Innovative competence, exploration and exploitation: The influence of technological diversification

Available online at www.sciencedirect.com

Research Policy 37 (2008) 492–507

Innovative competence, exploration and exploitation: The influence of technological diversification Cristina Quintana-Garc´ıa a,∗ , Carlos A. Benavides-Velasco b,1 a

Dpto. Econom´ıa y Administraci´on de Empresas, Facultad de Ciencias Econ´omicas y Empresariales, Campus El Ejido, s/n, 29071 M´alaga, Spain b Dpto. Econom´ıa y Administraci´ on de Empresas, E.T.S. Ingenieros Industriales, Campus El Ejido, s/n, 29071 M´alaga, Spain Received 9 June 2007; received in revised form 27 July 2007; accepted 10 December 2007 Available online 1 February 2008

Abstract This paper investigates how technological diversification influences the rate and specific types of innovative competence. We test a set of hypotheses in a longitudinal study of a sample of biotechnology firms. Our findings provide strong support for the premise that a diversified technology base positively affects innovative competence. Furthermore, technological diversification is found to have a stronger effect on exploratory than on exploitative innovative capability. This empirical evidence suggests that technological diversity may mitigate core rigidities and path dependencies by enhancing novel solutions that accelerate the rate of invention, especially that which departs from a firm’s past activities. © 2007 Elsevier B.V. All rights reserved. Keywords: Technological diversity; Invention; Exploration; Exploitation; Patent data

1. Introduction Over the last few decades, firms and industries have witnessed technological diversification due to increases in the complexity of products (Rosenberg, 1976; Giuri et al., 2002; Breschi et al., 2003). The range of disciplines relevant to firms’ innovative processes is expanding in both breadth (the number of relevant disciplines) and depth (their sophistication and specialization) (Wang and von Tunzelmann, 2000). For developing product innovation, firms use various scientific and technological sources, embodying different characteristics and aiming at different corners of the market (Dosi, 1988). ∗ Corresponding author. Tel.: +34 95 213 41 47; fax: +34 95 213 12 93. E-mail addresses: [email protected] (C. Quintana-Garc´ıa), [email protected] (C.A. Benavides-Velasco). 1 Tel.: +34 95 213 25 28; fax: +34 95 213 12 93.

0048-7333/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.respol.2007.12.002

Such “technological diversification” can be defined as the diversity in the knowledge system and principles underlying the nature of products and their methods of production. It is related to a corporation’s expansion of its technological competence into a broader range of technical and discipline areas (Granstrand and Oskarsson, 1994, p. 355), although such expansion does not necessarily have to be associated with product diversification (Granstrand et al., 1997; Gambardella and Torrisi, 1998; Andersen and Walsh, 2000). In general, technological diversification has only recently attracted the attention of researchers (Granstrand, 1998; Gemba and Kodama, 2001; Suzuki and Kodama, 2004). At the firm level, analyzing samples of large companies (many of them related to the information and communication technology sector), some descriptive studies have empirically demonstrated that, nowadays, multi-field competency and technological diversity is the dominant feature (Rao

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

et al., 2004; Mendoc¸a, 2006; Palmberg and Martikainen, 2006). Another group of works provides interesting insights into the relationship between technological diversification and some organizational dimensions such as size, product diversification or corporate internationalization (Cantwell and Piscitello, 2000; Cantwell and Santangelo, 2000; Piscitello, 2000; Le Bas and Patel, 2004). For example, Piscitello (2000, 2004) and Valvano and Vannoni (2003), examining patenting activities of large and leading industrial companies, confirm that technological diversification does not proceed in a random way but coherently. Moreover, coherence of corporate diversification strategies, which positively influences economic performance, is higher when firms are active in sectors sharing similar technological resources. Particularly, there has been little research on how technological diversification affects innovation performance (Nesta and Saviotti, 2005; Garcia-Vega, 2006). These studies find that diversification of the technology base enhances R&D intensity and the number of patents. The aim of this paper is to advance in the knowledge of the impact of technological diversity on innovation. This investigation is novel as it explores, in a particular high-technology sector, the influence of a diversified technology portfolio on specific types of innovative capabilities: exploitation and exploration. Exploitative innovation is based on intensive search, which means experimentation along an existing knowledge dimension. Exploration is rooted in extensive search that pursues potential new knowledge (March, 1991). Exploitative innovation improves the methods or materials used to achieve the firm’s objectives of profitability and satisfying customer needs. In contrast, exploratory technological innovation involves novel methods or materials that are derived either from a completely different knowledge base or from a recombination of parts of the firms’ established knowledge base with a new stream of knowledge (Freeman and Soete, 1997). Based on evolutionary theory and organizational learning research (Dosi, 1982; Nelson and Winter, 1982; March, 1991; Levinthal and March, 1993), we test a set of hypotheses about how technological diversification enables organizations to improve their innovative capacity and influences exploratory and exploitative inventions in a different way. We address this issue because is relevant for theory and practice. From the technology management perspective, compared to concentration, developing a diverse R&D portfolio implies integrating new technical staff, assimilating new technological knowledge (Lee and Allen, 1982),

493

and, in general, a more complex process of building research competences. Consequently, technological diversification constitutes a critical decision especially in high-technology industries where research projects demand large investment and developing successful innovations can take several years. On the other hand, evidence to improve our understanding of how a diversified technology base influences particular types of innovative capabilities is relevant for organizational learning theory. Search with high scope enriches the firm’s knowledge base by adding distinctive new variations (Fleming, 2001; Katila and Ahuja, 2002). However, extremely high levels of technological diversity may damage the desired balance or combination between exploitation and exploration (March, 1991; Levinthal and March, 1993), and hence, the firm’s capacity for sustaining its competitive position through mixed processes of knowledge creation, adaptation and consolidation. The paper is in four sections. First, we present the theoretical framework. After a section on research methodology to test the hypotheses, the results from a sample of biotechnology companies in a longitudinal study are offered. Finally, the findings are discussed and areas of future research are proposed. 2. Theory and hypotheses The complexity of many modern innovations necessitates a pooling and integration of multiple strands of knowledge (Subramaniam and Youndt, 2005). It is suggested that due to shortening product life cycles and expanding numbers of technology options, the ability to effectively integrate technology is now more important (especially in some sectors such as the computer industry) than the ability to develop new technology (Iansiti and West, 1997; Christensen, 2002). Technological diversification can be beneficial to organizations for a number of reasons (Granstrand et al., 1997; Patel and Pavitt, 1997). For example, organizations can benefit from introducing new technology into existing products and systems to improve performance and develop new functionalities. Or, organizations can take advantage of the continuing relevance of old technologies by combining them with pertinent and necessary emerging technological opportunities. Furthermore, organizations can benefit by coordinating innovation in core products with complementary changes in the production system and supply chain. Hence, technological diversity may influence firms’ capacity for combining and recombining their stock of existing knowledge with new components that result in new breakthroughs. Moreover, as such diversifica-

