A critical appraisal of global innovation management literature: Is there really a "trend towards a global innovation network"?

1

1. Some Allegations in Literature With Respect to R&D Internationalization

It is purported that there is "a trend towards the international R&D network". According to this view, R&D internationalization started with the search for additional markets, then tapped cheap labor for production, localized existing products and finally ended with a fully deve l- oped R&D function [1]. This approach is an applic ation of concepts developed before [2,3]. It is claimed that there is a general trend towards more cooperation and more decentralization of R&D. Given this trend, "five types" of R&D organization are claime d to be emerging. The company undergoes a linear development process across these types that ends with the emer- gence of a global R&D network [4]. This development process has mainly been explained by intra-firm optimization of global cost and transaction cost functions such as to minimize total organizational cost. This can even imply positive external effects by affiliates mutually learn- ing from one another, thus increasing the firm's competitiveness [5]. This view is consistent with models based on transaction cost arguments [6] and with the process school's develo p- ment model towards the "transnational" firm [7].

R&D internationalization motives other than cost were mainly summarized under the head- lines demand, supply and external factors [8,9,10,11,12] and the categories local markets, resources and synergy potentials [13]. Indeed there has been evidence that market size [14] is an important location decision factor and that foreign R&D investment is correlated to general host country advantages [15]. In Germany, the share of external R&D expenditures spent off- shore almost doubled in favor of expenditures within the industrial sector [16]. This "trend towards R&D internationalization" seems to contradict classical findings [17] that advocate inventors be granted local market monopolies by patents, logically implying R&D to be most centralized to maximally exploit economies of scale and to secure vital knowledge from unco- trolled outflow. Historically, this was indeed the case, the former German chemicals con- glomerate IG Farben being an example for a Schumpeterian monopolist. Some have tried to explain this paradox by resorting to a "trend that R&D had to adjust to world-wide dispersion of knowledge and technology creation", with i ncreasing competitive pressure for new and innovative products and the fusion of formerly unrelated technologies reinforcing this trend [18]. Discussion on this paradox has been ample and has often drawn on arguments known from general internationalization theory [19,20,21].

Another reason for this paradox is seen in the fact that increased competition from within and outside their industries has forced companies to globally source technological knowledge and human resources (i.e., researchers that can generate knowledge valuable to the firm). How-

2

ever, there is also a spatial dimension of this problem, because it is purported that the relevant knowledge needed is not available everywhere, but rather globally dispersed in certain re- gions or clusters ("centres of excellence"), which are characterized by a high rate of new technology output. Thus , the "hunt for the best heads" would lead to the establishment of local research and development, mainly in order to absorb critical competencies from centres of excellence and to support organizational learning on a global scale [22,23]. This standpoint basically views R&D internationalization as a sourcing problem, thus returning to competitive advantage theory [24], with technological advantage in the form of superiority of knowledge to form the essential asset basis of MNCs' competitiveness in world markets [25]. However, the notion of R&D internationalization to be a linear development process or a "goal" in a design process to be reached ultimately is questionable given the latest develo p- ments i n global economic order. In addition, empirical evidence related to the above- mentioned arguments has been debatable . Therefore, the remainder of this paper is organized as follows:

Section 2 will critically review the allegations introduced above, presenting empirical evi- dence that does not go together with existing R&D internationalization theory. Section 3 will take into account the emergence of China to show further phenomena hardly recognized by existing theory. Section 4 will introduce concepts known from other disciplines that can help to develop new theory that more aptly responds to real -world developments.

2. Critical appraisal: Methodological, Historical, Size and Technology Effects

The arguments mentioned above effectively trace back to the "centralisation versus decen- tralisation" and "local versus global" discussions which assume that a company "chooses" its "optimal" international R&D setup, and "decides" on the degree of (de-)centralization and globalisation. These allegations have been reviewed in the context of international R&D or- ganization [26]. They have been discussed amply and controversially [27,28,29,30,31,32,33,34,35,36,37,38]. The case is not that clear, however, as empirical as well as theoretical phenomena are show ing.

Firstly, there is ample evidence available as to the methodological flaws of many studies ad- vocating certain organizational designs. Most of these have dealt with non-randomly selected small samples of first-world MNCs within the Triad only, providing rather subjective case studies and anecdotes, not taking into consideration macroeconomic and international trade relations effects. Some have not pr ovided any data at all. Extensive review and proof for this fact is available [39,40,41,42,43,44,45,46]. In addition, they made little effort as to provide

3

valid inferential conclusions that could be drawn from an adequate sampling design taking into consideration size, industry and technology effects. Studies that did so could show that there are cases where "non-globalisation" is efficient [47,48].

Secondly, the neglect of the historical changes in global economic regulation has led to error- prone assumptions. It has been shown [49] that the degree of firms' internationalization was by no means lesser in 1914 than it was in 1980, with plenty of multinational firms existing in Europe and in the US. There was even an equivalent to an international currency (the gold standard) in Europe till 1918. Although research amongst these firms was centralized, deve l- opment and sales had a high degree of internationalization. These links were destroyed by the world wars, leading to the set-up of national R&D bases again. Reacting to US pressure in 1946 trying to prevent its still booming war economy from a hard landing, the GATT as the basis for the WTO to follow was created in 1947, granting American companies favourable trade agreements, making it possible for them to export into markets abroad exempt from tar- iff barriers. As soon as they had gained market share, development activities were interna- tionalized first in order to substitute exports by products adapted to local markets, however, due to investment barriers, the main market entry was by joint venture. Still, the Bretton Woods order grounded on the US dollar's gold coverage provided a hindrance to free interna- tional investment. This is why large-scale M&A activities - the premier mode by which R&D internationalization took place - set off only after the system was aba ndoned in 1973. Only with the post-Bretton Woods system of international finance, together with world-wide FDI policy liberalisation, it was possible for Triad firms to quickly take over existing labs and re- sources instead of having to resort to greenfield investments. For pharmaceutical firms - among the first that set up international R&D facilities - this provided a convenient way to exploit R&D economies of scale and investment efficiency. Indeed, there is plenty of evi- dence that R&D organization patterns were hardly planned [50,51,52,53], but represent rather organic ally grown structures that evolved over time, reacting to relative economic order and regulation [54,55,56]. The fact that global deregulation led to increasing externalization of R&D has been empirically validated, discovering that the "global networks" following M&A activities were largely the consequence of international restructuring and not due to some ide- alized global network design process [57,58]. Interestingly, in some cases i nternational de- regulation has effectively led to "non-globalisation" in certain industries [59,60].

Thirdly, it need not be that R and D are internationalized according to some process that puts research as the function to follow development. Instead, the two can be separated spatially and timely. This becomes clear when comparing US and Japanese firms' R&D internationali-