An Analysis of the Relationship between Product Diversification, Geographical Diversification and Technological Diversification

(1)

Discussion Paper on International Management and Innovation

Michael Stephan¹

An Analysis of the Relationship between Product Diversification, Geographical Diversification and Technological Diversification

Discussion-Paper 02-02

Stuttgart, December 2002

ISSN 1433-531X

1 Dipl.-oec. Michael Stephan, Center for International Management and Innovation. Contact: Department of International Management (510K), University of Hohenheim, D-70593 Stuttgart, Tel: ++49-711-459-3761, Fax: ++49-711-459-3446, E-mail: stephanm@uni-hohenheim.de

(2)

Abstract

Building upon a firm-level empirical investigation we conduct a phenomenological analysis of corporate diversification patterns. We take up a more holistic, multilevel approach towards corporate diversification: In our investigation we capture diversification at the output level, at the input level, and at the geographical market level and analyze the relationships between the individual dimensions. The analysis is based on a study of trends in corporate diversification over a fifteen years period ranging from 1983 to 1997. Our sample comprises 46 multinational corporations headquartered all over the triad countries. Based on our three-dimensional view of diversification, we set up three propositions on the relationships between the individual diversification strategies. We then test the correlations between these diversification strategies and derive different phenotypes of corporate diversification. Each phenotype is characterized by a distinctive corporate diversification pattern. From a dynamic perspective these phenotypes translate into paths of evolution of corporate diversification patterns. It can be shown, that there are several homogenous groups of sample companies with similar corporate diversification patterns and similar paths of evolution.

Keywords:

Product Diversification Technological Diversification Geographical Diversification Growth Strategies

(3)

Introduction

The world’s largest multinational corporations are actively engaged in a multitude of businesses and usually manage a broad spectrum of technological resources. In contrary to international diversification and product diversification, the latter phenomenon - technological diversification - has attracted comparatively little attention in management research (Breschi, Lissoni, and Malerba, 1998). A few empirical studies have come to the conclusion that technological diversification had been on the rise in the 1970s and 80s (Fai, 1999; Granstrand and Oskarsson, 1994; Kodama, 1995; Oskarsson, 1993). This increase has been observed at corporate level across all industries and across all triad countries. Various empirical inves- tigations in international management research have revealed that most of the multinational corporations have further increased the geographical range of their business activities during that period of time. However, and most interestingly, at the same time numerous studies have revealed that product diversification has been declining during the same period of time (e.g. Markides, 1996). Common to most of these studies is that they take up a partial view on corporate diversification. Product diversification, technological diversification and geographic diversification have usually been analyzed separately as distinctive phenomena or alternative routes to growth. Studies either focus on output diversification, i. e. product or business diversification, or they concentrate on input diversification, e. g. technological diversification, or they analyze international diversification, i. e. geographical diversification. Only a few studies in the international and strategic management disciplines have tried to take up at least a two-dimensional view of corporate diversification. Beyond these two-dimensional approaches there exists no detailed multilevel-analysis of the corporate diversification phenomenon.

This paper tries to overcome this deficiency and takes up a more holistic, multilevel approach. In our empirical investigation we will capture both diversification at the output level, at the input level and at the geographical level. Building upon a corporate-level empirical investigation we conduct a phenomenological analysis of corporate diversification patterns.

The analysis is based on a study of trends in corporate diversification over a fifteen years period ranging from 1983 to 1997. Our sample comprises 46 multinational corporations headquartered all over the triad countries. Based on our three-dimensional view of diversification, we set up three propositions on the relationships between the individual diversification strategies. We then test the correlations between these diversification strategies and derive different phenotypes of corporate diversification. Each phenotype is characterized by a distinctive corporate diversification pattern. From a dynamic perspective these phenotypes translate into paths of evolution of corporate diversification patterns. It can be shown, that there are several homogenous groups of sample companies with similar corporate diversifi-

(4)

cation patterns and similar paths of evolution. Furthermore, most of the diversification patterns change only gradually over time.

The paper is organized according to the following plan. The first section gives an overview of the relevant strands of literature that incorporate more holistic, two-level perspectives on corporate diversification and address the relationship between different diversification dimensions. In the following we set up the resourced-based framework on the relationship between product diversification, technological and international diversification strategies. The resource-based framework incorporates three theoretical propositions about the relationships between each of the dimensions. The next section introduces our empirical data and discusses the methodology that we have employed in capturing corporate diversification. In addition we present the recent trends in product diversification, technological diversification, and geographical diversification of the sample companies. We then present the results of the correlation analysis and test our propositions. Building upon the empirical results we derive different phenotypes of corporate diversification patterns. We conclude the paper with an illustration of the individual paths in the evolution of the corporate diversification patterns.

Multilevel Perspectives on Corporate Diversification: Synopsis of the Literature Today, the literature on corporate diversification not only covers a wide range of research is- sues, but also represents a great variety of perspectives and disciplinary paradigms. Since the middle of the 70s, literature on product diversification and its effect on corporate performance has been a mainstay of strategic management research. The international management discipline has dealt exhaustively with the phenomenon of geographical diversification. And lastly, technological diversification has been a primary focus for quite a number of researchers in the technology management domain. However, most researchers have dealt with the individual corporate diversification dimensions in an isolated way. Only a few studies take on a more holistic, two-level perspective on corporate diversification and address the relationship between two diversification dimensions. In the following we give a short overview on the studies that incorporate a two-dimensional perspective.

