جهانی شدن R & D توسط شرکت های چند ملیتی مستقر در ایالات متحده

This paper examines patterns and determinants of overseas R&D investment by US-based manufacturing MNEs using a new panel dataset over the period 1990–2004. The analysis reveals that R&D intensity of operation of US MNE affiliates is determined mainly by the domestic market size, overall R&D capability and cost of hiring R&D personnel. There is no evidence to suggest that R&D specific incentives have a significant impact on inter-country differences in R&D intensity when controlled for other relevant variables. Overall, our findings cast doubts on the efficacy of efforts by host country governments to entice MNE affiliates to engage in domestic R&D activity, in a context where R&D is becoming a truly global activity.

Multinational enterprises (MNEs) play a pivotal role in the generation of technology and its transmission across countries.1 The potential contribution of MNE affiliates to innovatory capability of the countries in which they operate (the host countries) is therefore central to the contemporary policy debate on the developmental impact of foreign direct investment (FDI). There are two ways in which MNE affiliates provide technology to host countries; importing technology produced elsewhere within the global branch networks (technology transmission) and developing new technology locally (technology generation). The host-country governments generally attach greater importance to technology generation over technology transmission, in the hope that research and development (R&D) activities undertaken within the national boundaries may have important externalities for local scientific and technological capabilities. This expectation has resulted in a strong competition among countries to attract R&D-intensive FDI.2 The purpose of this paper is to complement and extend the existing literature on the determinants of the global spread of R&D activities of MNEs using a new panel data set relating to the operation of US-based MNEs during 1990–2004, a period characterized by significant changes in international production as part of the on-going process of economic globalization. The key research issue is the relative importance of policy-related variables in explaining inter-country differences in R&D intensity over and above the relevant non-policy (structural) variables. Compared to the previous studies on this subject,3 we examine inter-country variation in R&D intensity by taking into account a larger number of explanatory variables suggested by the theory of MNE behaviour, with a view to minimizing potential omitted variable bias in estimation. To the best of our knowledge, ours is the first attempt to examine patterns and determinants of overseas R&D activity using panel-data econometrics. The panel data approach offers a solution to the problem of bias caused by unobserved heterogeneity (in this case, country-specific peculiarities not captured by the explanatory variables), a common problem in the estimation of models with cross-section data as in the previous studies. Working with panel data also has the advantage of capturing dynamics that are difficult to detect with cross-section data. Another novelty of our analysis is the attention paid to the impact of the stage of development of host countries on the hypothesized relationship between the R&D intensity and the explanatory variables. There is no fully developed theory of inter-country differences in R&D intensity of MNE operation. As in previous studies, we therefore formulate our empirical model in an eclectic fashion, drawing upon the analytical foundations of MNE behaviour. We strongly believe that our approach is preferable to working with an optimizing model derived from first principles assuming a ‘representative’ firm. This approach, notwithstanding its analytical elegance, cannot adequately address issues that arise from imperfect information and heterogeneity relating to industry characteristics and government policies (Kirman, 1992, Dunning, 2000 and Vernon, 2000). The findings suggest that R&D intensity of operation of US MNE affiliates is determined mainly by domestic market size, overall R&D capability and cost of hiring R&D personnel. Domestic market orientation of production is found to be a significant positive determinant of R&D only in low-income countries, presumably reflecting the need for product adaptation to suit special demand conditions associated with low-income levels and the lower degree of global integration of these countries. It seems that, in an era of rapid global economic integration, the nature of market orientation is not a significant determinant of R&D patterns in advanced industrialized nations and newly industrialized countries. We also find that industry composition is an important determinant of the overall R&D intensity of MNE operation in a given country over and above the other variables considered here. There is no evidence to suggest that R&D-related tax incentives and intellectual property protection have a significant impact on inter-country differences in R&D intensity when appropriately controlled for the basic location factors. The paper is organized as follows. Section 2 provides a succinct review of the theory of overseas R&D activities of MNEs in order to set the stage for the ensuing empirical analysis. Section 3 examines trends and patterns of overseas dispersion of R&D expenditure of US MNEs. Section 4 deals with model specification, data sources and the econometric methodology used in the regression analysis of the determinants of inter-country differences in R&D propensity. Section 5 presents the results and interprets them in the context of the existing literature. The final section summarises the key inferences.

We have examined patterns and determinants of overseas R&D activity by MNEs using a new panel dataset relating to US-based MNEs over the period 1990–2004. It is found that domestic market size, geographic distance, overall R&D capability of the country and cost of R&D personnel are key determinants of the R&D intensity of operation of US MNE affiliates. There is also evidence that, contrary to the conventional wisdom, the impact of domestic market orientation of affiliates on R&D propensity varies among countries depending on their stage of development. The degree of domestic market orientation has a positive impact on R&D intensity only in developing countries other than the East Asian NICs. For the latter countries and developed countries, the two variables are negatively related, suggesting that greater export-orientation is associated with greater (not less) R&D intensity. There is also evidence that, once controlled for the other relevant variables, the industry composition does matter in explaining inter-country variations in R&D intensity. R&D related tax incentives do not seem important in explaining inter-country differences in R&D intensity when appropriately controlled for other relevant variables. Intellectual property protection seems to matter only for mature economies with complementary endowments. Overall, our findings cast doubts on the ability of host governments to entice MNEs affiliates become technology creators within their national borders as part of their FDI policy. Moreover, our results do not support the case for supporting MNEs engaged in domestic-market oriented production on grounds that they promote local R&D activities. In a context where R&D is becoming a truly global activity, MNEs’ decision to undertake R&D activities in a given country seems largely endogenous to its overall growth and development process. Excessive concern about whether R&D is performed locally or elsewhere within the MNE network runs the risk of downplaying the more important role of MNEs as a conduit of technology transfer. Even if MNE affiliates generate little or no technology locally, they are potentially well placed to play an important role in improving local innovative capabilities through technology transfer. In our examination of the determinants of R&D intensity, we have been able to bring to bear considerably richer data than have been used in prior research. However, the results need to be qualified for two major limitations of the data set. First, a two-dimensional panel dataset (arranged by country and time), while being a significant improvement over pure cross country data set, still fails to capture industry specificity of R&D intensity. Unlike in previous studies, we have controlled for industry specificity by including a R&D potential index of output mix, but this is admittedly a crude way of tackling a more complex issue. Second, we have used a perception-based index to measure R&D tax incentives, which probably does not consistently measure inter-country differences. There is certainly a need to check the robustness of the results using a statutory measure of tax incentives. Finally, it is important to be cautious when generalizing from our findings, which are specific to overseas operations of US-based MNEs. In particular, given the large domestic economy and the R&D resource base, US-MNEs may have a lesser tendency to internationalize R&D in aggregate or in specific industries overseas compared to MNEs based in a small country like Sweden.