494

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

tion favors new combinations and transforms prevailing knowledge, the likelihood of developing radical innovative capabilities may be higher (Abernathy and Clark, 1985). Despite the importance of researching these relationships for learning and decision-making processes related to the organization of R&D activities, there is little empirical evidence to show how differences in technology selection affect invention. In the following sections, we develop hypotheses about the effects on both the scale and the types of innovative capacity of firms’ technological knowledge diversification. 2.1. Effect of technological diversification on innovative competence Literature on evolutionary theory and technological trajectories (Rosenberg, 1969; Dosi, 1982; Nelson and Winter, 1982) suggests that maintaining positions in a diverse range of technologies (e.g. maintaining people with different kinds of background) is essential. Since many innovations are designed to solve unrelated problems, companies that are more technologically diversified, capture more opportunities and technical possibilities and so benefit largely from their own research activities (Nelson, 1959). In an emerging stage of regime, a set of technological possibilities consists of a number of quite different classes of technology. A firm’s R&D activities may focus on one class of technology with little attention paid to others. If technological advance is finally achieved through the ignored technological field, core competences of the firm might become obsolete. Thus, there is a “selection” effect because the more approaches there are to a given technological objective, the greater the contribution to technical advance of the approach that the market ultimately selects will be (Dosi, 1982; Nelson and Winter, 1982; Cohen, 1995). Organizational learning theory also suggests the benefits of a diverse knowledge base. To diversify the technology base in order to search for complementarities and novel solutions accelerates the rate of invention. On the contrary, the repeated application of a particular set of technologies eventually exhausts the set of potential combinations (Kim and Kogut, 1996). As Levinthal and March (1993, p. 103) affirm, it is valuable to create “inventories of competencies” that might be employed later without knowing precisely what future demands will be. This will positively influence the accumulation of the absorptive capacity that permits the firm to predict the nature and commercial potential of technological advances and to exploit technological opportunities (Cohen and Levinthal, 1990). If firms lack detailed knowledge of a particular set of tech-

nologies in some initial period, they will not be aware of the significance of new technological opportunities in related areas. Although such opportunities are recognized, the lack of sufficient background knowledge damages the firm’s ability to capitalize on new developments to generate innovations. Thus, technological diversification may play a preventive role against core rigidities (Leonard-Barton, 1992) by generating and renovating technological trajectories and taking advantage of cross-fertilization effects between different technologies (Granstrand, 1998; Suzuki and Kodama, 2004). Adding new knowledge to the firm’s repertoire is important for its continuity and the mitigation of path dependencies. Supporting these arguments, empirical evidence from the robotics industry shows that there is a linear and positive relationship between search scope and product innovation (Katila and Ahuja, 2002). Ahuja and Lampert (2001) demonstrate, for the chemicals industry, that experimenting with novel, emerging, and pioneering technologies is a way for organizations to overcome core rigidities and is associated with the subsequent number of inventions. A study by Nicholls-Nixon and Woo (2003) examines the relationship between breadth of technological knowledge and technical output (number of products and patents) in a sample of established pharmaceutical companies during a period corresponding to the emerging phase of biotechnology as a paradigm. They find no association between the variables, although they admit that the high level of aggregation at which technologies were counted (seven very basic classes) was insufficient to obtain adequate variance between the sample firms. Studies in the same industry suggest that the ability to access new knowledge and integrate it flexibly across disciplinary classes within the organization benefits the effectiveness of product or process development (Henderson and Cockburn, 1996; Cockburn and Henderson, 2001). Research programs initiated within a more diverse development effort and technological knowledge is significantly more likely to result in innovations than others initiated within a more narrowly focused effort. Hence, the performance advantage in R&D activities appears to lie in economies of scope rather than in economies of scale. Analyzing U.S. biotechnology patents applied for between 1990 and 1998, Nesta and Saviotti (2005) state that the scope and coherence of the knowledge base contribute positively to innovation performance (estimated by the number of patent applications). More recently, based on a panel data of 544 European firms, Garcia-Vega (2006) provides more evidence on the positive relationship between technological diversity and innovation at

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

the firm level (measured by R&D intensity and number of patents). What an organization knows how to do is a function of what it learned previously (Pisano, 2002). We hypothesize that accumulating technological knowledge with a diversified base, enables companies to improve their absorptive capacity and obtain more innovations in the future. In contrast, research concentrated on or closely related to core competences can produce lock-in effects. Hypothesis 1. The greater the firm’s technological diversification, the higher its subsequent innovative competence. Technological diversification is, however, not costless. A firm requires large investment and a number of years to integrate and assimilate new technological knowledge (Lee and Allen, 1982). This process involves several tasks such as setting up new types of manufacturing facilities and operations, scanning for promising new technologies, assessing the feasibility of such technologies, recruiting specialists in new technical areas, building strategic alliances and networks, integrating new technologies with the firm’s existing technologies, etc. (Danneels, 2002). Moreover, firms with a diversified technology portfolio are likely to involve high integration, coordination and communication costs (Granstrand, 1998; Ahuja and Lampert, 2001). In consequence, high technology firms, especially those that are small and medium sized, have to deal with the trade-off between higher innovative competence and greater costs produced by diversification. 2.2. Technological diversity, exploration and exploitation A well-known classification of innovative competences and search modes is that which distinguishes between exploration and exploitation (March, 1991). Exploratory invention requires distant search and a departure from the firm’s store of current skills and capabilities. Conversely, exploitation leverages a firm’s existing knowledge. That is, incremental innovative capabilities draw upon reinforced prevailing knowledge, whereas radical innovative competences draw upon transformed prevailing knowledge (Subramaniam and Youndt, 2005). Accordingly, technological diversification influences the rate of invention output but its impact might be stronger on exploration than on exploitation. Diversity in the knowledge system might enhance incremental innovation. A central assumption in evolutionary theory (Nelson and Winter, 1982; Sahal, 1985) is

495

that of “local search”, which means that firms focus on similar technology. As a result, they become more expert in their current domains. By enhancing current technological combinations, such new technologies could potentially improve products and processes, thereby leading to incremental innovation and indicating that a change in routine has taken place. However, Patel and Pavitt (1997) suggest that technological variety is a necessary characteristic of revolutionary rather than normal technological change. They consider variety essential for effective experimentation and choice under conditions of great uncertainty. Likewise, it is assumed that a more sustainable competitive advantage relies heavily on the ability to move beyond local search and to reconfigure its knowledge (Kogut and Zander, 1992; Rosenkopf and Nerkar, 2001). Access and exposure to a variety of new and alternative technological knowledge domains influence a firm’s propensity to transform knowledge and find new ways in which existing problems can be solved. Hence, it is considered that innovative asset creation by developing competences such as new technological fields promotes the capacity to produce more radical product and process innovation (Christensen, 2002). Through a case study, Danneels (2002) finds that for exploitative projects, firms can use their technological and market knowledge, but that successful exploratory innovation generally is based on a broad scope of technological search, that is, it occurs when the learning activity adds a new competence for the firm. Likewise, some works suggest that innovative potential can be managed by controlling the antecedent conditions, in particular by selecting a broad range of scientific specialties (Ettlie et al., 1984; Dewar and Dutton, 1986; Cardinal, 2001). The availability of multiple scientific areas of experience aids R&D professionals in developing new knowledge and expanding existing knowledge bases through the crossfertilization of ideas. Through a transversal analysis of a sample of pharmaceutical companies, Cardinal (2001) finds that scientific diversity is positively related to the likelihood of radical innovation but to a lesser extent than of incremental innovation. However, the rest of the empirical evidence shows that a diversified technology system (measured by the number of persons in different technical specialties) has a stronger effect on radical than on incremental innovation (Ettlie et al., 1984; Dewar and Dutton, 1986). Ettlie et al. (1984) studied radical and incremental process innovations in the food-processing industry. Analyzing research teams, they showed that adoption of radical innovation occurred more frequently in firms with an aggressive technology policy and a critical mass of technical specialists representative of diverse knowledge resources. Dewar and Dutton (1986)