Literature on the Relationship between Product and International Diversification

Within the plethora of strategic management literature on product diversification one can find two types of studies that consider international diversification as a further variable. The first category represents the majority of these studies and contains work that investigates the combined effects of product and international diversification on firm performance. These studies do not look directly at the relationship between product diversification and interna-

(5)

tional diversification since the primary focus is on performance. E.g., Geringer et al. (1989) conduct an explanatory analysis of potential links between MNCs’ performance and the diversification strategy and degree of internationalization which characterized their operations.

They tested their hypotheses using data on 200 of the largest MNCs in the U.S. and Europe.

The study’s results showed that in fact the degree of internationalization has an important role in understanding performance differences among diversified MNCs. Based on a sample of 62 MNCs Kim et al. (1989) analyzed the impact of global diversification strategy on corporate profit performance by integrating the product and the international market dimensions of diversification. They found that integrated international and product diversification strategies had a positive influence on firm profits growth. Sambharya (1995) examined the individual and joint effects of product and international diversification on firm performance. In a study of 53 U.S.-based MNCs the author made the observation that both international and product diversification strategies are not profitable by themselves. However, he found that the inter- action effects of product and international diversification lead to a substantial increase in firm performance. Hitt et al. (1997) tested the effects of international diversification on innovation and performance in product-diversified firms. They suggest that product diversification positively moderates the international diversification and performance relationship. Kim et al.

(1993) analyzed the influence of global diversification and product diversification on the risk- return trade-off, i.e., firms with high returns can have low risk (Bowman paradox). In a sample of 152 large U.S. multinationals the authors found that diversified firms can indeed achieve a favorable risk-return performance with global market diversification.

In contrast to the work that primarily focuses on the performance effects of combined diversification, the second category of research directly addresses the relationship between product and international diversification. This limited number of studies come to a somehow puzzling conclusion as they indicate a negative relationship between international diversification and product diversification. E.g., Madura and Rose (1987) reported that reverse product diversification occurred in favor of international diversification. In a study of 262 British manufacturing firms Grant et al. (1988) reported that multinational diversity was strongly negatively associated with product diversity. Lastly, Sambharya (1995) analyzed the relationship between product and international diversification. In his study of the combined effect of international diversification and product diversification strategies on firm performance he observed an inverse relationship between both diversification dimensions. Sambharya (1995) argues that product diversification and international diversification strategies require different types of skills and are risky, since in both strategies firms spread themselves out in terms of product proliferation or market proliferation. According to him it is unlikely that firms would take both risks simultaneously. Thus, MNCs that are diversified internationally are likely to be less diversified in terms of products. Though, finally it has to be noted that, apart from the

(6)

work of Madura and Rose (1987), none of the studies specifies the nature of the causality between product and international diversification.

To sum it up, the strategic management literature on the combined effects of product and international diversification on firm performance comes to the (comparatively unambiguous) conclusion that international diversification positively moderates the relationship between product diversification and performance. The authors from that field argue for this moderation effect from the resource-based perspective. An integration of product and international diversification helps firms exploit interdependencies across their businesses to achieve potential scope economies. Thus, from the resource-based view of the firm, the structures and capabilities developed to implement product diversification strategies should also help implement international diversification, and vice versa. In contrast to these findings stand the empirical observations of the authors that more directly address the relationship between international and product diversification. These studies indicate a negative relationship between international diversification and product diversification. In opposition to the arguments above it is argued that both diversification strategies require different types of skills and are too risky.

Literature on the Relationship between Technological and International Diversification

The number of studies that address the relationship between technological diversification and international diversification has been growing considerably during recent years. Most of the contemporary research suggests that these strategies have become essentially complementary to each other. As opposed to the work on the relationship between product and international diversification the overwhelming number of studies on the relationship between technological and international diversification also incorporate assumptions concerning the causality between both diversification strategies. Theoretical arguments supporting the positive relationship between both diversification strategies have been advanced that suggest two different positions: either technological diversification precedes internationalization or internationalization precedes technological diversification. In a study of large technology- based firms in Europe, Japan and US, Granstrand (1996) argues that in recent years a positive causation has run from technological diversification to internationalization. He hypothe- sizes that an increase in technological activities leads to higher R&D expenditures, which can be recovered through the internationalization of activity, serving a wider range of markets.

Especially in high-technology industries, firms operating only in domestic markets may find it difficult and time consuming to recoup initial R&D investments. From this point of view, international diversification of markets and activities has been found to be the main response to a broadening of the firm’s technology base.

(7)

On the other hand, quite a number of theoretical and empirical models conversely argue, that international diversification precedes technological diversification. This work has been devised of the process by which MNCs access locationally dispersed technological assets through their international operations. The models suggest that the geographical diversification of markets and activities has a positive influence on the firms’ innovative activities and heavily promotes the diversification of technological activities. Mitchell et al. (1996) conduct a study about the direction of causality between international diversification and investment in intangible assets in the case of 239 public American manufacturing companies. The authors find that firms increase their R&D expenditures after expanding their multinational structure, but not vice versa. Of course, the recorded increase in R&D expenditures may be attributable to activities that Kuemmerle (1996) terms home-base exploiting R&D activities. Home-base exploiting R&D activities support the local adaptation of the firm’s products and do not necessarily imply an increase in the span of technological activities and resources. Home- base exploiting R&D activities facilitate the local exploitation of the MNCs core capabilities.

