Firm Size And Dynamic Technological Innovation

Technovation 22 (2002) 537–549 www.elsevier.com/locate/technovation

Firm size and dynamic technological innovation Gregory N. Stock a

a,*

, Noel P. Greis b, William A. Fischer

c

Department of Operations Management and Information Systems, College of Business, Northern Illinois University, DeKalb, IL 60115, USA Kenan Institute of Private Enterprise, Campus Box 3440, Kenan Center, University of North Carolina, Chapel Hill, NC 27599-3440, USA c IMD — International Institute for Management Development, Ch. de Bellerive 23, P.O. Box 915, CH-1001 Lausanne, Switzerland

b

Received 24 March 2001; accepted 9 June 2001

Abstract Competitive strategy can be influenced by technological change and technological innovation over time, a process we refer to as dynamic innovation. Using data from the computer modem industry, we examine the relationship between firm size and dynamic innovation. Innovation performance is represented by the technological performance of the outputs of the firm’s innovation process, namely new products developed by the firm. In contrast, the extant literature typically considers the relationship between size and innovation at a single point in time, and innovation performance is generally characterized by the productivity of the innovation process. Our findings indicate that smaller firms in the computer modem industry exhibit higher levels of dynamic innovation performance.  2002 Elsevier Science Ltd. All rights reserved. Keywords: Dynamic innovation; Technology; Firm size

1. Introduction Technological change can be a critical driver of competitive strategy (Porter, 1985; Nelson and Winter, 1982). In addition, technological change can affect the evolution of industry structure (Abernathy and Clark, 1985; Tushman and Anderson, 1986) and economic growth (Schumpeter, 1934; Solow, 1957). Given its importance at these multiple levels of analysis, an important avenue of research would be an exploration of the drivers of technological change. Technological change is intertwined with the process of technological innovation, and a number of conceptualizations of innovation explicitly include the rate of change in the functional performance of a technology as a key element (Schumpeter, 1942; Utterback and Abernathy, 1975; Abernathy and Clark, 1985; Tushman and Anderson, 1986; Anderson and Tushman, 1990; Henderson and Clark, 1990; Christensen, 1992a,b; Methe, 1992; Henderson, 1995; Khanna, 1995; Foster, 1986; Tushman and Rosenkopf, 1992). In fact, one definition of inno* Corresponding author. Tel.: +1-815-753-9329; fax: +1-815-7537460. E-mail addresses: [email protected] (G.N. Stock), [email protected] (N.P. Greis), [email protected] (W.A. Fischer).

vation characterizes innovation as “a change in technology which is manifested in the development of new products” (Methe, 1992, p. 14, emphasis added). Technological change can therefore be characterized as a series of innovations over time. Our interest in what drives technological change, therefore, is in essence an attempt to understand the process of technological innovation over time, or more specifically, the process of creating a series of innovations over time. In this paper, we refer to this process as “dynamic innovation.” Dynamic innovation might be manifested in a number of ways, although in this research we follow the approach of Methe (1992) above and explicitly consider the technological change embodied in new products developed by the firm. A logical next step is to investigate empirically which factors are related to this process of dynamic innovation. In the technology management literature, the relationship between firm size and technological innovation has received a good deal of attention. This general topic can be traced back at least as far as Schumpeter (1942). The straightforward quality of the question and the fundamental nature of firm size as an organizational variable likely account in large part for the attractiveness of the topic for research. Although there is a large literature investigating the

0166-4972/02/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 6 6 - 4 9 7 2 ( 0 1 ) 0 0 0 6 1 - X

