some countries. This may lect data on intramural R&D expenditure by foreign-owned result in different samples for both surveys. The collection affiliates for every second year and, to a limited degree, of data on R&D expenditure of domestic firms abroad (out - on outward R&D expenditure starting with the year 2007. ward R&D) seems to be a challenge in particular. The methodology used is in line with the OECD Frascati manual. Foreign-owned affiliates are defined as ‘enterprises The problem of institutional separation and different sam - resident in the compiling country over which an institutional ples of foreign-owned firms multiplies when R&D inward unit not resident in the compiling country has control, or an and outward data is compared bilaterally. A study by OECD enterprise not resident in the compiling country over which NESTI (Colecchia 2006) compared outward BERD of the US an institutional unit resident in the compiling country has and some European countries with corresponding data for control’ (Regulation (EC) No 716/2007, Article 2 a)). The inward BERD in several European countries. The compari- concept of the ultimate controlling institutional unit (UCI) son showed considerable differences both in the number is used to determine foreign control of an enterprise. The of foreign-owned firms surveyed and in R&D expenditure UCI is the institutional unit in a chain of control which is not between the two data sources. controlled by another institutional unit (Eurostat 2007). In their Foreign Affiliates Statistics (FATS), EUROSTAT distin - With increased efforts put into the identification of foreign- guishes between inward FATS data, statistics describing the owned firms, additional issues arise for the analysis of this data activity of foreign-owned affiliates resident in the compiling over time. We have to be careful in the interpretation of time se- country, and outward FATS data, statistics describing the ries, because increasing levels of internationalisation for a given activity of foreign-owned affiliates abroad controlled by the country or industry over time may be (partly) rather the result compiling country (Eurostat 2007). Values for a reporting of increased efforts put into the collection of this data than of firm are allocated completely to the majority owner’s coun - increased levels of internationalisation. try; splitting of values according to owner shares or double counting is not recommended. Data on the R&D expenditure of foreign-owned firms in a detailed sectoral disaggregation should be interpreted with 13care as well. Data reported by the OECD (Colecchia 2006, owned firm for most European countries. In contrast, there p. 10) indicate that a handful of MNE subsidiaries account is considerably less data available for non-European coun- for a major share of foreign-owned R&D expenditures in tries: Only rudimentary data on inward BERD could be col- smaller countries. A change in the classification of only one lected for China and Israel. No data was available for South of these firms may result in shifts in the sectoral composi - Korea, Russia, India, and Brazil. This is a major shortcoming tion of inward BERD. and a serious obstacle to a global analysis of R&D interna- tionalisation, since emerging economies may rapidly gain- The statistical unit of R&D surveys is the enterprise and all ing importance in the process. R&D expenditure of the enterprise is assigned to the sec- tor of the main economic activity of the enterprise. Some In most countries, inward BERD data is collected by busi- countries, however, further split R&D expenditure if enter- ness sector R&D surveys. From our perspective, this is the prises are active in more than one product field, for example preferred organisational form, because it ensures the com- chemicals and pharmaceuticals. This may be a source of parability of the data with total BERD or sectoral BERD. inconsistencies in international comparisons of total BERD Separate R&D surveys for multinational firms (like in Israel) and inward BERD at the sectoral level. Another challenge should only be considered as a second strategy. related to the sectoral classification of foreign-owned firms is the classification of non-producing affiliates. If a foreign- Sectoral data is mostly available at NACE two-digit level. owned affiliate generates the majority of its value added by For a few selected industries (mainly pharmaceuticals and selling the products of its parent company, it is classified aeronautics), some statistical offices also provide data at as wholesale and retail trade, even if it belongs to a parent the NACE three-digit level. The availability and quality of company from the manufacturing sector. inward BERD data is better for manufacturing than for ser- vice industries. In our opinion, this is a second major short- Finally, another relevant issue is the treatment of non-R&D coming of the available data. Some countries exclude the performers. R&D surveys only report information