عوامل تعیین کننده انتخاب مکان فعالیت های تحقیق و توسعه در شرکت های چند ملیتی چیست؟

We analyse 446 location decisions of R&D activities by multinational firms incorporated in the European Union over 1999–2006. Our results suggest that on average, the location probability of a representative R&D foreign affiliate increased with agglomeration economies from foreign R&D activities, human capital, proximity to centres of research excellence and the research and innovation capacity of the region. Further, our evidence suggests that in comparison to European multinational firms, the effects of patents intensity and proximity to centres of research excellence were stronger in the case of North American multinational firms. While government R&D expenditure intensity increased the probability of location of R&D activities by European multinational firms in the region, it did not have a significant effect on the probability of location of R&D activities by North American multinational firms.

There has been a growing internationalisation of enterprise R&D activities over the last two decades. Multinational enterprises (MNEs) are the main drivers of this growing internationalisation of enterprise R&D and in many countries foreign affiliates carry out more R&D than domestic firms (OECD, 2007 and Abramovsky et al., 2008). While traditional cross-border R&D enterprise activities have tended to locate in developed economies, an increasing amount of R&D outward investment in recent years has gone to emerging economies (OECD, 2007, European Commission, 2008 and Sachwald, 2008). The type and motivations of R&D investment vary depending on whether R&D activities by multinationals were located in developed or emerging economies (Thursby and Thursby, 2006, Shimizutani and Todo, 2008 and Sachwald, 2008). In recent years, the speed and extent of the internationalisation of R&D have increased (von Zedtwitz and Gassmann, 2002, UNCTAD, 2005 and Abramovsky et al., 2008). This increased mobility of R&D and innovation activities has been linked to increased global competition, technological change, in particular the use of information and communication technologies (ICT) and the availability and costs of skills (Abramovsky et al., 2008 and OECD, 2008). In addition to the traditional role of R&D foreign investment in diffusing technology (demand-driven) related to adapting products and services to local market conditions and supporting MNEs local manufacturing operations, R&D foreign investment is being increasingly motivated by tapping into worldwide centres of knowledge (supply-driven) as part of firms strategies to source innovation globally (Wortmann, 1990, Hakanson and Nobel, 1993, Florida and Kenney, 1994, Florida, 1997, Patel and Vega, 1999, Le Bas and Sierra, 2002, Iwasa and Odagiri, 2004, von Zedtwitz and Gassmann, 2002, Ambos, 2005, Abramovsky et al., 2008 and OECD, 2008). Over the period 1995–2005, the share of foreign affiliates in total business R&D expenditure has increased substantially in almost all European Union's countries (European Commission, 2008). In 2005, this share was over 70% in Ireland, over 50% in Belgium and the Czech Republic, over 40% in Austria and Sweden. In contrast, the share of R&D expenditure by foreign affiliates was lower, less than 25% in Slovakia and Finland. The European Union (EU) is the largest recipient of R&D investment by US multinationals. In 2005, the EU accounted for 62.5% of the R&D expenditure of affiliates of US parent companies abroad. Abramovsky et al. (2008) show that in comparison to 1990, over the period 2000–2004, the average level of innovative activity of multinational firms from EU countries located abroad grew faster than their innovative activity conducted in the home country. This dynamics has lead to a growing share of the innovative activity located abroad in the total innovative activity of multinational firms. This increasing internationalisation of R&D activity in the EU raises a number of questions which are interesting and relevant for both research and policy making: Where are the R&D activities of multinational enterprises located? Who are the main foreign investors in R&D activity? What factors drive the location choice of multinational R&D activity? To answer these questions, we analyse the determinants of the location choice of R&D activities by multinational firms across regions in the European Union. By considering regions as location choices we account for heterogeneity of locations within countries and avoid the aggregation bias which might arise when using country averages in cross-country analyses. We use a large firm-level data set1 which enables us to consider a wide range of location choices of R&D activities by multinational firms. Specifically, we analyse the location choice of R&D activities of 446 new foreign affiliates incorporated in the European Union over the 1999–2006 period. The large number of location choices (233 regions) enables us to obtain robust estimates of determinants of the attractiveness of regions to foreign investment in R&D activity.2 The traditional theory of multinational firms has modelled the location decision of multinational firms assuming that R&D activity is located where production takes place and it has not addressed specifically the case of the location choice of R&D activities by multinational firms.3 Notable exceptions are Markusen (2002) and Ekholm and Hakkala (2007). These latter theoretical contributions allow the geographical separation of knowledge-based (R&D) activities and production facilities in a two-country general equilibrium setup. The theoretical model proposed by Markusen (2002) known as the “knowledge capital model” of multinationals firms predicts that when trade costs are low, international production is likely to locate in large economies while knowledge-intensive activities will concentrate in small skills-intensive economies. The model developed by Ekholm and Hakkala (2007) allows agglomeration forces to arise in both production and R&D activities and predicts that international production will locate in a larger economy while R&D activities by multinationals will locate in a smaller economy to benefit from R&D spillovers. In contrast to the slow development of the theoretical literature on the location choice of R&D activities by multinational firms, a growing number of empirical studies have analysed the internationalisation of R&D and the development of R&D global networks (Florida and Kenney, 1994, Patel and Vega, 1999, Frost, 2001, Ambos, 2005, Abramovsky et al., 2008 and Sachwald, 2008). Given that multinational enterprises are concentrated in R&D-intensive industries, many factors driving the location choice of foreign affiliates are also relevant and important in the case of R&D activities of multinationals. However, as documented in a number of recent studies in international business, in addition to demand-side factors, such as market access, factors specific to the R&D sector such as knowledge-sourcing have become increasingly important as a motivation for establishing R&D units abroad (Florida, 1997, Patel and Vega, 1999, Frost, 2001, Le Bas and Sierra, 2002, von Zedtwitz and Gassmann, 2002, Ambos, 2005, Ito and Wakasugi, 2007 and Belderbos et al., 2008). Most existing studies analyse determinants of the location choice of foreign R&D in a single country setup. Cantwell and Iammarino (2000) analyse the location patterns of multinational networks for innovation in the UK regions. Frost (2001) examines the origin of external sources of innovation of US greenfield subsidiaries. Ito and Wakasugi (2007) and Shimizutani and Todo (2008) investigate determinants of Japanese R&D investments abroad and Iwasa and Odagiri (2004) analyse determinants of Japanese R&D investment in the US. Ambos (2005) analyses motivations of German-owned multinational enterprises with international R&D activities. This paper builds on and extends these two strands of literature, namely the existing theoretical and empirical literatures, on international trade and investment on one hand, and on the internationalisation of R&D and global R&D networks on the other hand. We add to the empirical literature on the location choice of multinational enterprises in three ways. First, in contrast with most existing empirical studies mentioned above which consider both demand-driven (market access) and supply-driven (knowledge sourcing) motivations for foreign direct investment in R&D in a single country setup, we estimate location choice models in a multi-country setup. Second, in contrast to existing cross-country analyses, we account for heterogeneity of locations within countries and avoid aggregation bias in the estimates of the location choice determinants. Third, in contrast to previous studies, we use an improved econometric methodology to account for spatial correlation among location alternatives and firms due to unobserved location-specific characteristics. Our results suggest that on average, the probability to locate in an EU region increased with agglomeration economies from foreign R&D activities, human capital, proximity to centres of research excellence and the research and innovation capacity of regions. There is also evidence of a geographical structure in the location choice of R&D multinational firms across the European Union. Further, our evidence suggests that in comparison to European multinational firms, the effects of patents intensity and proximity to centres of research excellence were stronger in the case of North American multinational firms. While government R&D expenditure intensity increased the probability of location of R&D activities by European multinational firms it had no significant effect on the location of R&D activities by North American multinational firms. The remainder of this paper is organised as follows. Section 2 describes the empirical methodology and testable hypotheses. Section 3 presents our data and summary statistics. The results of our econometric analysis are presented in Section 4. Finally Section 5 summarises our results and concludes.