496

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

examined process innovations in the footwear industry and reported that a broad array of disciplines benefits both radical and incremental innovation although the impact is stronger on the former. Similarly, a longitudinal research focused on the pharmaceutical industry (Wuyts et al., 2004) found that the technological diversity of a firm’s alliance portfolio positively influences both radical and incremental product innovation; however, in the model estimation, it is possible to observe a higher impact of such diversity on radical innovative outcomes. Based on this evidence and the theoretical framework, we hypothesize that technological diversification from several sources benefits the scale of both exploratory and incremental innovation, but its effect on the former is larger. Complexity and knowledge scope should be less important for incremental innovation because adoption of this type requires minor changes to existing knowledge resources in the organization. Hypothesis 2. The greater the firm’s technological diversification, the higher its subsequent exploitative innovative competence. Hypothesis 3. The greater the firm’s technological diversification, the higher its subsequent exploratory innovative competence. Hypothesis 4. Technological diversification has a larger positive impact on exploratory innovative competence than on exploitative innovative competence. 3. Methodology 3.1. Research setting Innovative competence is strongly sector-specific. The knowledge base and learning processes related to innovation differ across sectoral systems of innovation (Malerba, 2002). Knowledge differs across sectors in terms of domains, which refer to the specific scientific and technological fields at the base of innovative activities (Dosi, 1988). Science-based firms are heavily dependent on diverse knowledge, skills and techniques. Their main tasks of innovation strategy are to monitor and exploit advances emerging from basic research, obtaining as a result a relatively high proportion of their own process technology, as well as a high proportion of product innovations that are used in other sectors (Pavitt, 1984; Tidd et al., 2001). Particularly, our study focuses on the biotechnology sector to test the predicted effects of technological

knowledge diversity on the rate and types of innovative capability. Some factors and features justify the importance of technological diversification and continuous innovativeness in this sector. Biotechnology itself is a revolutionary technology, and rapid and radical technological innovation within biotechnology threatens to render even current products obsolete within a relatively short time. Therefore, biotechnology firms can sustain a competitive advantage only by continuous innovation that results in valuable and patentable products (Powell et al., 1996). Furthermore, in the biotechnology sectoral system, a wide variety of science and engineering fields play relevant roles in renewing the search space; such diversity demands a full range of systemic interaction mechanisms with science and scientific institutions to foster the innovation process (Cooke, 2002; Malerba, 2005). Nowadays, genetic engineering and biotechnology rely on a wide knowledge system and diverse competences (molecular biology, microbiology, enzymology, cell biology, etc.) each of which has its own dynamics in relation with other domains. The integration of novel technologies in the firm’s knowledge base is likely to induce profound changes in the prevalence of certain bodies of knowledge, the relationships between them and the way knowledge is organized and used in the innovation process (Nesta and Dibiaggio, 2003). However, this variety of technologies has been increasing over time. In the beginning, in the mid-1970s, compared with other industries (e.g. chemical synthesis, pharmaceutical sector) the biotechnology process was a regime characterized by relatively immature theory and thin practical experience. This weak knowledge base underlying biotechnology production meant that the ability to bring together, integrate, and generate relevant knowledge was limited (Pisano, 1994). From the mid1970s on, however, substantial advances in fields such as physiology, enzymology, cell biology, bioengineering, and so on led to great progress in the ability to understand the mechanism of action of some existing drugs and had a profound impact on the process of discovery of new drugs. These advances explain that modern drug research relies upon the input of scientists skilled in a very wide range of disciplines, increasing the importance of the scope economies (Henderson and Cockburn, 1996; Cockburn and Henderson, 2001; Malerba and Orsenigo, 2002). The youth of the biotechnology companies makes them an excellent venue in which to study the organizational processes influencing the initial accumulation of capabilities (Pisano, 2002, p. 129). Particularly, this feature together with the increasing diversity of scientific

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

and technical knowledge experienced by the biotechnology sector justifies its selection as a suitable research setting to contrast the effects of technological diversification on innovative competence over the 1976–2002 time period. 3.2. Patent data We use invention as a proxy indicator of innovative competence, which refers to the discovery of new methods or materials (Freeman and Soete, 1997) rather than its subsequent commercialization. Studying the determinants of inventions is of importance because they represent valuable sources of competitive advantage (Barney, 1991) and the “creation of opportunities” (Ahuja and Lampert, 2001). Patent data represent the main source of information for our study. Some limitations of patent data are (Griliches, 1990; Silverman, 1999): (a) part of technical knowledge may remain unpatented either because it is unpatentable or because a firm may choose not to patent to keep in secrecy; (b) differences in the comprehensiveness of patenting may exist across firms, industries and time2 ; (c) patents differ greatly in their technical and economic significance; (d) the extensive diversification of many firms and also the various merger waves create severe technical problems in using the patent data at the individual corporate level.3 Nevertheless, patent data are very useful to analyze the process of innovation and technical change. They are by definition related to inventiveness, and they are based on an objective and only slowly changing standard. Several studies (see Griliches, 1990) have demonstrated a strong relationship between patent numbers and R&D expenditure at the cross-sectional level, across firms and industries, implying that patents are a good indicator of differences in innovative competence across different firms. The fact that a firm applies for a patent in a given technological field means that it is at, or close to, the technological frontier and has advanced technological competencies (Breschi et al., 2003). In particular, some authors who are critical of the use of patents admit that this measure can be an appropriate indicator in the context of many high-tech sectors (e.g. Mansfield, 1986). Among the data available in patents, classification codes identify the type of technol-

2

The first two problems can be taken care of by industry dummy variables, or by limiting the analysis to a particular sector or industry. 3 See Section 3.3 that explains how we deal with this problem in the empirical study.

497

ogy embodied in the invention. Subclasses can be used to observe indirectly the process of recombinant search and learning (Fleming, 2001). Compared to the exclusive use of R&D expenditure, patents offer richer information on the particular range of technological strengths possessed by a firm (Silverman, 1999). Our measure of technological diversification goes beyond existing research, which partially captures this concept assessing the variety of technical specialists or partners. As invention, patent can be considered the end result of a research project. Thus, patent codes may be used to observe more comprehensibly a firm’s technological diversity resulting from any kind of source. Since this paper deals with a sample of US companies, we use the United States Patent and Trademark Office (USPTO) as the main information source. The patent office usually assigns each patent into multiple subclasses; it also establishes and updates new classes and subclasses each year, as technology changes. This retrospective updating enables historical consistency in the measurements across time (Fleming, 2001). 3.3. Sample The level of analysis is the firm-year. We have aggregated the set of patents granted by the firm in a year. For the year of aggregation, we used the application date of the granted patents to consider the time when technological knowledge was created by the firm. The aggregation is worthwhile because it gives an overall picture of exploration strategies that predominate in the firm (Rosenkopf and Nerkar, 2001). The research sample of US dedicated biotechnology firms (DBFs) was drawn from the USPTO. Biotechnology patents often fall into the codes 435 and 800 of the U.S. Patent Office and Trademark’s Classification. This Office offers a list of companies and other organizations (universities, research labs, etc.) that have at least five patents granted during the period 1969–2003 in all codes of the classification. We conducted searches in CorpTech Explore, D&B International Million Dollar Database, and Worldscope to identify companies included in the lists related to codes 435 and 800 that are fully dedicated to biotechnology activities. This process yielded a list of 525 DBFs; this sampling frame has an international scope. Historical information from Mergerstat, CorpTech Explore and Worldscope allowed us to identify inactive firms due to failure or merger and acquisition transactions. These companies were not excluded in our sample to avoid survival bias. Besides, through the international directory Corporate Affiliations we came across companies that actually were subsidiaries of others. This

