Introduction: the regionalization hypothesis

We can summarize as follows the essence of our regionalization hypothesis: most international business action is characterized by semi-globalization. Semi-globalization is situated between the theoretical extremes of: (a) firms expanding across the world as if it were an integrated marketplace, not in terms of homogeneity of location advantages, but in terms of unlimited MNE access to these advantages. Here, a strategy of full integration yields large benefits of scale, scope and exploitation of national differences; (b) firms focusing primarily on their home nation and operating quasi-autonomously in each individual host country, whereby national responsiveness is the sole strategy required.

The empirical evidence for regionalization suggests two points. First, much of the international activity of multinational enterprises (MNEs) appears to be conducted at the intra-regional rather than inter-regional level. The triad of regions we consider most relevant consists of North America, Europe and Asia, because these regions are both the home of the vast majority of large MNEs in the world and the geographic locus of most business innovations. In addition, these regions are economically very different, geographically distant from each other, and characterized by internal attempts to achieve greater regional cohesion (at present most advanced in Europe and least in Asia).Footnote 1 Second, many MNE international operations (to the extent that these span more than a single region) are organized at the regional level rather than the global level.

The first point implies that the scope and pervasiveness of the globalization phenomenon have been vastly exaggerated, in spite of numerous academic publications on topics such as ‘global investment’, ‘global sales’, ‘born global companies’, etc. Most MNEs are simply not capable of deploying and exploiting their firm-specific advantages (FSAs) around the world in an indiscriminate fashion. Even though most large firms would like their products and services to conquer the world and to earn high market shares around the globe, the unfortunate reality of international business is that only a handful of firms (nine firms in the world's top 500, to be precise, namely IBM, Sony, Philips, Nokia, Intel, Canon, Coca-Cola, Flextronics, and Moët Hennessy–Louis Vuitton) have been able to achieve a balanced distribution of their sales in North America, Europe and Asia. For most other firms, selectivity in national and international expansion has clearly remained critical. Much of that selectivity has been guided by regional considerations such as the preference of European firms to grow their international operations mainly in Europe; of North American firms to give priority to NAFTA-based expansion; and of Asian firms to explore business opportunities in Asia first before venturing into other parts of the world. The main reason is that the liability of intra-regional expansion appears to be much lower than the liability of inter-regional expansion: the additional costs of doing business abroad are often much higher when venturing into other regions of the world than when expanding intra-regionally, in the home triad region. In the latter case, location-bound FSAs can often easily be augmented so as to become deployable in the entire home region. In addition, the firm's non-location-bound FSAs can easily be fine-tuned so as to optimize their exploitation potential in the home region.

The second point suggests that it is difficult to manage an internal network spanning more than a single region: the ‘distance’ separating North America, Europe, and Asia remains substantial. This distance has a mix of cultural, administrative, geographic and economic roots with important implications for the internal functioning of the MNE network (Ghemawat, 2001, 2005). Corporate-level management faces enormous bounded rationality constraints when managing operations in host triad regions: see Rugman and Verbeke (2004b, 2005a, especially Chapter 9). Therefore, many of the world's largest MNEs exhibit a regional component in their operations, such as regional headquarters or regional operating divisions.

In theory terms, building upon Rugman and Verbeke (1992a, 1992b, 2001, 2004a, 2005a, 2005b and 2005c), many of these MNEs’ FSAs – although non-location bound in principle, that is, deployable and exploitable beyond home country borders – appear to be home region bound rather than having global deployment and exploitation potential. This holds especially for FSAs at the downstream side of the value chain, where products and services need to be marketed to buyers. In the case of markets seeking foreign direct investment (FDI), an MNE's resource allocation to the host market can be interpreted as a set of one-sided commitments without equivalent commitments from potential purchasers, and therefore carrying substantial risks, especially in cases where a high liability of inter-regional foreignness is present. In contrast, other types of FDI, focused on the upstream side of the value chain, especially sourcing and manufacturing, are often accompanied by resource commitments from other economic actors, thereby reducing the challenges posed by the liability of inter-regional foreignness, at least if these other actors are embedded in the relevant host region. Especially if inter-regional FDI in sourcing or manufacturing is meant primarily to facilitate home region sales, and is thus efficiency seeking (as is the case with much North American and European investment in China and India), the liability of inter-regional foreignness mostly does not constitute a major risk, but such investment will then not necessarily improve the MNE's position in terms of sales in the host region either.