538

G.N. Stock et al. / Technovation 22 (2002) 537–549

relationship between firm size and innovation, this study provides two important contributions. First, we consider the relationship between firm size and what we term dynamic innovation. There is a substantial body of research that has examined innovation in a dynamic context (e.g., Abernathy and Clark, 1985; Tushman and Anderson, 1986; Abernathy and Utterback, 1978), but the connection between firm size and dynamic innovation has received comparatively little attention. A second contribution concerns the assessment of innovation performance. In most of the literature on firm size and innovation, the focus is on the efficiency, or productivity, of the process of innovation and whether larger or smaller firms are able to innovate at a lower cost per innovation. Obviously, this issue is extremely important. For two firms developing equivalent new products, the one that is able to accomplish this task at the lower cost will likely reap a competitive advantage. However, what happens if the new products are not equivalent? A firm that is able to develop more technologically advanced products more quickly might have a competitive advantage that could overcome a disadvantage in the cost efficiency of its innovation process. In fact, the uniqueness and performance of a new product is often identified as a success factor in new product development (Cooper, 1979, 1990; Cooper and Kleinschmidt, 1986). Therefore, we consider a second perspective in the performance of a firm’s technological innovation process. We explicitly consider in our innovation performance measure whether the innovations created by a firm are more or less technologically advanced. The most straightforward approach to assess whether an innovation (a new product innovation in this paper) is technologically advanced is to consider its technical performance. Moreover, because we focus on the process of dynamic innovation, we consider the rate of change in the technical performance of a firm’s new products introduced over time. This second perspective we refer to as the effectiveness of the innovation process. To examine this topic, we have to chosen to focus on dynamic innovation within a single industry, namely the computer modem industry. Restricting the study to a single industry and product type may limit the extent to which its results can be generalized. However, limiting the study to modem manufacturers allows us to control for industry effects that might otherwise confound our analysis. A “modem” (modulator–demodulator) is a device that allows a computer to send and receive data over telephone lines. Specifically, a modem converts (modulates) the digital signals used by computers into analog signals that can be transmitted over telephone lines. At the receiving computer, another modem converts the analog signal back (demodulates) into a digital signal that can be used by the receiving computer. Because telephone lines were designed for voice transmission many years before computers existed, a key

challenge in providing greater data communications performance is in the ability of the modem to overcome the physical limitations for data transmission inherent in a telephone line. The modem industry provides an attractive setting for this study for a number of reasons. The industry is technology-intensive, meaning that the technological performance of a firm’s products is probably a key determinant of the firm’s market success. There is also a fairly wide range of electronic technologies embodied in a modem (i.e., both analog and digital circuitry; both hardware and software; large, complex integrated circuits such as microprocessors as well as simpler integrated circuits). The technical performance of modems has increased many-fold over the industry’s lifetime. The underlying technology itself in this industry has undergone substantial change as well. In particular, the basis for technological competence for firms in this industry has evolved from the design and use of proprietary, dedicated hardware to the design and use of proprietary software running on general-purpose hardware. This evolution has mirrored similar changes seen in computers and peripheral products such as printers and disk drives. From an organizational perspective, there is a diverse group of large and small firms developing and manufacturing modems. The industry includes wellknown firms such as IBM and Motorola, as well as lesser-known companies such as Boca Research. The remainder of this paper will be organized in the following manner. The next section will further consider the concept of dynamic innovation, review background literature on firm size and innovation, and develop our research hypothesis. The fourth section describes the collection and analysis of the data. The fifth section reports the results of the data analysis and the evaluation of the research hypothesis. The final section discusses implications and directions for future research.

2. Conceptual framework 2.1. Dynamic innovation Above, we introduced the concept of dynamic innovation as a process of creating a series of innovations over time. In this section, we explore this idea in more detail. To begin, we first examine the general concept of technological innovation and associated literature. We will then consider dynamic innovation and literature related to topic. In general, an innovation refers to a new way of accomplishing some task, or the implementation of an idea. Therefore, “[i]nnovation does not occur when a new idea is generated, but rather when that new idea is put into use” (Damanpour, 1987, p. 676). A technological innovation, which is the focus of this paper, is the