In this paper we estimated the determinants of the location choice of new R&D activities by multinational firms across regions in the European Union over the period 1999–2006. With respect to methodology improvements, in addition to conditional logit models we estimated nested logit models to account for the fact that in relation to many alternative location choices, conditional logit models might lead to biased estimates if the location choices are not independent. In contrast to most existing cross-country analyses, we consider regions within countries as location choices and thus avoid aggregation bias in the estimates of the location choice determinants. We find that in our spatial choice framework the independence of the location choices does not hold. Consequently we base our conclusions on the estimates of determinants of the location choice of R&D activities by multinationals obtained with the nested logit models. We find evidence of a geographical structure in firm's location choice across the European Union. Our results suggest that on average, the probability of the location of a representative R&D foreign affiliate in an EU region increases with agglomeration economies from foreign R&D activities and the region's knowledge base measured by human capital, proximity to centres of research excellence, research and innovation capacity. It appears that, over the analysed period, regional characteristics such as market potential, compensation per employee and unemployment rate, had no significant effect on the attractiveness of regions to R&D foreign investment over and above other determinants. Some of these results might be influenced by the fact that our data does not distinguish between research and development activities. Our evidence also suggests that country level corporate tax rates had no significant effect in fostering the attractiveness of regions to R&D foreign investment over and above other determinants. This result might be explained by the fact that the sensitivity of the location probability to taxation in a country/region is higher in the case of a small number of location options (Barrios et al., 2008). Also, multinationals locate foreign affiliates in more than one country and they optimise the tax on a global base. A large number of European countries have introduced R&D tax incentives to foster investment in R&D activities. Notable exceptions are Germany, Sweden, Finland, Luxembourg and Estonia. R&D tax incentives can foster as well as discourage international investment. Furthermore, the benefits to firms of R&D tax incentives are strongly linked to the overall corporate tax rates. While it is desirable to test the effects of these R&D tax incentives on the location choice of foreign R&D activities, for the purpose of this paper, data on R&D tax subsidies over 1990–2008 are available for only 12 of the 21 countries analysed in this paper. The determinants of the location choice of R&D foreign affiliates vary depending on the country of origin of the foreign investor. The region's innovation capacity measured by patents intensity and business R&D expenditure intensity as well as the presence of centres of academic research excellence were important for the location choice of R&D activities by European as well as North American multinationals. In comparison to European multinational firms, the effects of patents intensity and proximity to centres of research excellence were stronger in the case of North American multinational firms. These results suggest that in comparison to R&D activities by European multinationals, the location of R&D activities of North American multinationals in EU regions were more likely to aim at knowledge-sourcing. While government R&D expenditure intensity mattered for the location choice of R&D activities by European multinationals, it did not play a significant role in the case of the location decisions for R&D activities by North American multinationals. Our research results suggest a number of policy implications. First, policy aiming to increase the knowledge base of regions is likely to foster the attractiveness of regions to R&D foreign investment. Second, positive externalities from clustering of R&D foreign affiliates outweigh competition effects. Third, given the heterogeneous behaviour of foreign investors, differentiated policy depending on target partner countries can increase the success of such policies.