To control for this exploitative kind of technological activities, several studies directly address the causal relationship between a firm’s internationalization of technological activities and the degree of technological diversification. Cantwell (1995) has suggested a positive causal relationship running from international networks of technological development to corporate technological diversification. His idea is that MNCs with internationally integrated research facilities are able to access locationally dispersed technological assets in the relevant host countries. Hence, geographical diversification of MNCs becomes an exogenous source of technological diversification. In a study of 24 Swedish MNCs, Zander (1997) argues likewise that the geographical dispersion of innovative activities may come to facilitate the technological development of the firm, since the MNC can tap into alternative streams of innovation in different centers, and integrate technologies from different national sources on an international scale. More generally, Birkinshaw and Hood (2001) point out that growth-triggering innovation often emerges in foreign subsidiaries – from employees closest to innovative cus- tomers and least attached to the procedures and politesse of the headquarters.

While most of the studies on the relationship between international and technological diversification have relied on prior assumptions on the variables’ exogenity and thus provide little evidence about the real determinants of the interrelationship between technological diversification and geographical diversification, Cantwell and Piscitello (1997) analyzed the causal relationship between both dimensions. In their empirical study comprising a sample of 40 U.S. and European MNCs in the 1980s, the authors come to the conclusion that the causality between geographical and technological diversification runs in both directions. Increases in technological competence, together with the greater dispersion across fields of technological

(8)

activities and across geographical sites, are likelier to be combined, and to mutually reinforce one another. Cantwell and Piscitello (1997) argue from a network perspective and point out that in the MNC network each affiliate specializes in accordance with the specific characteristics of local production conditions, technological capabilities and user require- ments. In this event international diversification of (technological) activities and technological diversification (competence creation) become interconnected and thus mutually positively related parts of a common process. Supplementary to these results, Cantwell and Piscitello (1999a) found that there is a trade-off between geographical complexity and technological complexity in managing international networks. Firms which have developed internationally integrated networks for technological development have tended to concentrate either on a wider geographical dispersion of competence creation, or alternatively instead on a wider sectoral dispersion of interrelated competencies across technical fields but in a more con- fined set of geographical locations.

Most of the research work cited above has focused on the relationship between technological diversification and international diversification from a quite recent perspective as they analyzed the situation in the 80s and early 1990s. In contrast, Piscitello and Cantwell (1999) take on a historical perspective on corporate diversification and explore the evolution of the relationship in 166 large European and U.S. industrial firms from 1901 to 1995. They found that only since the 1980s geographical diversification and technological diversification (competence creation) have become interconnected and mutually positively related parts of a common corporate development process. Prior to the 1980s they could not identify a positive relationship between both strategies. According to them, in the inter-war and early post-war periods, there was no particular linkage running from corporate technological diversification to the internationalization of the firm’s capability. Thus, in the earlier stages of development of large firms it is reasonable to depict technological diversification and internationalization of markets as two alternative strategies for corporate growth. While large firms were commonly diversifying their technological competence in the normal course of growth, their internationalization was aimed at the wider exploitation of the basic technological competencies and resources they had already established at home rather than at extending their competencies and resources into new fields abroad.

Literature on the Relationship between Technological and Product Diversification

Only a few theoretical and empirical studies have focused on the relationship between technological diversification and product diversification. Most authors implicitly assume that technological diversification and product diversification should be intrinsically linked (Stephan,

(9)

2001). Pioneering insights into the phenomenon of technological diversification in multi-business corporations beyond such basic assumptions have been contributed by Kodama (1986), Pavitt et al. (1989) and Granstrand and Sjölander (1990). They have shown that today the world’s largest multinational corporations are not only actively engaged in a multitude of businesses but also manage a broad spectrum of technological resources. Subsequently, a few empirical studies have come to the conclusion that technological diversification in multi-business corporations had been on the rise in the 1970s and 80s (Fai, 1999;

Granstrand and Oskarsson, 1994; Kodama, 1995; Oskarsson, 1993, Sjölander and Oskars- son, 1995). This increase has been observed at corporate level across all industries and triad countries.

Fai and Cantwell (1999) as well as Fai (1999) have analyzed the development of the relationship between technological diversification on the one hand, and corporate growth and product diversification on the other hand from the beginning of the last century up to the 1980s. They have shown that technological diversification was historically associated with increases in firm size and product diversification, whereas in the 1980s it appeared that diversification into new technologies may or may not lead to diversification into new product areas. Obviously there is a certain degree of decoupling of technological diversification from product diversification and corporate growth in the 1980s. These results are in line with empirical observations made by other authors who revealed that technological diversification has been on the rise in the 1980s while during the same period (unrelated) product diversification has been declining in many MNCs. These MNCs have refocused their business lines and simultaneously expand the span of their technological activities (Markides, 1996; Gran- strand and Oskarsson, 1994; Gambarella and Torrisi, 1998).

Some authors have analyzed the causality between technological diversification and product diversification. According to Granstrand and Sjölander (1990) technological diversification was a major determinant of corporate growth in Swedish MNCs in the 1980s. In a another empirical study from the 1980s Chatterjee and Wernerfelt (1991) were able to explain product diversification in the 1981-1985 period with the technology profiles of their sample companies prior to 1981. There seems to be a time lag between product diversification and technological diversification steps: Companies that plan to diversify their range of businesses first have to tap into the new technological resources that are necessary to develop new products. Thus, technological diversification is largely determined by product diversification.