G.N. Stock et al. / Technovation 22 (2002) 537–549

use of new technology to produce changes in products or services, or the ways in which products or services are produced (Damanpour, 1987). Therefore, a technological innovation can be thought of as the incorporation of technology into the development of new products or processes. This definition of innovation has specific implications for how innovation should be measured. A good deal of prior research in the relationship between firm size and innovation has used various proxy measures for innovation, including indicators of innovative effort or activity, such as R&D spending or R&D intensity (Acs and Audretsch, 1991a; Kamien and Schwartz, 1982). One conceptual problem with this type of measure is that it is an input to the innovative process, not an output. Other research has attempted to remedy this shortcoming by using other measures, such as patents, patent citations, or scientific publications (Scherer, 1965; Halperin and Chakrabarti, 1987; Narin and Noma, 1987).1 While these measures are probably better, they still are more closely related to inventive activity than innovation. They represent the generation of ideas, not necessarily the application of ideas to new products or processes. Another body of literature has used the number of innovations created by a firm as the measure of innovation (Acs and Audretsch, 1987, 1988; Audretsch and Acs, 1991; Pavitt et al., 1987). These studies, however, make no distinction between innovations with respect to their quality or impact. In a similar vein, there has been research focusing on the pharmaceutical industry that has used the number of new drugs brought to market as a measure of innovation (Graves and Langowitz, 1993; Jensen, 1987). This paper, as we discussed above, focuses on what we refer to as the effectiveness of the innovation process, which is reflected by the extent to which innovations developed by this process are technologically advanced. Therefore, innovation performance is based on the technical performance of the products actually developed and introduced into the market by the firms in the industry of interest. This conceptualization of innovation thus incorporates both elements of our definition of a technological innovation. It reflects the implementation of an idea as a new product, and it explicitly recognizes the application of technology in the new product. This approach is consistent with the definition of Methe (1992) above. Beyond the general concept of innovation, we are concerned in this study with dynamic innovation, which explicitly considers the process through which a firm develops a series of innovations over time. There is a good deal of literature examining innovation in a

1 See Griliches (1990) for an extensive review of the literature employing patent statistics.

539

dynamic context. This literature explores a number of areas including the relationship between changes in industry structure and changes in innovation (Schumpeter, 1942; Tushman and Anderson, 1986; Abernathy and Clark, 1985); the relationship between organizational changes and the relative importance of product and process innovation over time (Utterback and Abernathy, 1975); the relative competencies of new entrants or established firms in responding to technological change (Tushman and Anderson, 1986; Henderson and Clark, 1990; Christensen, 1992a,b); firm size, market structure, and evolutionary innovation (Methe, 1992; Nelson and Winter, 1982); technology life cycles (Anderson and Tushman, 1990; Khanna, 1995; Henderson, 1995); and the specific form of technological trajectories over time (Dosi, 1982; Foster, 1986). Underlying much of this prior research is an implicit consideration of the rate of technological change. For example, the “era of ferment” found in the technology life cycle of Anderson and Tushman (1990) is characterized by rapid technological change; it is followed by a period of slower, incremental change. Similarly, the technology “S-curve” examined by Foster (1986) exhibits distinct periods of both slow change and rapid change in technical performance. Barnett and Clark (1996, p. 264) summarize the interconnected relationship between technological change and product development in the following statement: “The entire sweep of technological changes over the past one hundred years is, in essence, the sum of thousands of individual product development projects in thousands of firms.” This focus on technological change leads us to select the rate of technical change in product performance as our approach to measuring dynamic innovation performance. Therefore, higher levels of dynamic innovation performance will be indicated by higher rates of technological change in the performance of products developed by a firm. 2.2. Firm size and innovation The question of whether large or small firms are more technologically innovative has been the subject of a great deal of controversy and research. The arguments in favor of large or small firms as the engine of innovation can take on an almost “theological” tone, as illustrated by the lively and somewhat contentious debate between Gilder (1988) and Ferguson (1988) in the Harvard Business Review and Ferguson (1988). The issue seems especially unclear to managers and the popular business press. A recent sampling of general business publications shows a recent wave of mergers and consolidations that reflects a belief that size provides a number of advantages, many of them related to technological capabilities, and is necessary to compete effectively, particularly over the long term (Colvin, 1999; Kupfer, 1998; Harrington et al., 1998; Anon., 1998b; Greising et al., 1998). The view