Given the above theoretical points, we now turn to Dunning et al.'s (2007) contribution to the regionalization vs globalization debate, and especially to these authors' insights that may be useful for international corporate strategy purposes and for improving our understanding of the geo-political reality of world business. Rather than dwelling on differences in statistical methodology, we shall focus on three key conceptual points.

The interpretation of firm- and country-level data

Rugman and Verbeke's (2004a) contribution is that the firm-level data for the world's largest 500 firms (most of which are MNEs) show a strong regional dimension in these firms’ activities. As suggested above, these firm-level data refute any simplistic notion of global integration and its alleged benefits. The country-level data by Dunning et al. broadly support the observation that regions matter, but with somewhat less clarity. Figure 1 explains the advantages of our approach vis-à-vis Dunning et al.'s analysis.

Figure 1
figure 1

Firm- and country-level data. HOME, sales in the home country; ROR, foreign sales in the rest of the home region; ROW, foreign sales in the rest of the world; F/T, value of foreign (F) to total (T) sales, where F=S+X, where S=sales by foreign subsidiaries and X is exports by the firm from the home country (as recorded in annual reports of the firms); R/T, intra-regional sales (HOME plus ROR).

Full size image

In Figure 1, we make a distinction between the ratio of foreign (F) to total (T) sales, F/T, and the ratio of regional (R) to total sales, R/T. The F/T ratio is the traditional metric for internationalization. It has been used in hundreds of empirical studies, especially in those testing the (elusive) relationship between multinationality and performance (Geringer et al., 1989; Sullivan, 1994; Contractor et al., 2003; Lu and Beamish, 2004; Hennart, 2007). Essentially, most of John Dunning's unparalleled academic oeuvre has consisted of describing and explaining the observed F/T data, but with a specific focus on country level FDI data rather than sales data. Here again, Dunning et al. provide macro (country-level) information on the FDI equivalents of our F/T sales data.

In contrast to the Dunning et al. focus on F/T data at the macro level, we introduced micro (firm-level) data on R/T in Rugman and Verbeke (2004a). We also developed theoretical arguments to demonstrate that, from the viewpoint of international corporate strategy, the R/T ratio may be particularly relevant, in both its micro or macro variants. This is because the R/T measure includes the home sales of the firm (which are often critically important from a corporate strategy perspective, as they reflect the nature of the MNE's FSAs and the related past mix of national and international expansion opportunities), as well as the proportion of foreign sales in the home region (ROR, for rest of region). The remaining foreign sales are those in the rest of the world (ROW).

For illustrative purposes, we have included in Figure 1 the actual percentages for the variables above, derived from a recent study by Yip et al. (2007). This study identified the home, regional and international sales of the 27 UK MNEs in the world's top 500 firms for which these data are available for 2003. In Figure 1, the average home country sales of these UK MNEs is 52%; the F/T is 48%. In contrast, the R/T is 64%. This implies an average of 12% of the F/T is in the rest of the European region (ROR). Finally, the foreign sales of UK MNEs to the rest of the world (ROW) represent 34%. In short, while the macro data by Dunning et al. bring to the table the traditional F/T data used in many international business studies (as proxied by FDI country equivalents), Figure 1 with its focus on regional sales paints a different picture. Dunning et al.'s data tell the simple story that international activities do matter, and should receive adequate attention in strategic decision-making. Our approach suggests that the home region, which represents almost two-thirds of the firms’ sales should be front and center in any strategic decision-making process.

