Analytical Differences in the Economics of Geography: The Case of the Multinational Firm

The subject matter of the overview paper co-authored with Philip McCann of Reading University that I will talk about today is an examination of various specific aspects, focusing mainly on the multinational firm, particularly research and development (R&D). It addresses approaches to multinational firms as focused on within the international business literature, and about the geography of international business. These are two complementary research streams which have very strong connections to each other, but which have developed very largely in parallel. To our knowledge, there is relatively little overlap between these two areas. Therefore, there is a very fruitful avenue of research trying to link the developments in these areas together. The business literature has developed a great deal of understanding with regard to international and multinational economic activity. The economic geography literature has developed a great deal of understanding of economic activity, which spans over space. However, each one of these has much to gain from utilizing insights of the other.

Following that, I will address knowledge networks specifically within the multinational firm and then move on from there to talk about connections between multinational knowledge development and clusters. There has been a recent explosion in the literature on clusters in the management area following some seminal work by Michael Porter (1998). I will try to point out some salient problems with this literature and indicate avenues along which these problems can be addressed.

Why do we need to integrate these approaches? When we have different intellectual endeavors aimed at different objectives, integration is only necessary if there is something to be gained. The reason we need some integration in this case is that international business tends to be fundamentally aspatial or geographically artificial. Geography in international business is understood to be the boundary of the nation-state. These borders are drawn on the basis of politics, history, et cetera and do not necessarily have any core spatial idea within them.

Economic geography and regional science, on the other hand, take a very spatial view of economic activity where the main objective of study is geographic distance. The nation-state appears only as a geographic construct within the economic geography literature. However, the drawback to this approach is that the focus of economic geography and regional science literature is on economic activity and not the firm. The firm thus appears in a very simplistic form. Organizational issues in the context of firms operating over vast distances are not addressed at all.

While the international business literature takes a relatively simplistic view of geography, the economic geography literature takes a relatively simplistic view of the firm. These are both crucial weaknesses. Porter's work on clusters attempts precisely to address this weakness in the international business literature, but the assumption is that clusters imply advantage. This is not necessarily the case and it is too simplistic because it is not adequately grounded in the economic geography literature. It does not take into account the research findings of economic geography.

The international business approach tries to provide an understanding of the logic of the multinational firm. This approach draws its roots from the Reading school of international business and the seminal work of John Dunning, Mark Casson and Peter Buckley. The Dunning model says that firms are driven to multinational activities by three sets of advantages, typically denoted as the OLI paradigm: ownership advantages, location advantages and internalization advantages.

The first two are relatively straightforward, drawing directly from the underlying economics. Ownership advantages are generated predominantly from industrial economics: directly or indirectly from the control of resources or assets, economies of scale, product differentiation, management expertise, and access to capital. Location advantages are derived from economic geography or advantages from low wages, natural resources, labor productivity, market size, transport cost, access, et cetera.

The third set of advantages, those arising from internalization, is the unique contribution of the international business literature. These internalization advantages are derived from internalizing economic activities across national borders. In a standard model where there is a home country and a host country, one possible avenue of activities between these two is the market. However, the interfirm transactions can be replaced whenever the advantages of doing the activities within the firm are great enough. Whenever it is advantageous to control the downstream host country operation, or conversely, whenever the costs of utilizing the market mechanism are high enough, this kind of activity may be preferable.

In the traditional business approach, firms are driven to become multinationals whenever these three sets of advantages are large enough. Geography appears in the international business literature fairly tangentially. However, there is one model within international business that attempts to address this issue, the so-called Product Cycle Model associated with Ray Vernon (1966). This is the idea that there is a core-periphery approach to the technological ladder within multinational firms, whereby the home country activities are at the technological frontier and the host or foreign country activities are standardized or obsolescent. This 1960s model was essentially a product of its time, during a period of ascendancy of American multinational firms, where cutting-edge technology was developed in the vast American home market and then exported to other countries.

However, this model implies a hierarchical ordering and by the 1970s it was recognized as outmoded because there are two necessary factors in order for the model to work. First, the multinational must be able to depend wholly on its home market for its technological advantage. In a highly competitive world this dependence on a single source for R&D advantage became impossible. By the late 1970s, most multinational firms were introducing new products in home and host markets virtually simultaneously, and the gap had disappeared entirely by the 1980s. One of the drivers for this is that both host location and subsidiary operations are sources of advantage for the firm. In order for multinationals to compete with other multinationals, they need to tap into these technological competencies globally.