However, technological diversification steps precede product diversification (Stephan, 2002).

(10)

The synopsis of the literature adopting a two-level perspective on corporate diversification has revealed inconsistencies in the core results of the studies. Inconsistencies in the results do not only appear within the individual strands of literature on the bilateral relationships but, and even worse, also across the three two-level perspectives on corporate diversification.

These inconsistencies remain, even if we exclude the issue of causalities between the diversification strategies. In the following we draw on the resource-based view of the firm and try to resolve some of these inconsistencies.

A Resource-based Framework on the Relationship Between Product Diversi- fication, Technological and International Diversification Strategies

We start our resource-based framework with propositions about the relationship between technological diversification and product diversification since the technology management literature offers only little inconsistencies with regard to the core results. Although some authors have observed a certain degree of decoupling between technological diversification and product diversification in the 1980s we still propose a strong positive relationship between both diversification dimensions. We further consider a natural time lag between input and output diversification steps. Firms from high-technology sectors that intend to diversify their lines of businesses into more or less related areas may of course draw on their existing resource-base to realize economies of scope, but, in addition, will have to accumulate new, complementary technological resources that are necessary for developing new products.

Proposition 1a

Product diversification and technological diversification are tightly coupled strategies to corporate growth into new fields in dynamic, technology intensive industries: Compa- nies that plan to diversify their range of businesses must tap into new technological re- sources that are necessary to develop the new products.

Proposition 1b

Technological diversification precedes product diversification: Companies that plan to diversify their range of businesses first have to tap into the new technological re- sources.

More or less consistent results are also presented by the contemporary research on the relationship between international and technological diversification. Most of the studies in this domain suggest that these diversification strategies have become essentially complementary to each other. The extent to which a firm has diversified its portfolio of technological competencies and resources influences (as well as being influenced by) the extent to which those capabilities are internationalized. Increases in technological competence, together with the greater dispersion across fields of technological activities and across geographical sites, are

(11)

likelier now to be combined, and to mutually reinforce one another. International diversified firms may be better able to broaden their technological capabilities by tapping the various technological resources available globally. At the same time, international diversification may generate the resources necessary to sustain large-scale R&D operations that are necessary for innovating simultaneously in multiple technological fields. The costs associated with technological diversification can be recovered through the internationalization of activity, serving a wider range of markets. We therefore propose a positive relationship between both diversification dimensions.

Proposition 2

Technological diversification and international diversification are complementary ways of competence and resource building: Diversifying the technology base and expanding the geographical scope of the operations are prerequisites for corporate growth and development.

The strategic management literature on the relationship between product diversification and international diversification has produced rather inconsistent results. The strand of literature that deals with the combined effects of product and international diversification on firm performance comes to the conclusion that international diversification positively moderates the relationship between product diversification and performance. In contrast, studies that directly address the relationship between international and product diversification indicate a negative relationship between international diversification and product diversification. The results of these latter studies are somehow puzzling since they suggest that, despite a good chance to achieve positive performance effects, firms will not risk to pursue combined product and international diversification strategies. Furthermore, from a holistic, three-dimensional perspective on corporate diversification, the proposition of a negative relationship between product diversification and international diversification produces model inconsistencies. Since we propose positive relationships on the other two bilateral diversification levels, a negative relationship on the third level would not fit into the complete model. We thus rely on the indications offered by the literature on the combined effects of product and international diversification on firm performance and propose, from a resource-based view, a positive relationship between product and international diversification. An integration of product and international diversification strategies helps firms exploit interdependencies across their businesses to achieve potential scope economies.

Proposition 3:

Product and international diversification are complementary ways to corporate growth and development. Resources and competencies developed to implement product di- versification strategies should also help implement international diversification, and vice versa.

(12)

In the following we will test our propositions with a simple correlation analysis between the individual diversification levels. Before we start our analysis, we shortly introduce our sample data and the methodology used.

Methodology – Sample Selection and Data Collection

To analyze corporate diversification patterns of firms we selected 46 multinational corporations from the international R&D Scoreboard ranking of the top 300 companies by R&D expenditures (random selection).² The sample companies are headquartered in the U.S. (15), Europe (12) and Japan (11). For each company we assembled an extensive data set cover- ing statements about financial measures (revenues, R&D expenditures), technological resources (technology portfolio), product diversification (product portfolio) and international diversification (geographical market portfolio) for a time period of 15 years (1983-1997) on an annual basis. Data on product diversification and international diversification were assembled from the annual reports and other primary company sources. Product diversification is captured by the dispersion of the firms’ product sales across 4-digit-ISIC codes. Corre- spondingly, international diversification is captured by the dispersion of the firms’ sales across major regional (geographical) markets.³ In contrast to product diversification and international diversification, a direct measure of technological resources or technological activities of firms is not available. The approach to grasp technological diversification is to view technology as consisting of a number of distinct ‘technological areas’ (for a similar approach see e. g. Jaffe, 1989). A multi-technology corporation will typically engage in R&D in a number of such areas. From this perspective, technological activities can be measured by two fundamental proxies: patents and R&D expenditures. While R&D expenditures are an input measure, patent filings are an indicator of the firm’s innovative output (Gavetti, 1994).

The paper uses patent data to characterize the technological position of firms and to measure technological diversification.