on R&D service sector for some or all years (Denmark, Spain, Fin- active firms. If a regional headquarter of a multinational land, the Netherlands, Bulgaria, Hungary and Turkey), while firm sponsors R&D abroad, but does not have own domestic others only report data for broad service sector aggregates. R&D activities, it is unlikely that this R&D expenditure en- Moreover, there are considerable differences in the share of ters the survey results. service industries on inward BERD, and their development over time between countries which may also raise concerns about data quality. From our perspective, it is difficult to tell 3.3. Experiences from the data if these differences reflect a different economic structure or different survey designs. collection In many countries, inward BERD data is also available in a A main task of the project was the collection of data on split by the home country of the foreign-owned firm. Na - R&D expenditure of foreign-owned firms from national tional statistical offices tend to offer this data in more detail statistical offices, EUROSTAT and the OECD. We collected than OECD and EUROSTAT databases. Some smaller coun- data on inward BERD which captures R&D expenditure of tries do not fully publish inward data in a home country split foreign-owned firms in a particular host country, as well as due to data confidentiality when there are only a few R&D outward BERD data which includes R&D expenditure of do- active foreign-owned firms from a particular country. This mestic firms abroad. is a considerable obstacle for the analysis of cross-country patterns of R&D internationalisation. Inward BERD data Over the last two years there has been a noticeable in- is also provided in a home country x sector dimension by crease in attention of national statistical offices for R&D many EU countries. Confidentiality issues are even larger in internationalisation, in particular in EU member countries. this case, and even medium-sized countries like the Nether- As mentioned before, a major impetus for the extension lands have to omit data due to data confidentiality. of survey programmes is EU regulation on FATS statistics starting with (EG) No. 716/2007 which requires EU member In some countries large changes in the shares of different states to collect data on the R&D expenditure of foreign- home countries between two years can be observed. This owned firms every two years. Data from national statistical may be the result of the concentration of foreign activity offices is considerably more actual and much more detailed in a few large firms. It may, however, also be the result of in a number of countries than it was some years before. a better identification of the ‘ultimate controller’ and the However, this is not yet fully reflected in the OECD AFA and ultimate country of ownership. This is often hard to capture EUROSTAT databases. Thus, we can expect a better cover- because of the multi-level structure of many multinational age of these databases in the near future. companies. Data on the inward BERD is available at the aggregate and There is much less data on outward BERD than on inward the sectoral level and by the home country of the foreign- BERD. Detailed outward BERD data is available only for 14two countries, the US and Japan. One reason for this poor gregates at the EU-27 level prepared by EUROSTAT based coverage is the fact that collecting outward BERD data is on the national data would be extremely helpful. Such ag- more difficult than collecting inward BERD data. A firm-level gregates will provide valuable information on the role of survey of outward BERD addresses R&D performing firms the EU in the process of R&D internationalisation. It is not located abroad. Statistical offices may have very little in - possible for researchers to generate such aggregates by formation about this population, because they typically ad- summing up national data; though it seems very unlikely dress the firm population located within a country. that confidentiality is an issue at the EU level. Second, there are some serious data gaps in the service 3.4. The most pressing data needs sector. In some countries the service sector is not included at all. More complete data on inward BERD in services, in The experiences from the data collection allow us to iden- particular in knowledge-intensive services, would enhance tify four areas where improvements in data quality and our picture of R&D internationalisation where most atten- availability can considerably increase our knowledge of the tion still focuses on manufacturing. internationalisation of business R&D: Third, we see a considerable difference in data quality and First, we believe that less – not more - detailed data may availability between EU and non-EU member countries, in give a more complete picture of R&D internationalisation. particular emerging economies. Any measure to increase With an increasing level of detail of the data reported by the awareness for the