498

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

information together with that related to mergers and acquisitions permitted us to aggregate patent data at the corporate level, that is, company structure was rebuilt from the patent units. We gathered yearly data of control variables during the period 1976–2002 for 115 USparent DBFs. We decided to include all the companies in the final sample. There were included both private and publicly traded companies. Moreover, the sample contains firms of different sizes. The collection of data related to the dependent variables and main independent variable was conducted in the first semester of 2005. At that time, the number of patents obtained by the companies after 2002 was very small; most of the patent applications remained under consideration to be granted. This fact justified closing the period of study in 2002. Furthermore, we use two lagged independent and control variables; that is why the first 2 years’ observations of each dependent variable are lost. As a result, the total number of observations is 985. 3.4. Measures

reflects a firm’s willingness to explore technology spaces and adopt pioneering or unprecedented approaches in its innovation strategy. “Exploitative innovative competence” is the number of patents that include one or more citations or self-citations and are thus close to existing innovations that have been produced by the firm or other organizations. 3.4.3. Independent variable: technological diversification There are alternative ways to measure technological diversity. We decided to use the Herfindahl index of diversification (Berry, 1975) which is derived from the Herfindahl–Hirschman Index (HHI). The HHI index is conventionally used to approximate industry concentration, and it is becoming popular to measure technological diversification (e.g. Garcia-Vega, 2006; Mendoc¸a, 2006; Palmberg and Martikainen, 2006). The Herfindahl index of diversification can be expressed as follows:  D = 1 − HHI = 1 − Pi2 i

3.4.1. Innovative competence This dependent variable is built through the number of patents granted by the firm in a year. To analyze the effect of knowledge diversity on this variable, we have used the application date of the granted patents. So doing, we take into account the time when innovation activity was carried out. We conducted searches in the online database of U.S. Patent and Trademark Office which includes all U.S. patents from 1976 through the present. We collected for each firm the total number of patents assigned into any technology class over the 1976–2002 time period. 3.4.2. Exploratory and exploitative innovative competence We classified the total number of patents granted for each firm into two groups: exploratory and exploitative inventions. Citation data contained in patents were used to perform this differentiation. Exploratory innovative competence represents the ability to perform extensive searches that result in novel methods or materials technologically distant from existing innovations. In contrast, exploitation improves methods or materials. Following the definition provided by Ahuja and Lampert (2001), “exploratory innovative competence” was measured by counting the patents that depart completely from prior firm knowledge base and represent new knowledge to other firms and to the market. That is, this dependent variable is the number of granted patents by the firm in a year that cite no other patents because they indicate high levels of creativity. The creation of these patents

where Pi denotes the proportion in a firm of patents in technical field i. The index equals zero when a firm researches only in a single technology, and it is close to one when the firm spreads its research activity over a broad technological knowledge base. The Herfindahl index of diversification has the distinct advantage of depending only on the distribution of patents of a firm. Any change in the distribution of patents of all other companies has no influence on this diversification measure (Rao et al., 2004). Additionally, it allocates less weight to patent classes that are less significant in a firm’s technology portfolio. Other common measures of technological diversification are the coefficient of variation (CV) of the Revealed Technology Advantage (RTA) index,4 and the Jacquemin-Berry entropy measure of diversification.5 There is a strict relationship between Herfindahl index and the CV. The relationship is Herfindahl index = (CV2 + 1)/N (Hart, 1971; Cantwell and Piscitello, 2000; Cantwell and Santangelo, 2000), thus both measures share similar advantages. Technological diversification has been measured using patent classifications that distinguish, for example, 30 types (OECD classification), 34 technical fields 4

RTA index is defined as the firm’s share of patenting in the field, divided by the firm’s aggregate share of patenting in all fields (Patel and Pavitt, 1997). 5 The formulation of the entropy measure of diversification is the  following: D = i Pi ln(1/Pi ) where Pi is the proportion in a firm of patents in technical field i.

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

(SPRU taxonomy), etc. We consider these taxonomies to be global, quite similar to sectoral categories, consequently technological diversity becomes a similar concept to business or market diversification; thus, it does not allow in-depth explorations of how interactions of knowledge pieces impact on the innovation process. From an organizational point of view, a more useful study would have to consider a more detailed classification using sub-fields of technologies. We followed the threedigit USPTO’s classification, which distinguishes over 400 technology classes. Consistent with previous studies (Argyres, 1996), patents assigned to more than one technical field have been treated as different applications in order to better capture firm-level technical diversity. The Herfindahl index of diversification has been measured taking into account the subclasses of the total of patents accumulated by a firm. Analyzing the sample of 115 companies, it is possible to observe that only one company patents in a single technology class (it has a Herfindahl index of diversification of 0). The rest of the companies patent in at least 6 different fields, and the maximum value is 62 technology classes. Thus, it is appropriate to use the selected index in studying our sample of DBFs. We assume that the accumulated variety of knowledge influences the subsequent capacity for performing innovations. Thus, “technological diversification” is a lagged variable and it is measured for the period 1976–2001. It also allows a reduction of concerns about reverse causality. 3.4.4. Control variables Numerous factors beyond technological diversity may influence innovative competence. We included two lagged control variables in the panel data related to within-firm changes over time such as R&D intensity and stock of patents. The intensity of research effort is used as a proxy for the level of technological opportunity in several studies (Benner and Tushman, 2002; Valvano and Vannoni, 2003). We estimate R&D intensity through the ratio of annual research and development expense (millions of dollars) over annual revenues (millions of dollars) multiplied by 100. We obtained R&D expenditure and revenue data at the firm level from Compustat, Worldscope and Investext Plus. Codified forms of technological capabilities are complementary to non-codified knowledge (partly captured by R&D intensity). Thus, we introduce the “stock of patents” as a control variable that is related to the cumulative condition of learning. Prior accumulated knowledge permits the firm to better understand and evaluate the sig-

499

nificance of technological advances, and hence it may benefit the rate of innovation. The total accumulated patent stock can also be used as a proxy for the firm size (Cantwell and Santangelo, 2000). Moreover, as this variable captures the total number of exploratory and exploitative inventions in t − 1, it is an additional control to address the threat of potential unobserved firm heterogeneity. Table 1 lists and describes all variables used in our study. 3.5. Statistical method The dependent variables of the study count the number of total, exploitative and exploratory patents. They take non-negative integer values, exhibit overdispersion and are longitudinal, including observations per firm and year. In this case the assumptions of the linear regression model related to homoskedasticity and normally distributed errors are violated and count models are more appropriate, specially the Poisson model. However, the Poisson model assumes that the mean and variance of the observed distribution are equal (Long, 1997). To adjust for overdispersion, we used the negative binomial model, a generalization of the Poisson model where the assumption of equal mean and variance is relaxed (Hausman et al., 1984). To control for firm heterogeneity, we used the generalized estimating equation (GEE) regression method. The GEE algorithm accounts for correlation between records within the same cluster (data collected about the same firm during successive periods of time) thus providing improved standard error estimates (Liang and Zeger, 1986; Zorn, 2001). The GEE approach is less computationally intensive than either fixed effects or random effects. Therefore, it often proves less subject to instability and convergence problems. A dummy variable for each year was included to control for factors that are the same for all cross-sectional units but vary over time (e.g. economic magnitudes). In order to reduce concerns about unobserved heterogeneity, we used an alternative measure of technological diversification to test the validity of its effect on innovative competence. Since we have panel data, we tried to understand what is happening to an individual firm’s innovation over time as it has more diversification in its own technological base. We created a dummy variable “diversification increase” that takes the value “1” when the Herfindahl index of diversification has increased in the previous year and 0 otherwise. That is, we study how change in the diversity of a firm’s technology base in prior years triggers change in the subsequent rates of patenting for that firm (see models 3, 7 and 11). The construction of such a vari-

500

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

Table 1 Variables and measures Variables

Type

Method used to measure the variables

Source of the data

Innovative competence

Dependent

Database of granted patents of the United States Patent and Trademark Office (USPTO)