The method of measuring technological diversification in our investigation is based on the notion that patents are a more valid indicator to assess technological resources and activities of firms. According to Gavetti (1994) and Pavitt (1988) we can synthesize the major advan- tages of the use of patents as follows:

2 The R&D scoreboard is published annually by the UK Department of Trade and Industry (DTI).

(13)

Patent data are effective and valid measures of the technological activities of companies;

Patent data also capture technological activities that are not rooted in the formal R&D or- ganization;

Patents offer detailed information on the relevant technological area, which is of particular relevance in order to assess the spectrum of technological activities of companies.

However, the author is well aware that patents are far from being a perfect proxy. The con- struct validity of the indicator is weakened by the fact that patents do not grasp technological activities that are not characterized by technical novelty (originality). The contents validity of patent data is impaired by the fact that the propensity to patent differs amongst technical fields, reflecting differences in the relative importance of patenting as a protection against imitation. In the absence of a more appropriate alternative, we believe that the granting of a patent reflects the judgment that the applicant has the competence to improve technology in a given field significantly.

To minimize the home country bias in comparing the technological activities of the sample companies headquartered across the U.S., Japan and Europe, we analyze the patent filings at the European Patent Office (see also Schmoch, 1999). The primary field of technological activity of each patent can be derived from the International Patent Classification system (IPC). To assess the diversity of the patent portfolios of the sample companies we have adopted a technology-oriented classification system which has been elaborated jointly by the German Fraunhofer Institute of Systems and Innovation Research (ISI), the French Patent Office (INIPI) and the Observatoire des Science and des Techniques (OST). The so called

“OST/INPI/ISI”-technology classification is based on the IPC and distinguishes between 30 different fields of technology and five higher-level technology areas (see figure 1). For each patent filing we collected information about the IPC class which was assigned by the patent examiners at the European Patent Office and then reassigned it to the corresponding OST/INPI/ISI-classification.

3 As alternative measures of international diversification one may use the global dispersion of firms’ assets and international employment data. However, these indicators generally suffer from the lack of availability of firm-level data. For a detailed discussion of the scope and limits of indicators to measure the degree of internationalisation see Stephan and Pfaffmann (2001).

(14)

Figure 1. OST/INPI/ISI-technology Classification Defined by IPC symbol

Electrical machinery and apparatus, electrical energy Audio-visual technology

Telecommunications

Information technology

Semiconductors Electrical engineering

Optics

Analysis, measurement, control technology Medical technology

Nuclear engineering Instruments

Organic fine chemistry Macromolecular chemistry, polymers Pharmaceuticals, cosmetics Biotechnology Agriculture, food chemistry Chemistry, Pharmaceuticals

Chemical and petrol industry, basic materials chemistry

Chemical engineering

Surface technology, coating

Materials, metallurgy Materials processing, textiles, paper Process engineering

Handling, printing Agricultural and food processing, machinery and apparatus Environmental technology

Machine tools Engines, pumps, turbines

Thermal processes and apparatus

Mechanical elements

Transport Space technology, weapons

Consumer goods and equipment

Civil engineering, building, mining Mechanical engineering

I. II. III. IV. V.

Electrical machinery and apparatus, electrical energy Audio-visual technology

Telecommunications

Semiconductors Electrical machinery and apparatus, electrical energy Audio-visual technology

Telecommunications

Semiconductors Electrical engineering

Optics

Organic fine chemistry Macromolecular chemistry, polymers Pharmaceuticals, cosmetics Biotechnology Agriculture, food chemistry Chemistry, Pharmaceuticals

Chemical and petrol industry, basic materials chemistry

Chemical engineering

Handling, printing Agricultural and food processing, machinery and apparatus Environmental technology Chemical engineering

Handling, printing Agricultural and food processing, machinery and apparatus Environmental technology

Machine tools Engines, pumps, turbines

Mechanical elements

Civil engineering, building, mining Mechanical engineering Machine tools Engines, pumps, turbines

Mechanical elements

Civil engineering, building, mining Mechanical engineering

I. II. III. IV. V.

For each sample company we compiled the technology portfolios for the three five-years periods 1983-1987, 1988-1992 and 1993-1997. The aggregated observation of five-years periods served to eliminate cyclical fluctuations in the individual firms’ patent filings. Although the EPO database covers information about patent filings dating back to 1978, our analysis starts in 1983. In fact, the sample of patents from 1978-1982 could be biased by the fact that only large European firms are likely to have used the European Patent Office since its very beginning. Figure 2 illustrates the profiles of the technology portfolios of the sample companies for the period 1993-1997 according to their ‘core’ technology fields. By employing a non- hierarchical cluster analysis we were able to identify eight homogenous groups of companies with similar technology profiles (“Telecom”, “Computer/Electronics”, “System Technology”,

“Automotive”, “Engines/Pumps/Turbines”, “Materials”, “Chemicals”, “Pharmaceuticals”). Each cluster is characterized by a specific set of core technologies that the member companies have focused on. Besides the eight clusters with homogenous technology profiles we have a residual of companies that do not fit into any of these groups. The nonconforming companies are grouped into the “hybrid” cluster.

(15)

Figure 2. Technology Clusters in the Sample (1993-1997 Period)

Engineering/

System Techn.

Telecom.