By the 1990s, this led to an observation that the mandates in the subsidiaries varied dramatically and differed from each other. Subsidiaries can be "home base exploiting" (traditional subsidiaries) or "home base augmenting" (creating an advantage for the home base). A number of research papers, including ours, point out that these mandates can change over time. Subsidiaries have a competitive drive to expand their mandates over time. As subsidiaries compete with each other within the same firm, there is an intrafirm dynamic of R&D mandate change. Subsidiaries have an incentive to present themselves to headquarters as being increasingly valuable as they seek to increase their importance and access to resources within the firm.

The outmoded Vernon model was replaced to a certain extent by the recent Porter clusters model, which we argue has some significant problems. These problems can be characterized by pointing out a stylized format for knowledge-creation and flows within a multinational firm. In a modern multinational, in a home-host setting, there are at least three stylized knowledge flows in addition to the core-periphery flow. One is called "knowledge-transfer" (reverse knowledge flows from subsidiary to parent for competitive advantage) generated by learning (knowledge flow into the subsidiary from the host location), and inevitably accompanied by spillovers (knowledge flow out of the subsidiary into the host location and to other firms).

One of the important messages is that the incentive structure of the firm is very important. For a multinational firm characterized by a high level of competition with other large multinational firms, it is extremely important to maximize learning from the host country and minimize spillovers. Very often in the economic geography literature and the technology management literature, there is an assumption that spillovers occur naturally as a matter of course. A key point of emphasis is that this assumption may be misguided. Spillover occurrence is extremely dependent on the nature of the firm.

If a subsidiary is a center of excellence, has a world product mandate, and is responsible for the technology R&D strategy of a particular business line of a multinational firm, the knowledge flow into the subsidiary is expected to be very large and it is less important for knowledge to flow from the subsidiary to the parent, as the subsidiary controls the technology strategy. If, on the other hand, technology strategy is under the control of the parent, the dominant flow of knowledge is expected to be from subsidiary to parent.

Parallel to this, space is extremely important in the economic geography and regional science literature where firms are modeled as points in space. Since economic geography does not focus on the organization of the firm, the only way in which firms appear is through locational configuration and this is a weakness associated with the model: It is quite possible that an unchanged locational configuration can be accompanied by a completely changed strategic configuration, as can be appreciated by examining the preceding models.

The core-periphery model in economic geography, a variant of the product cycle model at the subnational level, states that there is a difference in the sophistication of activities, with advanced activities driven more by resource requirements and standardized ones by cost. The problem with this approach is that it assumes a very centralized optimality approach to firm decision-making. Recent work in the area of R&D management and international business management indicates that this is not the case. In reality, multinational firms are large, complex, multidivisional entities, where divisions compete with each other and jostle for influence. Location optimality is often not the basis upon which decisions are made, but rather decisions are based on the realpolitik of activities in the organization. There is a need to recognize that large divisionalized firms are messy, complicated entities that do not function in a precise fashion with a single objective.

Putting together this messy, locational sophistication of economic geography with the messy, complex analysis of the firm in international business, gives a central role to knowledge as the basis for multinationals and the basis for clusters. Work in this area is being done by the Uppsala school, which is unique because it is formed by a dialogue among regional scientists and management scholars.

Two aspects of multinationals operating in clusters need to be highlighted. In the multinational knowledge network, one aspect is the notion of transfer. This is the idea of moving knowledge or leveraging knowledge across units within the same firm. Multinationals try to create competitive advantage by leveraging knowledge-transfer from one unit to another.

The second, equally important aspect is integration. All the knowledge that is coming in from diverse sources (local inflows, inflows from other subsidiaries, inflows from the parent country, and inflows from the parent units) has to be integrated in order to create something of value.

However, the general focus in international business is on transfer of knowledge. The Uppsala school is extremely connected with regional scientists, so their focus is much more on knowledge integration, such as embeddedness. Subsidiaries are valuable because they become embedded in the local system of innovation. For a complete analysis, to understand multinationals and clusters, it is necessary to incorporate both transfer and integration. Otherwise, the result will be suboptimal subsidiary performance. Subsidiaries with too much integration, to the extent that they lose contact with their parent, become starved of influence within their parent multinational, which is suboptimal. These are problems from the perspective of the firm because multinational firms are aimed at maximizing their own value.