Exploitative innovative competence

Dependent

Exploratory innovative competence

Dependent

Technological diversification (t − 1) and (t − 2)

Independent

Diversification increase (t − 1) and (t − 2)

Independent

Diversification increase Endog (t − 1) and (t − 2)

Independent

R&D intensity (t − 1) and (t − 2)

Control

Stock of patents (t − 1) and (t − 2)

Control

Number of total patents granted by the firm in a year. For the year of aggregation, it is used the application date of the granted patents Number of patents granted by the firm in a year that include one or more citations or self-citations. For the year of aggregation, it is used the application date of the granted patents Number of patents granted by the firm in a year that cite no other patents. For the year of aggregation, it is used the application date of the granted patents Herfindahl index of diversification taking into account the subclasses of the total patents accumulated by a firm Dummy variable that takes the value ‘1’ when the firm’s Herfindahl index of diversification has increased in the previous year and ‘0’ otherwise Observed “diversification increase” predicted by using a logit regression which includes two lags of “R&D expenditure” as instrumental variables Ratio of annual research and development expense (millions of dollars) over annual revenues (millions of dollars) multiply by 100 Total number of patents granted by the firm

R&D expenditure (t − 2) and (t − 3)

Instrumental

Annual research and development expense (millions of dollars)

able implies losing another year’s observation per firm, decreasing the total number of observations to 844. Moreover, we have implemented two methods to correct potential “endogeneity” in the empirical test. First of all, the key explanatory variables (technological diversification, diversification increase and stock of patents) have been lagged for two periods. Secondly, we have estimated the relationship with techniques that take the problem of endogeneity into account (models 4, 8 and 12). The two-stage Heckman estimation procedure is a common technique to correct endogeneity. This method requires full parametric specificity, thus it

To aggregate patent data at the corporate level (including all patenting units of the company) we conducted searches in Corporate Affiliations, CorpTech Explore, D&B International Million Dollar Database, Mergerstat and Worldscope

Compustat, Worldscope and Investext Plus

Database of granted patents of the United States Patent and Trademark Office (USPTO) To aggregate patent data at the corporate level (including all patenting units of the company) we conducted searches in Corporate Affiliations, CorpTech Explore, D&B International Million Dollar Database, Mergerstat and Worldscope Compustat, Worldscope and Investext Plus

is not suitable for count data models. Using a simultaneous equations model where one dependent variable is a count, and which has a binary variable as endogenous regressor can provide consistent estimates for the parameters (Windmeijer and Santos-Silva, 1997; Terza, 1998; Romeu-Santana and Vera-Hern´andez, 2005). This is a two-stage method analog to the popular Heckman estimator that avoids the computational requirements of a full information maximum likelihood (FIML). Thus, we have considered as the endogenous regressor the dummy variable “diversification increase” (d). The specification of a simultaneous model with the count

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

dependent variable and the binary endogenous regressor is: y = e(d∗α+xi β) + εi

(1)

d ∗ = zi δ + ωi

(2)

where εi is the heterogeneity term, y represents the dependent variables (innovative competence, exploitative innovative competence and exploratory innovative competence), zi are the instruments that explain “diversification increase”, and d* (diversification increase Endog) is an unobserved latent variable. The observed diversification increase (d) is modeled such that d=1 d=0

if zi δ + ωi > 0 otherwise

We ran the models estimation using the STATA statistical software. 4. Results Table 2 shows descriptive statistics and correlations for the variables of the study. Table 3 provides the regression results using GEE negative binomial regression analysis with levels of significance reported for two-tailed tests. Models 1, 5 and 9 include only the control variables in order to provide a baseline. It is noteworthy that adding the independent variable accounting for technological diversification in models 2, 6 and 10 significantly increases goodness-of-fit when compared to models 1, 5 and 9, respectively. Models 3, 7 and 11 report the results obtained using two lags of “diversification increase” as the main independent variable. Models 4, 8 and 12 include two lags of the variable “diversification increase Endog” instead of “diversification increase”. In the first stage, we estimated δ by logit (see Table 4) to predict the probability of diversification increase. We used as instrumental variables two lags of R&D expenditure (annual research and development expense expressed in millions of dollars). The second stage was to substitute the observed diversification increase outcomes (d) into (1) to estimate the model by negative binomial regression using the GEE algorithm. The corrected results reject exogeneity of d because the coefficients are significantly different from zero. Hypothesis 1 proposed that technological diversification has a positive effect on a firm’s innovative competence. Models 1, 2, 3 and 4 show results of regressions to analyze this hypothesis. In models 2 and 3

501

as well as in model 4, which deals with endogeneity, the significant positive coefficients of the lagged variables “technological diversification”, “diversification increase” and “diversification increase Endog” demonstrate that such diversity benefits innovative competence, supporting Hypothesis 1. Hypotheses 2–4 propose that although technological diversification positively influences both exploitative and exploratory innovative competence, the effect on the latter is higher because of its capacity to enhance new ideas and novel combinations to solve technologically complex problems. These hypotheses are strongly supported. The coefficients for the independent variables in models 6 and 10 are positive and significant, but the value of the coefficients in model 10 (β1 = 8.736, β2 = 3.622) are higher than in model 6 (β1 = 2.954, β2 = 1.064). These differences are statistically significant. A Wald test of the equality of the technological diversification coefficients yields, for β1 , a χ2 of 190.91 with one degree of freedom, which can be rejected at the one-percent level, and for β2 , a χ2 of 77.74 (p < 0.001) with one degree of freedom. The results reported in models 7 and 11, and in models 8 and 12 reinforce the previous findings. A Wald test for equality of “diversification increase” coefficients in models 7 and 11 produces, for the coefficients of the first lag, a χ2 of 85.46 (p < 0.001) with one degree of freedom, and for the coefficients of the second lag, a χ2 of 78.63 (p < 0.001) with one degree of freedom. Finally, the differences are again statistically significant in models 8 and 12, yielding a χ2 of 36.80 (p < 0.001) with one degree of freedom for the first lag of “diversification increase Endog”, and a χ2 of 15.80 (p < 0.001) with one degree of freedom for the second lag. Technologically diversified companies innovate at a higher rate and largely create knowledge new to the firm and to the industry. That is, they have a better ability to generate novel solutions that completely depart from the firm’s previous knowledge base rather than consisting of extensions of their established innovative domains. Hence, to increase the variety of the technology base may avoid core rigidities by renovating organizational routines and competences that underlie the production of innovations. The effects of the control variables were as expected. R&D intensity was found significant in all models shown in Table 3. The stock of patents as indicator of the firm’s prior experience has the expected positive sign. It seems to aid in the development of both exploitative and exploratory inventions. Considering the stock of patents as a proxy of company size, it is possible to observe that large firms patent at a higher rate than smaller ones. Moreover, the correlation analysis presented in Table 2 suggests that size is positively related to the degree of

502

Table 2 Descriptive statistics and correlations Variables

*

p < 0.05.

S.D.