Computer, Electronics Data processing, Audio-visual technology

Telecommunication Electronics

Optics, Cybernetics

Motor, Turbine Technology

Transport

Machine Tools Thermal Processes,

Equipm.

Materials Production

Technology Organic

Chemistry

Polymers Pharma-

ceut.

Basic Chemistry Biotech-

nology

Nokia Ericsson Motorola Alcatel

Phillips Sony Matsushita

Mitsubishi Toshiba IBM

HP NEC Sharp

ABB Siemens Hitachi

Kyocera

Automotive Ford General Motors

Toyota Daimler Benz

Bosch

Materials

Thyssen Alcoa Krupp

Saint Gobain Transport Techn.

Pharma Glaxo

Kao Lilly Pfizer Novo Nordisk

Roche Merck USA

Chemistry Bayer

Novartis Akzo Nobel Dow Chemical

BASF DuPont

Toray Merck D

General Electric Allied Signal

UTC Medical

Technology

Sulzer Johnson&Johnson

Engineering/

System Techn.

Telecom.

Computer, Electronics Data processing, Audio-visual technology

Telecommunication Electronics

Optics, Cybernetics

Motor, Turbine Technology

Transport

Machine Tools Thermal Processes,

Equipm.

Materials Production

Technology Organic

Chemistry

Polymers Pharma-

ceut.

Basic Chemistry Biotech-

nology

Nokia Ericsson Motorola Alcatel

Phillips Sony Matsushita

Mitsubishi Toshiba IBM

HP NEC Sharp

ABB Siemens Hitachi

Kyocera

Automotive Ford General Motors

Toyota Daimler Benz

Bosch

Materials

Thyssen Alcoa Krupp

Saint Gobain Transport Techn.

Pharma Glaxo

Kao Lilly Pfizer Novo Nordisk

Roche Merck USA

Chemistry Bayer

Novartis Akzo Nobel Dow Chemical

BASF DuPont

Toray Merck D

General Electric Allied Signal

UTC Medical

Technology

Sulzer Johnson&Johnson

Analogue to the technology portfolio we compiled the geographical market portfolios and product portfolios for each company. To compile the geographical market portfolios we collected the firms’ revenues in the major global regions. In our study, the major global regions correspond to relatively heterogonous geographical market areas based on their economic and political conditions. We differentiate between 6 geographical market areas. These are:

North America, Latin America, Europe, Africa/Middle East, Asia/Pacific and the individual home country of each sample corporation. To compile the product portfolios we collected the firms’ revenues in different ISIC classes (International Standard Industrial Classification) on a 4-digit level. Figure 3 illustrates the profiles of the product portfolios of the sample companies for the period 1993-1997 according to their top-selling businesses. Again, by employing a non-hierarchical cluster analysis we were able to identify homogenous clusters of companies. On the whole, we have identified ten industry clusters with more or less homogenous product profiles: “Chemicals”, “Pharmaceuticals”; “Materials”; “Metal Products”; “Automotive”;

“Engines/Machinery”; “Diversified Electrical Engineering”, “Telecommunications”; “Consumer Electronics”, and “IT/Computer”. Surprisingly, the product clusters only partially correspond to the technology clusters. While some of the clusters find rough correspondents in the other dimension, with moderate differences in terms of composition, others do not have similar counterparts. Obviously there is a certain degree of decoupling. The phenomena of decoup-

(16)

ling will be picked up again when we observe the latest trends in technological diversification and product diversification.

Figure 3. Product Clusters in the Sample (1993-1997 Period)

Divers. Electrical Engineer.

Machinery&Equipm.

Telecom.

Computer

Data Processing/IT Telecommunications Consumer Electronics

Electrical Engineering Products

Mechanical Engineering

Basic Materials

Turbines, Engines, Machinery

Basic Metals, Metal Components Non-metallic

Materials Agro-

chemicals Pharma- ceuticals

Chemicals

Nokia Ericsson Motorola NEC

Sharp

Hitachi Mitsubishi

Auto- motive

Materials Kyocera Saint Gobain Pharma

Merck KgaA Novartis

Chemicals

Toray Medical

Products and Services

Bayer BASF Dow

Kao

Chemical fibres, Oil DuPont Akzo J&J

Merck Co.

Metal Products

Alcoa Thyssen

Krupp

Auto- motive Allied Signal

Bosch Consumer Electr.

IBM HP Sony

Matsushita Phillips

Toshiba Optical

Products, Fine Mechanics, White Goods

Daimler Ford

GM Toyota Glaxo

Lilly Pfizer Novo Nordisk

Roche

Alcatel Siemens

GE Sulzer

UTC ABB

Divers. Electrical Engineer.

Machinery&Equipm.

Telecom.

Computer

Data Processing/IT Telecommunications Consumer Electronics

Electrical Engineering Products

Mechanical Engineering

Basic Materials

Turbines, Engines, Machinery

Basic Metals, Metal Components Non-metallic

Materials Agro-

chemicals Pharma- ceuticals

Chemicals

Nokia Ericsson Motorola NEC

Sharp

Hitachi Mitsubishi

Auto- motive

Materials Kyocera Saint Gobain Pharma

Merck KgaA Novartis

Chemicals

Toray Medical

Products and Services

Bayer BASF Dow

Kao

Chemical fibres, Oil DuPont Akzo J&J

Merck Co.

Metal Products

Alcoa Thyssen

Krupp

Auto- motive Allied Signal

Bosch Consumer Electr.