In Rugman and Verbeke (2004a), we suggested that a focus on R/T is critically important to international corporate strategy, as a complement to F/T. International corporate strategy is not based solely on the straightforward, globally dispersed exploitation of FSAs in the MNE's foreign subsidiaries. Instead, as first analyzed by Buckley and Casson (1976), Rugman (1981), and Dunning (1981) himself, many FSAs are normally first developed in the MNE's home market, and even in the case of non-location-bound FSAs the MNE will still incur additional costs of doing business abroad. Here, it is precisely the relative familiarity with – and the experience in – the home country and the home region vis-à-vis other host markets that explains the liability of inter-regional foreignness.

While more recent literature suggests that some FSAs may be developed in foreign subsidiaries, these are often location bound, meaning that they cannot be readily deployed and exploited beyond the host country's borders or a perhaps somewhat broader region, unless the MNE has developed complex internal knowledge augmenting and diffusion systems (Birkinshaw, 2000; Rugman and Verbeke, 2001, 2003). Strategically speaking, a substantial portion of the MNE's FSAs will still be developed in the home base, and the empirical evidence of the MNE's sales data suggests that these FSAs, in order to be exploited successfully in host regions, require substantial additional investments, typically leading to location-bound FSAs, deployable and exploitable in the relevant host region only.

The above conceptual points cannot be derived from the macro (country-level) data of Dunning et al. Such macro data do not allow unbundling of the complex sets of FSAs present at the level of individual companies. Moreover, the macro data could unfortunately also be wrongly interpreted as providing support for globalization strategies, whereas the firms involved tell us they operate regionally. The firm-level data, providing evidence of regional (rather than global) business activity, represent the most relevant metric today.

Problems of artificial cultural clusters and ‘regional’ data adjustment

The problem with macro data is that such data cannot reflect adequately the selection processes that individual MNEs engage in when choosing their path of international expansion and their organizational structure and governance mechanisms. This problem is compounded in Dunning et al.'s work because of their use of statistical artifacts that alter the macro data sets used. One such artifact is the index of revealed investment comparative advantage (RICA). The RICA index measures the extent to which a country, relative to its share of the world's inward and outward FDI, commands a higher or lower share of that FDI in six so-called ‘regional’ clusters. Unfortunately, the results (shown in various tables with adjusted macro data and in a few diagrams) are somewhat misleading as they are derived from a hypothetical set of regional clusters, based on cultural archetypes: see for example Ronen and Shenkar (1985). Such archetypes do have some value, especially in work on international human resources management and organizational behavior studies, but they are less appropriate for studies that attempt to understand the essence of international corporate strategy and geo-political reality. In contrast, the broad triad regions defined by Rugman and Verbeke (2004a) are based on the revealed empirical evidence of international business, and the observation of a mix of cultural, administrative, geographic and economic distance among triad regions. The empirical evidence is that MNEs operate on the basis of these triad regions, and not across the six artificial cultural regions used by Dunning et al. to calculate RICA.

The six clusters recognized by Dunning et al. are: an Anglo cluster, a Germanic/Nordic cluster, a Latin American cluster, a Latin European cluster, a Far Eastern cluster, and a ‘mixed’ cluster, which includes all other countries. As a result, Australia is now in the Anglo cluster, even if, at the macro level, Australia competes with rivals in Asia as a location for regional head offices for inward FDI, while much of its outward FDI in resource-based industries goes to the rest of Asia. Similarly, South Africa is also placed into the Anglo cluster, even though MNEs such as Anglogold have relocated their head offices to London (UK) and are related more to the European Union than to a hypothetical Anglo cluster. But, paradoxically, perhaps the worst fit for the Anglo cluster is the UK itself, as Dunning et al.'s own data indicate that UK FDI is now highly intra-regional, more so than that of many other European countries in the so-called Nordic and Germanic cluster. As another example, Dunning et al. position Mexico in the Latin American cluster, but in reality it is in NAFTA, with over 80% of its trade and FDI there, not in Latin America.