Minimum

Maximum

1

2

3

4

5

6

7

8

9

10

11

12

13

14

9.34

21.33

0

244

1.00

7.72

15.38

0

233

0.87*

1.00

1.62

9.77

0

169

0.70*

0.31*

1.00

0.64

0.15

0

0.98

0.34*

0.31*

0.38*

1.00

0.64

0.16

0

0.98

0.33*

0.31*

0.36*

0.82*

1.00

0.50

0.50

0

1

0.10*

0.11*

0.15*

0.75*

0.75*

1.00

0.50

0.50

0

1

0.12*

0.12*

0.17*

0.79*

0.82*

0.75*

1.00

0.64

0.47

0

1

0.10*

0.07*

0.13*

0.69*

0.59*

1.00

0.58*

1.00

0.64

0.47

0

1

0.08*

0.06*

0.11*

0.66*

0.62*

0.63*

1.00

0.61*

1.00

47.68

73.95

0.02

279.08

0.21*

0.25*

0.20*

0.10*

0.10*

0.09*

0.10*

0.02

0.08*

1.00

49.77

76.81

0.02

279.08

0.20*

0.23*

0.19*

0.09*

0.10*

0.09*

0.09*

0.03

0.07*

0.73*

1.00

47.39

104.87

0

760

0.44*

0.47*

0.22*

0.28*

0.32*

0.07*

0.07*

0.03

0.01

0.06

0.06

1.00

42.09

96.24

0

733

0.32*

0.37*

0.12*

0.24*

0.28*

0.06*

0.09*

0.06*

0.04

0.06

0.05

0.89*

1.00

30.69

74.35

0.01

875.1

0.28*

0.33*

0.26*

0.27*

0.29*

0.12*

0.15*

0.12*

0.12*

0.35*

0.33*

0.56*

0.55*

1.00

27.25

65.01

0.01

822.8

0.26*

0.31*

0.23*

0.27*

0.30*

0.10*

0.15*

0.10*

0.15*

0.37*

0.35*

0.57*

0.57*

0.91*

15

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

1. Innovative competence 2. Exploitative innovative competence 3. Exploratory innovative competence 4. Technological diversification (t − 1) 5. Technological diversification (t − 2) 6. Diversification increase (t − 1) 7. Diversification increase (t − 2) 8. Diversification increase Endog (t − 1) 9. Diversification increase Endog (t − 2) 10. R&D intensity (t − 1) 11. R&D intensity (t − 2) 12. Stock of patents (t − 1) 13. Stock of patents (t − 2) 14. R&D expenditure (t − 2) 15. R&D expenditure (t − 3)

Mean

1.00

Table 3 Regression results for effect of technological diversification on innovative competence Innovative competence

Technological diversification (t − 1) Technological diversification (t − 2) Diversification increase (t − 1) Diversification increase (t − 2) Diversification increase Endog (t − 1) Diversification increase Endog (t − 2) R&D intensity (t − 1) R&D intensity (t − 2) Stock of patents (t − 1) Stock of patents (t − 2) Annual dummies (25 years in models 1, 5 and 9, and 23 years in the rest of models) Constant N (number of firm-years) Number of firms Wald χ2

0.014** (0.004) 0.014** (0.004) 0.013*** (0.001) 0.011*** (0.001) 11 years*

Model 2

Exploitative innovative competence Model 3

Model 4

Model 5

Model 6

Exploratory innovative competence Model 7

Model 8

Model 9

Model 10

3.118*** (0.421)

2.954*** (0.425)

8.736*** (0.929)

1.082*** (0.319)

1.064** (0.324)

3.622*** (0.811)

0.009* (0.001) 0.009* (0.001) 0.013*** (0.001) 0.012*** (0.001) 10 years*

Model 11

0.292*** (0.063)

0.183** (0.053)

0.678** (0.079)

0.438*** (0.064)

0.269*** (0.054)

0.753*** (0.080)

0.023*** 0.021*** 0.017*** 0.018*** 7 years*

(0.005) (0.003) (0.001) (0.001)

Model 12

0.277** (0.056)

0.213*** (0.055)

0.447*** (0.090)

0.219** (0.057)

0.174** (0.056)

0.317*** (0.094)

0.019** (0.005) 0.019** (0.005) 0.018*** (0.001) 0.017*** (0.001) 8 years*

0.015** (0.004) 0.014** (0.003) 0.014*** (0.001) 0.013*** (0.001) 11 years*

0.014* (0.004) 0.013* (0.004) 0.012*** (0.001) 0.012*** (0.001) 9 years*

0.023*** (0.005) 0.022*** (0.005) 0.013*** (0.001) 0.013*** (0.001) 11 years*

0.015* (0.009) 0.016* (0.009) 0.017*** (0.001) 0.016*** (0.001) 9 years*

0.010** (0.003) 0.009** (0.001) 0.011*** (0.001) 0.012*** (0.002) 8 years*

0.011* (0.005) 0.012* (0.005) 0.012*** (0.003) 0.010*** (0.002) 8 years*

0.021*** 0.020*** 0.012*** 0.015*** 9 years*

(0.006) (0.006) (0.002) (0.002)

0.016* (0.004) 0.015* (0.004) 0.016*** (0.001) 0.015*** (0.002) 7 years*

1.179*** (0.097) −1.305*** (0.271) 1.179*** (0.097) 0.726*** (0.133) 0.904*** (0.126) −1.323*** (0.274) 0.698*** (0.137) 0.578*** (0.134) −1.115*** (0.205) −9.193*** (0.606) −0.564*** (0.185) −1.189*** (0.217) 1212 985 851 851 1212 985 851 851 1212 985 851 851 115 676.02***

115 859.86***

115 766.02***

Notes: standard errors shown in parentheses beneath regression coefficients. * p < 0.05.

115 681.58***

115 672.00***

115 896.84***

115 749.70***

115 679.86***

115 699.56***

115 852.02***

115 735.48***

115 678.84***

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

Model 1

** p < 0.01. *** p < 0.001.

503

504

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

Table 4 First-stage logit estimates Diversification increase Endog (t − 1) R&D expenditure (t − 2) R&D expenditure (t − 3) Constant N (number of firm-years) Number of firms Wald χ2

Diversification increase Endog (t − 2)

0.040** (0.001) −0.105 (0.080) 851 115 9.15**

0.074*** (0.002) −0.136 (0.088) 740 115 13.08***

Standard errors shown in parentheses beneath regression coefficients. ** p < 0.01. *** p < 0.001.

technological diversification. This is confirmed by the logit estimation shown in Table 4. A firm’s volume of R&D investment appears to increase the diversity of the technology base. These results are consistent with the claim that technological diversification involves high costs, and small and medium-sized companies have to manage the trade-off between the benefit and the greater costs of such a strategy. 5. Discussion and conclusion The main aim of this paper was to examine the influence of the breadth of technological knowledge on different types of innovation. Our empirical study focused on a particular high-technology sector where innovation processes demand the integration of a wide range of skills and technical disciplines to promote the development of new products. This study has theoretical implications for the organizational learning theory, and it represents an advance in the understanding of the route to specific sorts of innovative capability. Our findings provide strong support for the premise that a diversified technology portfolio positively and significantly affects a firm’s competence to innovate. Our result demonstrates that introducing new technologies into the firm’s knowledge system favors the search for complementarities and novel solutions that increase the rate of invention and avoid learning traps. This evidence supports the theoretical notion that it is valuable to create inventories of competencies to permit effective utilization of the new knowledge, and positively influence the accumulation of absorptive capacity that allows the firm to predict the nature and commercial potential of technology advances and to exploit technological opportunities (Cohen and Levinthal, 1990; Levinthal and March, 1993). Technologically diversified companies have more strategic options in terms of purchase, licensing, and alliances and internal development to build research competences and generate innovations. Accordingly, a high level of tech-

nological diversification may be a necessary condition for firms to sustain their competitive advantage. Moreover, this study contributes to knowledge on the effects of technological diversification on exploitation and exploration. The scope of technological resources is found to have a stronger effect on exploratory than on exploitative innovative capability. Hence, different approaches of technological diversity precede different innovative outcomes. Our findings suggest that exploratory invention requires more information processing and exposure to a variety of technological knowledge domains. The different points of view, backgrounds, and types of training inherent in a diversified technological knowledge base facilitate complex problem solving, the generation of new ideas and novel combinations, and hence the development of exploratory innovative competence. The synthesis of different perspectives permits a better understanding of new technical processes, encouraging their adoption. Scope economies and knowledge depth are less important for incremental innovation because its adoption requires less knowledge resources in the organization for development or support. Although any type of innovation process entails substantive change, exploitation is focused on improving efficiency within an existing technological trajectory. Our study has implications for practice. Managers should recognize that developing a broad technological knowledge base to incorporate sources of variation and novel combinations is more likely to yield a greater impact on exploratory than on exploitative invention. This assumption has important implications and creates a dilemma for the design of formal R&D programs. Exploration and exploitation imply specific decision and adoption processes that involve different levels of risk and ambiguity. Diversifying the technological knowledge base to improve exploratory innovative competence requires greater investment in different research projects, and in information processing and integration mechanisms to ensure the cross-fertilization and combination