topic in these countries may help EU member states over the last years, confidentiality issues to improve our understanding of R&D internationalisation. further increased and became a main constraint when ana- Some of these countries, in particular China and India, may lysing data on R&D of foreign-owned firms. Inward BERD already be major host locations for overseas R&D of Euro- from non-European countries (except the US), for example, pean firms – however, data to test this assumption is not is only poorly available in many small EU member states available. because there is usually only a handful of R&D active for- eign-owned firms from a particular home country and, as a Finally, the outward perspective remains poorly covered result, many results are confidential. Another example is in - and is not included in most reporting countries. Outward ward data in a country x sector split, which may very useful data can be substituted by the corresponding inward data for analytical purposes, but also leads to a large number of (mirror flows). This is feasible for some large countries like confidential values, even in medium sized reporting coun - Germany or France, which appear in nearly all inward BERD tries. Thus, we propose to publish (but not collect) data in data as home countries. But even in these cases big nu- larger aggregates to avoid confidentiality issues. merical differences between reported inward and outward BERD are visible. Statistical agencies can overcome this issue by providing data in higher country group and sectoral aggregations. Constructing outward data by the corresponding (mirrored) Rather than reporting inward BERD data for individual coun- inward data in a home country split may be helpful to see tries of origin (at the risk of suppressing data due to con- which countries have large R&D activities abroad and which fidentiality), the statistical agencies should publish country have not, or to observe shifts in the distribution of outward aggregates, specifying the countries included in these ag- BERD between the US, Europe and Asia. However, such an gregates. An aggregate for foreign-owned firms from Asian approach is only a second-best solution to outward data countries excluding Japan, for example may provide much from enterprise surveys. more valuable information than data for individual coun- tries where most information has to be suppressed because of the low number of firms. Moreover, several country ag - 154. CURRENT STATUS AND 4.1. Inward BERD across countries and over time FUTURE DIRECTIONS OF R&D INTERNATIONALISATION We measure the intensity of R&D internationalisation by overall inward R&D intensity. This indicator measures the ratio of inward BERD to total BERD (including foreign-owned This chapter investigates variations in the degree of R&D and domestically owned BERD). It thus shows the relative internationalisation across countries and sectors. The first importance of foreign-owned firms in different national in - section looks at differences in inward BERD across coun - novation systems. tries and over time to identify the countries which are most internationalised. Section 4.2 gives insights in the shares Figure 2 depicts overall inward R&D intensity for different of various home countries in total inward BERD of the EU countries. The figure reveals that the internationalisation of countries. In particular, we focus on the question in which R&D is increasing in the majority of countries. Only Hungary countries EU- or non-EU firms have the largest share on and the United Kingdom experienced a decrease in overall inward BERD. Whereas Section 4.3 proceeds with a cross- inward R&D intensity between 2003 and 2007. However, country analysis of the existing outward BERD, section 4.4 the internationalisation of R&D emerges only slowly, as in- investigates the sectoral perspective of business R&D ex- ward R&D intensities remained stagnant in a number of penditure. Section 4.5 analyses the relationship between countries, including large countries such as France, the US, the European Union and The United States of America in Japan or Germany. Huge changes between 2003 and 2007 R&D internationalisation. Section 4.6 closes this chapter are only observable in small countries. with a look at new players from emerging economies in the internationalisation of R&D. Figure 2: Overall inward R&D intensity in the business sector (inward BERD / total BERD, 2003 and 2007) 90% 2007 80% 2003 70% 60% 50% 40% 30% 20% 10% 0% Note: No 2003 data for Malta, Israel, Netherlands, Switzerland and Denmark; 2008 instead of 2007; 2006 instead of 2007; 2004 instead of 2003 Source: OECD, Eurostat, national statistical offices, own calculations 16 Overall inward R&D intensity Malta Ireland Belgium Israel Czech Republic Austria Sweden Slovakia Hungary United Kingdom Canada Italy Poland Slovenia Romania Estonia Germany Spain Norway Portugal Netherlands France Finland, United States of America Switzerland Denmark Japan Bulgaria Latvia