540

G.N. Stock et al. / Technovation 22 (2002) 537–549

is not universally accepted, however, as many believe that size achieved through merger and consolidation is fraught with difficulties (Loomis, 1999; Taylor, 1999; Garten, 1998; Anon., 1998a). At the same time, a number of firms, even in the midst of a robust economy, have reduced workforce size through actions known by a variety of names, including layoffs, downsizing, reengineering, and outsourcing (Koretz, 1998; Greising, 1998). Too often, however, reducing the size of the firm is motivated only by a desire to cut costs. The danger of such a narrow focus is that it may result in a weakening or elimination of core technological capabilities that can determine the success of the firm over time (Prahalad and Hamel, 1990; Anon., 1995; Bernstein, 1998). Scholarly study of the relationship between firm size and innovation can be traced to the work of Schumpeter (1942), specifically to the idea of the Schumpeterian hypothesis. This “hypothesis” actually is a set of two hypotheses. The first states that there will be a positive relationship between innovation and monopoly power; the second states that large firms will be more than proportionately more innovative than small firms. Kamien and Schwartz (1982) also note that this second hypothesis, while attributed to Schumpeter, was developed more fully by Galbraith (1956). In the remainder of this paper, when we refer to the “Schumpeterian hypothesis” we will be referring to the second hypothesis. In this section we first examine the reasoning for and against the Schumpeterian hypothesis, as well as the research evidence supporting these conflicting arguments. Then we consider how prior research in dynamic innovation relates to the Schumpeterian hypothesis. Research examining the relationship between firm size and innovation has come primarily from an economics perspective. The arguments in favor of the Schumpeterian hypothesis draw primarily on the concept of economies of scale in R&D activities, which has a number of possible explanations. For example, a larger firm will have the ability to employ a larger R&D staff, which will lead to economies of scale in R&D. These economies of scale can be traced to a number of factors. The first is that engineers and scientists will be more effective when they have more colleagues with whom to interact. In addition, a large staff can allow the division of labor in research and development. A third possible advantage of size is that a larger R&D group is more likely to recognize the importance of unforeseen discoveries (Kamien and Schwartz, 1982). More generally, larger size will allow a firm to accumulate a larger store of technological knowledge and capabilities (Damanpour, 1992). These factors relate principally to the technical development of innovations. Perhaps as important is the advantage a larger firm will enjoy in the exploitation of the technical advances. The reasoning here is that a larger firm will be in a better position to exploit an unforeseen innovation because it can more easily enter a new market. In

addition, a large multiproduct firm will have more opportunities for diversification of R&D projects, and will therefore be able to realize a higher yield from the resources devoted to R&D (Kamien and Schwartz, 1982). Another way to think of this point is that a larger firm will have the resources to tolerate an occasional unsuccessful R&D project (Damanpour, 1992). On the opposing side, there are arguments that smaller firms have greater advantages in innovation. In general, a smaller firm might be more innovative because it would be expected to be more flexible and therefore be better able to accept and effect change (Damanpour, 1992). In a large firm, there is a good deal more bureaucracy, which leads to more difficult communication and coordination of R&D. Gilder (1988) similarly attributes the innovative advantage of small firms to an ability to avoid the “bureaucratic inertia” found in large companies and to a greater ability to adapt to changes in markets. Also, unforeseen research findings may be more likely to go unrecognized among the larger volume of R&D activity occurring in a large firm. Another consideration is that engineers and scientists in a smaller firm may be more highly motivated than in a large firm. In a small firm, the compensation of an individual may be more tightly linked to performance than in a large firm, particularly in those entrepreneurial firms where a scientist or engineer receives stock or stock options as part of a compensation package. Moreover, the contribution of an individual in a small firm is likely to have a more visible impact on the firm’s overall performance than in a larger firm, which would also lead to a higher degree of motivation (Kamien and Schwartz, 1982). As noted above, there is an extensive literature that empirically examines the relationship between firm size and technological innovation. In their literature review, Kamien and Schwartz (1975, p. 15) characterize the objective of this stream of research in the following manner: “A statistical relationship between firm size and innovative activity is most frequently sought with exploration of the impact of firm size on both the amount of innovational effort and innovation success.” We will not attempt to review this entire body of literature.2 However, we will highlight some of the more relevant research. Consistent with our observations above concerning the conceptual difficulties in using inputs as a measure for innovation, we consider here only the research relating firm size to innovation outputs. Unfortunately, the literature relating firm size to innovative or inventive outputs shows decidedly mixed results. Kamien and Schwartz (1982, p. 84) find that “beyond some magni-