It is clear that the cultural clusters adopted are merely an academic artifact, intellectually appealing but relatively far removed from the practice of international corporate strategy and geo-political reality. A careful reading of the world's top 500 firms’ yearly reports demonstrates that a large number of firms with substantial international activities do operate mainly in terms of the triad regions of North America, Europe and Asia. We have yet to identify a single firm operating along the clusters used by Dunning et al.

The use of statistical artifacts also produces rather implausible results at the macro level, where most of Dunning et al.'s contribution is situated. For example, the classification of countries in Figure 1, although correct given the variables chosen, appears somewhat inconsistent with other sources of UN data and certainly at odds with geo-political reality. Here, China appears in the lower-left quadrant with a low outward and inward transnationality index. Yet the UN has also reported that China has been the largest inward recipient of FDI flows over the last 10 years. It is difficult to understand why Chinese inward FDI would be positioned in the same quadrant as Poland and Italy, or what strategy or policy implications can be derived from such positioning. Rather, the use of more simple and straightforward variables to measure its inward FDI position would make it appear in the lower-right quadrant. Similarly, the US has one of the world's largest outward FDI flows, and it should therefore appear in the upper-right quadrant, not the lower-left one, as should Japan. It is of course because the relevant data are expressed relative to parameters such as GDP and gross fixed capital formation that an intuitively implausible and ultimately unhelpful positioning of countries is obtained.

As another example, the adoption of so-called adjusted globalization indices (AGIs) alters the actual data sets on inward and outward FDI stocks. Those stocks are now ‘adjusted’ for the size of the home regions receiving or generating them. The problem is that once FDI figures are adjusted (and this holds for any other parameter), it becomes difficult to understand what the adjusted data really mean. In contrast, the regional sales data in Rugman and Verbeke (2004a) have the virtue of being the real data used by managers in international corporate strategy, and accurately reflect geo-political reality.

How home region bound are FSAs?

In their conclusions Dunning et al. pose the important question whether lower investment in distant regions should be interpreted as the result of reduced deployability and exploitability of the MNE's FSAs or as the outcome of increased spatial or transaction costs. We have provided the answer to this question in our earlier work (see especially Rugman and Verbeke, 2004a, 2004b, 2005a, 2005b and 2005c): much of the international business literature that adopts an internalization theory/eclectic paradigm perspective assumes that FSAs are non-location bound, and can be exploited in host countries subject to additional costs of doing business abroad. In reality, many FSAs, especially at the downstream end of the value chain, experience a rapid decay when penetrating a host region, because of the lack of complementary FSAs required to operate successfully there.

Complementary FSAs are mechanisms to reduce transaction costs (e.g., investments in location-bound brand name development and reputation building). In other words, transaction costs increase as ‘distance’ increases, and a reduction of transaction costs requires investment in new FSA development to complement imperfectly deployable and exploitable, non-location-bound FSAs. As a result, the MNE's operations in the host region function on the basis of an overall FSA bundle that may differ significantly from the one deployed in the home region. The difference between both reflects the liability of inter-regional foreignness. A related issue raised by the authors is the observation of a large number of non-equity cross-border relationships. In many cases, these should be interpreted as attempts by MNEs to access inexpensively, and without the normal risks of equity involvement, the required complementary FSAs needed to operate effectively in host regions.

Finally, Dunning et al. observe that the transfer of institutional strengths by US firms to Japan has made Japan a more desirable location in which to invest. This is an important observation, as it suggests that early MNE investment in host regions may lead to spill-over effects benefiting subsequent investors. This occurs not only because the latter firms can gain access to the complementary FSAs crafted by the former, but also because the new FSA bundles introduced by the early investors somehow act as a best practice, diffused at the macro level in the host region. The empirical question then becomes whether such diffusion is sufficient to reduce fundamentally the liability of inter-regional foreignness faced by outsiders. The answer to this question will require more analysis, building upon micro (firm-level) data as used in Rugman and Verbeke (2004a).