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

of different technologies. A high degree of technological diversification may become a source of information overload implying high coordination and communication costs. The level of such investment will be lowest when the firm uses known procedures and accumulated experience. In consequence, compared to exploitation, returns from exploration are more remote in time, distant, and uncertain. Exploitation provides efficient solutions and supports current organizational viability through near and clear returns. However, exploration improves the ability to adapt to a changing environment because it increases the variance of organizational activities (McGrath, 2001). Because of the changing nature of innovation requirements embedded in technology cycles, firms must develop capabilities to balance exploration and exploitation. It is necessary to promote a strategic integration between different research projects establishing new cognitive models for exploration, while allowing experiential learning to improve efficiency. Accordingly with the firm’s specific goals related to such a balance, managers should examine whether the firm has the necessary technologies (and other resources) to develop particular levels of innovative competence, and to determine which technologies need to be built. We recognize several limitations of this study. It is limited by its focus on the biotechnology sector. It has unique characteristics that raise questions about the generalization of our conclusions. It would be of interest to extend this research to other industries, and thus to identify sectoral differences in the importance of technological diversification, and its impact on a firm’s innovation performance. Another limitation comes from using patent data to measure the main variables as we explained in Section 3. Combining secondary source with information collected through questionnaires, to study the mentioned question would be interesting. Additionally, with this method, we would try to highlight the different impacts on innovative capability of specific sources of technological diversity. These findings would have managerial implications about how firms can optimally configure the combinations of those sources to generate different levels of exploratory and exploitative innovative competences. Another subject into which this study can be extended is the examination of how the coherence of the technology base (Christensen, 2002; Breschi et al., 2003) influences innovative capability. This analysis should be connected to evolutionary models of technical change (Dosi, 1982; Nelson and Winter, 1982; Sahal, 1985), because the requirements of a firm’s level of technological specialization or diversification may depend on the evolution and maturity of the technological domain.

505

Our research contributes to a better understanding of the relationship between a diversified technological knowledge base and specific types of innovative competence. The findings have theoretical and managerial implications since they demonstrate that variance seeking may affect the balance between exploitation and exploration. These results provide the motivation to continue the study of some unexplored issues related to the effects of different sources of technological diversity, relatedness of knowledge components and evolution of technological trajectories, which constitute promising streams for future research. Acknowledgements We thank Mauro Guillen, Mary Benner, Daniel Levinthal, Gary Pisano, the editor Nick Von Tunzelmann and the anonymous referees for helpful comments and discussions on earlier versions of this work. We dedicate this paper to Alfredo A. Quintana-S´anchez, Engineer (In Memoriam) who provided excellent technical support in the creation and management of our research database. This paper is a study carried out within the research group ‘Technological Innovation and Quality’ (SEJ-414), which has financial support from the Council of Innovation, Science and Business of the Andalusian Government (Spain). The views in this manuscript are solely the responsibility of the authors. References Abernathy, W.J., Clark, K.B., 1985. Innovation: mapping the winds of creative destruction. Research Policy 14, 3–22. Ahuja, G., Lampert, C.M., 2001. Entrepreneurship in the large corporation: a longitudinal study of how established firms create breakthrough inventions. Strategic Management Journal 22, 521–543. Andersen, H.B., Walsh, V., 2000. Co-evolution within chemical technology systems: a competence bloc approach. Industry and Innovation 7, 77–115. Argyres, N., 1996. Capabilities, technological diversification and divisionalization. Strategic Management Journal 17, 395–410. Barney, J., 1991. Firm resources and sustained competitive advantage. Journal of Management 17, 99–120. Benner, M., Tushman, M., 2002. Process management and technological innovation: a longitudinal study of the photography and paint industries. Administrative Science Quarterly 47, 676– 706. Berry, C.H., 1975. Corporate Growth and Diversification. Princeton University Press, Princeton. Breschi, S., Lissoni, F., Malerba, F., 2003. Knowledge-relatedness in firm technological diversification. Research Policy 32, 69– 87. Cantwell, J., Piscitello, L., 2000. Accumulating technological competence: its changing impact on corporate diversification and internationalization. Industrial and Corporate Change 9, 21–51.

506

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507

Cantwell, J., Santangelo, G.D., 2000. Capitalism, profits and innovation in the new techno-economic paradigm. Journal of Evolutionary Economics 10, 131–157. Cardinal, L.B., 2001. Technological innovation in the pharmaceutical industry: the use of organizational control in managing research and development. Organization Science 12, 19–36. Christensen, J.F., 2002. Building innovate assets and dynamic coherence in multi-technology companies. In: Foss, N.J., Robertson, P.L. (Eds.), Resource, Technology and Strategy. Explorations in the Resource-based Perspective. Routledge, London, pp. 123– 152. Cockburn, I.M., Henderson, R.M., 2001. Scale, scope in drug development: unpacking the advantages of size in pharmaceutical research. Journal of Health Economics 20, 1033–1057. Cohen, W.M., 1995. Empirical studies of innovative activity. In: Stoneman, P. (Ed.), Handbook of the Economics of Innovation and Technological Change. Blackwell, Oxford, pp. 182–264. Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspective on learning and innovation. Administrative Science Quarterly 35, 128–152. Cooke, P., 2002. Biotechnology clusters as regional, sectoral innovation systems. International Regional Science Review 25, 8–37. Danneels, E., 2002. The dynamics of product innovation and firm competences. Strategic Management Journal 23, 1095–1121. Dewar, R.D., Dutton, J.E., 1986. The adoption of radical and incremental innovations: an empirical analysis. Management Science 32, 1422–1433. Dosi, G., 1982. Technological paradigms and technological trajectories: a suggested interpretation of the determinants and directions of technical change. Research Policy 11, 147–162. Dosi, G., 1988. Sources, procedures, and microeconomic effects of innovation. Journal of Economic Literature 26, 1120–1171. Ettlie, J.E., Bridges, W.P., O’Keefe, R.D., 1984. Organization strategy and structural differences for radical versus incremental innovation. Management Science 30, 682–695. Fleming, L., 2001. Recombinant uncertainty in technological search. Management Science 47, 117–132. Freeman, C., Soete, L., 1997. The Economics of Industrial Innovation. MIT Press, Cambridge, MA. Gambardella, A., Torrisi, S., 1998. Does technological convergence imply convergence in markets? Evidence from the electronics industry. Research Policy 27, 445–463. Garcia-Vega, M., 2006. Does technological diversification promote innovation? An empirical analysis for European firms. Research Policy 35, 230–246. Gemba, K., Kodama, F., 2001. Diversification dynamics of the Japanese industry. Research Policy 30, 1165–1184. Giuri, P., Hagedoorn, J., Mariani, M., 2002. Technological diversification and strategic alliances. Laboratory of Economics and Management Working Paper Series, no. 2002/4, Pisa. Granstrand, O., 1998. Towards a theory of technology-based firm. Research Policy 27, 465–489. Granstrand, O., Oskarsson, C., 1994. Technology diversification in “MUL-TECH” corporations. IEEE Transactions on Engineering Management 41, 355–364. Granstrand, O., Patel, P., Pavitt, K., 1997. Multi-technology corporations: why they have “Distributed” rather than “Distinctive Core” competencies. California Management Review 39, 8–25. Griliches, Z., 1990. Patent statistics as economic indicators: a survey. Journal of Economic Literature 28, 1661–1707. Hart, P.E., 1971. Entropy and other measures of concentration. Journal of the Royal Statistical Society, Series A 134, 73–85.