Recent cluster literature points to location in clusters as sources of competitive advantage. The problem is that the spatial aspect, the geographical dimension, is never really specified. Two further aspects are that, first, advantage can only occur if there is some kind of interaction. There can be agglomeration centers that do not interact with each other and do not have any cluster-type properties. Second, cluster location generates costs as well as benefits. This means that firms located in clusters will be those that believe that there is a lot of gain to be made and little to lose. Firms that feel otherwise will not locate in a cluster. That means that cluster dynamics are not simplistic and are strongly conditioned by market structure. Therefore, market structure has to be taken into consideration in any theory of clusters. Clusters are not all the same. Any one cluster is going to have varying characteristics but they can typically be defined as predominantly being either "pure agglomeration," "industrial complex," or "social network" clusters.

A pure agglomeration is one in which the market structure is fundamentally competitive; interfirm relations are fragmented and unstable; membership is open; access is simply rental payments, but location is necessary to reap the advantages; the notion of space is predominantly urban; and the dynamics are stochastic. One example is a competitive urban economy. An industrial complex structure is related to the idea of flagship firms where there are some large firms with stable trading arrangements; a closed membership system; internal investment and location are essential; the notion of space is local but not necessarily urban; and the dynamics are going to be planned and strategic. One example here is in manufacturing. A social network is based on the sociological perspective. Interfirm relations are based on trust, loyalty, and are non-opportunistic; membership is partially open; access is based on history and location is necessary but not sufficient; and the dynamics are going to be mixed.

Our view of knowledge is that knowledge has both public-good and private-good aspects. The public-good aspect is the idea that outflows of knowledge are good because knowledge itself is a public good. The more outflows of knowledge occur in a particular area, the more attractive it is for knowledge-seeking firms to locate in that area. This aspect is going to dominate for competitive firms. The private-good aspects are going to dominate for oligopolistic industries, because the cost is very high. Large multinationals are unlikely to benefit from either pure agglomeration or social network clusters, and there is likely to be adverse selection.

Therefore, the result is R&D co-location in competitive industries and complementary location in the case of oligopolistic industries. There is evidence that large multinationals have no co-location of R&D with their competitive rivals. This means that big multinationals only go into clusters with non-core R&D as "listening posts" and do their core R&D in a protected area with industrial complex arrangements.

The main message is that in order to understand the dynamics of multinationals and clusters, it is extremely important to understand these twin views of knowledge: knowledge as a private good for oligopolies and knowledge as a public good for competitive firms. Therefore, clusters and multinational foreign direct investment (FDI) do not necessarily go together. The motivation of multinationals is to prevent outflows of knowledge. This has to be understood for good policy to be put together.

Commentator: ASAKAWA Kazuhiro, Faculty Fellow, RIETI /
Professor, Graduate School of Business & Management, Keio University

The important message of the paper is that a cluster by itself does not necessarily bring advantages for a large firm. The cluster provides multinational firms with the traditional, industrial complex type of interactions, but not necessarily with knowledge-intensive interactions. Thus, the benefit of locating in a cluster should not be overestimated. The question then is how firms can move from the traditional cluster presence to the more active, knowledge-intensive innovation mode. For this, in addition to the international business and economic geography paradigms, the strategic management paradigm may elaborate on the mechanism to leverage innovation within clusters.

There are several points I would like to clarify.. First, once a firm is located in a cluster, what will be the next step for the firm to leverage its locational advantages? Second, to what extent can a firm enjoy the advantage of being inside a cluster through alliances, as compared with greenfield arrangement? Also, can a firm without ownership advantages (according to Dunning's framework) enjoy the benefit of being inside a cluster as well? Last, what are the advantages for multinationals to bridging different clusters located throughout the world?

Q: Could you elaborate on the predominance of Japanese companies that tend to locate their R&D in America and Europe despite the benefits offered by India, where other companies are locating?

A: A lot of Japanese R&D tends to be very market-focused, so the connectivity between the downstream marketing focus and the upstream R&D focus tends to be fairly tight. Japanese firms tend to be very strong in manufacturing, and the connectivity between the R&D activity and the market is very important. U.S. and European companies tend to do better on the service end of activities. Therefore, you would expect Japanese R&D to be co-located with the market.