IBM HP Sony

Matsushita Phillips

Toshiba Optical

Products, Fine Mechanics, White Goods

Daimler Ford

GM Toyota Glaxo

Lilly Pfizer Novo Nordisk

Roche

Alcatel Siemens

GE Sulzer

UTC ABB

Measures and Trends in Technological Diversification (1983-1997)

In our study we understand technological diversification as the extent of the spread of the firms’ technological activities across the 30 different fields of technology defined by the OST/INPI/ISI-classification. As described above, these technology fields aggregate into 5 technology areas. The degree of technological diversification is captured with the entropy measure of diversification. The single entropy measure grasps the degree of unrelated or total technological diversification and considers the spread of the patent filings across the 30 different technology fields (i=1...N; N ≤ 30). Let T_i be the share of the ith technology field in the total patent filings of the firm. For each sample company then the degree of unrelated technological diversification (DT) is computed as follows:

) / 1 ln(

1

i N

i

i T

T DT

∑

=

= .

(17)

This measure takes into consideration two elements of technological diversification: the number of technological fields in which a firm operates and the relative importance of each of these fields compared to the total number of patent filings. To include the degree of re- latedness among the various technology fields in which the firms are active, we assume that technological activities in different technology areas are unrelated, and that technological activities belonging to different technology fields but in the same technology area are related (Markides 1995). Let Tij be the share of the field in technology area j in the total patent filings of the firm. DR_j is then defined as the related technological diversification arising out of the patent filings in several technology fields within one technology area j:

∑

=

= ^N

i ij

j ij

T T DR

1

1) ln(

Since the firm has filed patents in several technology areas (j=1…M, ≤ 30), its total degree of related technological diversification is equal to

∑

=

= ^M

j

j jT DT DTR

1

where T_j is the share of the jth technology area in the total patent filings of the group.

Finally, we control for technology-specific differences in the propensity to patent, caused by differences in the relative importance of patenting as a protection against imitation, and firm- specific differences in the propensity to patent, caused by differences in firm-size and the country-of-origin effect, by the use of another diversification measure that is based on the Revealed Technological Advantage (RTA) index. The RTA of a sample company in a particular field of technology is given by the firm’s share in that field of all EPO patent filings of the sample companies, relative to the firm’s overall share of all patents filed by the sample companies. If P_ix denotes the number of patent filings of company x in technology field i, the RTAix is then defined as:

⎟⎟

⎠

⎜⎜ ⎞

⎝

=⎛

total x

itotal ix

ix P P

P

RTA P .

The index varies around unity, such that a value in excess of one shows that the firm is comparatively advantaged in that technology field in relation to other firms. In this manner technology-specific and firm-specific differences in the propensity to patent are normalized.

We have chosen to use the inverse value of a sample firm’s coefficient of variation of the RTA index across all technology fields as the RTA-based measure of technological diversification (RTAD):

(18)

( )

^⎟^⎟^⎠

⎞

⎜⎜

⎝

=⎛

⎟⎟

⎟

⎠

⎞

⎜⎜

⎜

⎝

⎛ −

=

∑

=

x x

RTA RTA

x N

i

ix x x

RTA

RTA N RTA

RTAD

σ

µ

1

1 2

1

Figure 4 provides an overview of the trends in unrelated technological diversification differentiated by industries of the sample companies. In the 1980s, the sample companies have (slightly) increased their degree of (unrelated) technological diversification. However, at the beginning of the 1990s, this trend has come to a halt. In the 1990s the sample companies began to refocus their technology portfolios. This observation is new: In contrast to the findings of prior empirical studies of previous periods, there has been a decrease in technological diversification by more than three per cent. Figure 4 also reveals that there exist considerable differences between the sample companies from different industries. Companies from the telecom cluster have dramatically refocused their technology portfolio in the 1983- 1997 period. Like the companies from the pharmaceutical cluster, telecom firms manage a technology portfolio that is considerably less diversified than the sample average. In contrast, companies from the automotive, diversified electric engineering and chemicals/materials clusters are engaged in a spectrum of technological activities that is high above the average.⁴

4 There also exist regional differences in the sample. Firms from the U.S. are less diversified than firms from Japan and Germany. On the average, U.S. firms began to refocus their technology portfolio already in the 1980s, whereas firms from Japan have still been broadening their activities during the 1990s.

(19)

Figure 4. Trends in (Unrelated) Technological Diversification in the Sample Differentiated by Industry Clusters

1,2

0,5 0,6 0,7 0,8 0,9 1 1,1

1993 - 1997 1988 - 1992

1983 - 1987 Divers. Electr.

Engineering

Automotive Single Entropy

Measure

Consumer Electronics

Pharma

Telecommunications Chemicals/

Materials Metal Products/

Machinery

IT/Computer Total Sample

1,2

0,5 0,6 0,7 0,8 0,9 1 1,1

1993 - 1997 1988 - 1992

1983 - 1987 Divers. Electr.

Engineering

Automotive Single Entropy

Measure

Consumer Electronics

Pharma

Telecommunications Chemicals/

Materials Metal Products/

Machinery

IT/Computer Total Sample

Measures and Trends in Product Diversification (1983-1997)

Analogue to the measure of technological diversification, the degree of product diversification is captured with the single entropy measure. This International Standard Industrial Clas- sification (ISIC) based index grasps the degree of unrelated product diversification. We differentiate between 68 industry segments that are based on 4-digit ISICs (k=1...S; S ≤ 68).