Hausman, J.A., Hall, B.H., Griliches, Z., 1984. Econometric models for count data with an application to the patents-R&D relationship. Econometrica 52, 909–938. Henderson, R.M., Cockburn, I., 1996. Scale, scope and spillovers: the determinants of research productivity in drug discovery. RAND Journal of Economics 27, 32–59. Iansiti, M., West, J., 1997. Technology integration: turning great research into great products. Harvard Business Review 75, 69–79. Katila, R., Ahuja, G., 2002. Something old, something new: a longitudinal study of search behavior and new product introduction. Academy of Management Journal 45, 1186–1194. Kim, D., Kogut, B., 1996. Technological platforms and diversification. Organization Science 7, 283–301. Kogut, B., Zander, U., 1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science 3, 383–397. Le Bas, C., Patel, P., 2004. Does internationalisation of technology determine technological diversification in large firms? SPRU Electronic Working Paper Series 128. Lee, D.M., Allen, T.J., 1982. Integrating new technical staff: implications for acquiring new technology. Management Science 28, 1405–1420. Leonard-Barton, D., 1992. Core capabilities and core rigidities: a paradox in managing new product development. Strategic Management Journal 13, 111–126. Levinthal, D., March, J.G., 1993. The myopia of learning. Strategic Management Journal 14, 95–112. Liang, K.Y., Zeger, S.L., 1986. Longitudinal data analysis using generalized linear models. Biometrika 73, 13–22. Long, J., 1997. Regression Models for Categorical and Limited Dependent Variables. Sage Publications, London. Malerba, F., 2002. Sectoral systems of innovation and production. Research Policy 31, 247–264. Malerba, F., 2005. Sectoral systems of innovation: a framework for linking innovation to the knowledge base, structure and dynamics of sectors. Economics of Innovation and New Technology 14, 63–82. Malerba, F., Orsenigo, L., 2002. Innovation and market structure in the dynamics of the pharmaceutical industry and biotechnology: towards a history-friendly model. Industrial and Corporate Change 11, 667–703. Mansfield, E., 1986. Patents and innovation: an empirical study. Management Science 32, 173–181. March, J.G., 1991. Exploration and exploitation in organizational learning. Organization Science 2, 71–87. McGrath, R.G., 2001. Exploratory learning, innovative capacity, and managerial oversight. Academy of Management Journal 44, 118–131. Mendoc¸a, S., 2006. The revolution within: ICT and the shifting knowledge base of the world’s largest companies. Economics of Innovation and New Technology 15, 777–799. Nelson, R.R., 1959. The simple economics of basic scientific research. Journal of Political Economy 67, 297–306. Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory of Economic Change. Belknap Press of Harvard University Press, Cambridge, MA. Nesta, L., Dibiaggio, L., 2003. Technology strategy and knowledge dynamics: the case of biotech. Industry and Innovation 10, 329–347. Nesta, L., Saviotti, P., 2005. Coherence of the knowledge base and firm innovative performance: evidence from the US. Pharmaceutical industry. Journal of Industrial Economics 53, 123–142.

C. Quintana-Garc´ıa, C.A. Benavides-Velasco / Research Policy 37 (2008) 492–507 Nicholls-Nixon, C., Woo, C., 2003. Technology sourcing and output of established firms in a regime of encompassing technological change. Strategic Management Journal 24, 651–666. Palmberg, C., Martikainen, O., 2006. Diversification in response to ICT convergence—indigenous capabilities versus R&D alliances of the Finninsh telecom industry. Info 8, 67–84. Patel, P., Pavitt, K., 1997. The technological competencies of the world’s largest firms: complex and path-dependent, but not much variety. Research Policy 26, 141–156. Pavitt, K., 1984. Sectoral patterns of technical change: towards a taxonomy and a theory. Research Policy 13, 343–373. Pisano, G., 1994. Knowledge, integration, and the locus of learning: an empirical analysis of process development. Strategic Management Journal 15, 85–100. Pisano, G., 2002. In search of dynamic capabilities. In: Dosi, G., Nelson, R.R., Winter, S.G. (Eds.), The Nature and Dynamics of Organizational Capabilities. Oxford University Press, New York, pp. 129–154. Piscitello, L., 2000. Relatedness and coherence in technological and product diversification of the world’s largest firms. Structural Change and Economic Dynamics 11, 295–315. Piscitello, L., 2004. Corporate diversification, coherence and economic performance. Industrial and Corporate Change 13, 757–787. Powell, W.W., Kogut, K.K., Smith-Doerr, L., 1996. Interoganizational collaboration and the locus of innovation: networks of learning in biotechnology. Administrative Science Quarterly 41, 116–145. Rao, P.M., Vemuri, V.K., Galvin, P., 2004. The changing technological profile of the leading ICT firms: evidence from US patent data, 1981–2000. Industry and Innovation 11, 353–372. Romeu-Santana, A., Vera-Hern´andez, A.M., 2005. A seminonparametric estimator for counts with and endogenous dummy variable. Econometrics Journal 8, 1–22. Rosenberg, N., 1969. The direction of technological change: inducement mechanisms and focusing devices. Economic Development and Cultural Change 18, 1–24. Rosenberg, N., 1976. Perspectives on Technology. Cambridge University Press, Cambridge, MA.

507

Rosenkopf, L., Nerkar, A., 2001. Beyond local search: boundaryspanning, exploration, and impact in the optical disk industry. Strategic Management Journal 22, 287–306. Sahal, D., 1985. Technological guideposts and innovation avenues. Research Policy 14, 61–82. Silverman, B.S., 1999. Technological resources and the direction of corporate diversification: toward and integration of the resourcebased view and transaction cost economics. Management Science 45, 1109–1124. Subramaniam, M., Youndt, M.A., 2005. The influence of intellectual capital on the types of innovative capabilities. Academy of Management Journal 48, 450–463. Suzuki, J., Kodama, F., 2004. Technological diversity of persistent innovator in Japan. Two case studies of large Japanese firms. Research Policy 33, 531–549. Terza, J.V., 1998. Estimating count models with endogenous switching: sample selection and endogenous treatment effects. Journal of Econometrics 84, 129–154. Tidd, J., Bessant, J., Pavitt, K., 2001. Managing Innovation. Integrating Technological, Market and Organizational Change. Wiley, Chichester. Valvano, S., Vannoni, D., 2003. Diversification strategies and corporate coherence evidence from Italian leading firms. Review of Industrial Organization 23, 24–41. Wang, Q., von Tunzelmann, G.N., 2000. Complexity and the functions of the firm: breath and depth. Research Policy 29, 805– 818. Windmeijer, F.A.G., Santos-Silva, J.M.C., 1997. Endogeneity in count models: an application to demand for health care. Journal of Applied Econometrics 12, 281–294. Wuyts, S., Dutta, S., Stremersch, S., 2004. Portfolios of interfirm agreements in technology-intensive markets: consequences for innovation and profitability. Journal of Marketing 68, 88– 100. Zorn, C.J.W., 2001. Generalized estimating equation models for correlated data: a review with applications. American Journal of Political Science 45, 470–490.