Q: Is there also a model where you incorporate intercompany service and goods flows to analyze business location decisions, in addition to the knowledge flows?

A: By focusing on knowledge, we do not mean to say that goods flows do not occur. We are trying to say that location tends to be driven by value-creation. Multinational value-creation tends to be strongly correlated with knowledge. Goods flows occur, but that side of the business tends to be more on the commodities end. There is not a lot of value being created there and it is being driven more by cost considerations.

Q: Please reply to the four questions raised by Dr. Asakawa.

Dr. Asakawa: Just to clarify, is it correct to assume that a multinational firm located in a traditional cluster cannot expect to extract benefits from the location in the same way as a small player?

A: What we are pointing to is that location in the knowledge context for large multinationals is driven by cost-benefit calculations. The cost-benefit calculus is that by locating the core activity in a chaotic setting with no protection, there is a lot to lose and very little to gain. For example, if BASF were to locate its software unit for chemical processes R&D, which is not a key but a necessary unit, in Silicon Valley, it could learn a lot and if anything leaked out to other firms, the effect would not be big.

How does the leakage occur? A lot of the knowledge-transfer occurs through the labor market. Fundamentally, when a person changes jobs, she or he takes that knowledge to the other company. Firms do not want this knowledge mobility to happen in the core activity. Therefore, the core activity is kept in a geographical area where the industrial complex nature of the R&D is such that if an engineer leaves, the chances are high that he or she will go work for a company that is connected to the first company. In this way, employees remain within the complex and go to one of the firm's suppliers or buyers or an associated firm. Therefore, knowledge does not escape. For a small start-up, this is not an issue.

With regard to cluster location, a forthcoming paper this year will look at biotechs. Pharmaceutical companies are very interested in biotechs because they are a source of cutting-edge innovation and growth that is lacking at large pharmaceutical companies. When biotechs discover something useful, pharmaceutical companies go and buy it. We have found that if you look at biotech hotspots in the United States, the way in which pharmaceutical companies operationalize their connection to this innovation activity is through joint partnership agreements.

There is a systematic difference between the drivers of partnership investments for domestic and foreign partners in biotech companies in hotspots in the United States. Domestic partners are driven by explanatory factors for revenue (patents, knowledge-creation, R&D spending, et cetera). For foreign partners, there was no explanatory parallel, and we found that they were driven by cluster characteristics. Our conclusion was that for domestic partners, they did not need any geographic benefits, so they were interested in the "what you know" as opposed to the "who you know" for foreign partners. In other words, for foreign partners, the firm is trying to get social capital and connections.

Dr. Asakawa: It has been found that most benefit is derived by those who have some sort of status as an entrepreneur or who know the key persons within a cluster. Otherwise, even if you are located in a hotbed of innovation, nothing happens. The key is who you know and what kind of reputation you have among the local entrepreneurs. In that respect, your paper has huge implications for foreign companies trying to get into local innovation. Perhaps network theory might be of some use to identify clusters in which "Who knows what?" is important.

A: I do not have a lot of expertise in network theory. We were trying to examine the old-fashioned social networks where there was some degree of relational capital that was necessary to minimize transaction costs. If you are a foreign firm, in order to acquire this relational capital, you need to generate a joint venture with a local firm.

One final thing I would like to point out is that in our research we focused predominantly on greenfield investments. However, in related research, we have looked at the different characteristics of R&D strategies within the context of acquisition versus the subsidiary mandate. When subsidiaries have a high mandate, then acquisition and acquisition entry tend to be associated with greater levels of R&D. When subsidiaries are home base exploiting, acquisition tends to be associated with decreased R&D.

If you have a multinational firm that is entering a location through acquisition, then it is important as a policymaker to understand the objectives of entry. If you have a firm entering essentially to utilize the acquisition as a market-seeking vehicle, then the objective of acquiring this firm is simply market share. Therefore, you are likely to see any duplication in terms of R&D cancelled out. There is a complete bifurcation depending upon the objective of entry for the multinational. If you are a policymaker and want to encourage FDI, you have to consider the kind of post-entry behavior to be expected. You can predict that based upon what the firm is looking to do.