Let Pk be the share of the kth industry segment in the total sales of the firm. For each sample company then the degree of unrelated product diversification (DP) is computed as follows:

⎟⎟⎠

⎜⎜ ⎞

⎝

=

∑

⎛

= k

k

k P

P

DP 1

ln

68

1

.

The entropy measure takes into consideration two elements of diversification: the number of segments in which the firm operates and the relative importance of each of these segments in the total sales of the firm. Figure 5 visualizes the trends in product diversification differentiated by industry of the sample companies.

(20)

Figure 5. Trends in (Unrelated) Product Diversification in the Sample Differentiated by Industry Clusters

0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1

1993 - 1997 1988 - 1992

1983 - 1987

Pharma

Automotive Chemicals/Materials

Telecom- munications IT/Computer

Metal Products/

Machinery Consumer Electronics

Diversified Electrical Engineering Single Entropy

Measure

Total Sample

0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1

1993 - 1997 1988 - 1992

1983 - 1987

Pharma

Automotive Chemicals/Materials

Telecom- munications IT/Computer

Metal Products/

Machinery Consumer Electronics

Diversified Electrical Engineering Single Entropy

Measure

Total Sample

At first sight, differences between the overall trends in product diversification and technological diversification become evident: the sample companies have constantly reduced their degree of product diversification. This observation corresponds to the findings of previous empirical studies: In the 1980s and 1990s, companies have been refocusing their product portfolios. Differences in the trends by industries are even more striking. In the automotive industry cluster for example, the companies have constantly expanded their technology portfolio in the 1980s and 90s. Contrary to this trend, the automotive sample companies have refocused their product portfolios in the 1990s. Furthermore, concurrent divergences have been ascertained in the chemicals/materials cluster, whereas reverse divergences exist in the consumer electronic industry cluster. Consumer electronic firms have increased the span of their products/businesses and at the same time decreased the spectrum of their technological activities. The divergences confirm the suspicion made earlier in the paper: there exists a considerable degree of decoupling between technological diversification and product diversification, at least in some industries.

(21)

Measures and Trends in International Market Diversification (1983-1997)

In contrast to the measures of technological and product diversification there is no agree- ment among scholars in the international business domain on how to measure international diversification (Sambharya 1995). A big variety of measures of international diversification have been employed in previous empirical studies. A prominent and simple form has been a unidimensional measure of international sales as a percentage of total sales (see e.g.

Geringer et al. 1989). Others have criticized using unidimensional measures, recommending instead more sophisticated multidimensional measures (e.g. Vachani 1999 and Sullivan 1994). However, the paucity of data limits the applicability of such measures at present.

Other measures, like the ones used by Kim et al. (1989, 1993), suffer for our purposes from the fact that they combine the product dimension and international dimension of corporate diversification. With regard to these concerns and to achieve homogeneity in the measures employed, we again used an entropy approach to measure the degree of international diversification of our sample companies. International diversification may be defined as expansion across the borders of countries and (global) regions into different geographic markets. Thus, a firm’s level of international diversification is reflected by the number of different markets in which it operates and their importance to the firm as measured, for instance, by the percentage of total sales represented by each market (see also Hitt et al. 1997). The entropy measure reflects the extent of the dispersion of the firms’ sales across the major global markets. As mentioned above, we differentiate between 6 geographical market areas (m=1...R;

R ≤ 6). Let Gm be the share of the mth geographical market area in the total sales of the firm.

For each sample company then the degree of international market diversification (DG) is computed as follows:

⎟⎟⎠

⎜⎜ ⎞

⎝

=

∑

⎛

= m

m

m G

G

DG 1

ln

6

1

Again, the entropy measure takes into consideration two elements of diversification: the number of the geographical markets in which the firm operates and the relative importance of each of these markets in the total sales of the firm. Nearly all of the sample companies had already globalized their businesses to a considerable degree at the beginning of the 1980s.

Nevertheless, most of the companies have further increased their degree of international diversification during the period of investigation. Figure 6 visualizes the trends in international diversification differentiated by home country of the sample companies.

(22)

Figure 6. Trends in International Diversification in the Sample Differentiated by Home Country

0,4 0,45 0,5 0,55 0,6 0,65

Single Entropy Measure

1993 - 1997 1988 - 1992

1983 - 1987

Japan USA Total Sample

Europe Germany

0,4 0,45 0,5 0,55 0,6 0,65

Single Entropy Measure

1993 - 1997 1988 - 1992

1983 - 1987

Japan USA Total Sample

Europe Germany

Figure 6 shows that companies have continuously increased the degree of geographical diversification in the 1980s and 1990s, irrespective of their country of origin. However, figure 6 also reveals that there exist considerable differences in the degree of internationalization between the different country clusters. On average, companies headquartered in Germany lead the pack in terms of geographical diversification, although, the degree of internationalization has stagnated in the German cluster at the beginning of the 1990s as a consequence of strong domestic growth in the course of the reunification. In contrary, companies from Japan rank far below the sample average. Considerable differences also exist between the industry clusters in the sample. On the average, multinational corporations from the pharmaceutical, telecom and machinery clusters have spread their sales more equally across global markets than other companies –automotive companies, in particular, have focused their sales on a limited number of core markets. Figure 7 visualizes the trends in international diversification differentiated by the industry clusters of the sample companies.