2

For a thorough review of the early literature see Kamien and Schwartz (1982). For a review of more recent literature, see Acs and Audretsch (1991b).

G.N. Stock et al. / Technovation 22 (2002) 537–549

tude, size does not appear to be especially conducive to either innovational effort or output.” More recent research has shown similar results. A number of studies have shown that patent counts increase at a rate that is less than proportional to firm size (Bound et al., 1984; Schwalbach and Zimmerman, 1991; Chakrabarti and Halperin, 1991). Other studies using the number of innovations (Acs and Audretsch 1987, 1988; Audretsch and Acs, 1991), the number of new drugs brought to market (Graves and Langowitz, 1993), and scientific publications (Halperin and Chakrabarti, 1987) have yielded similar findings. On the other hand, there is research showing that larger firms have an advantage in innovation, at least in some cases (Henderson and Cockburn, 1996; Lichtenberg and Siegel, 1991; Mansfield, 1980; Harrison, 1994; Cohen and Klepper, 1996). The prior research has examined the relationship between firm size and innovation. This literature provides a starting point from which to develop a conceptual framework for our subsequent analysis. However, its usefulness in our study is limited in three ways. First, the contradictory nature of both the conceptual and empirical findings related to firm size and innovation does not provide clear guidance of what to expect in general. Second, as we have noted above, the economics literature typically considers the relationship between firm size and the productivity of the innovation process, rather than the relationship between firm size and the performance of the technology that is the output of the innovation process. Although the two outcomes may very well be related, it is not clear that they necessarily would be. Third, for the most part, the empirical research in this area typically considers the relationship between firm size and innovation only at a single point in time. At this point, we also need to consider how the literature on dynamic innovation discussed above applies to the relationship between firm size and innovation. Although this literature provides a good base on which to conceptualize technological change, for the most part it does not consider the firm size explicitly as an organizational variable. However, the evolutionary framework of Nelson and Winter (1982) is an exception. We will use this framework in addition to the larger body of literature relating firm size and innovation reviewed above to develop a research hypothesis relating firm size to dynamic technological innovation. The evolutionary framework argues that sustained innovation will require more resources, which gives larger firms an advantage. In addition, larger firms are better able to appropriate returns from their innovation and are less vulnerable to temporary periods of lower innovative performance. The evolutionary framework therefore provides a theoretical justification for the expectation that larger firms will be more innovative. In addition, in his study of the semiconductor industry Methe (1992) found empirical support for this aspect of

541

the evolutionary framework. We recognize that there are arguments and a good deal of empirical evidence contradicting the Schumpeterian hypothesis. However, in this paper we find the theoretical arguments and empirical evidence compelling enough to adopt a position consistent with the evolutionary framework and the Schumpeterian perspective. Therefore, our research hypothesis can be stated in the following manner: H1: Firm size will be positively related to dynamic technological innovation performance. Recall that we have conceptualized and will measure dynamic innovation performance as the rate of change in new product technical performance. The research hypothesis suggests that the rate of change would be greater for larger firms, so to test this hypothesis, we need to determine whether the rate of technical change depends on firm size. We will discuss the details of our analysis below.

3. Methodology 3.1. Data and variables Data for this study were collected from the computer telephone modem industry for the period from 1974 through 1993. Two different sets of data are used in this study. One data set consists of information about the technical performance of products introduced by firms in this industry during the time horizon of the study. This data set was constructed from yearly modem industry overview reports from Datapro Research, Inc. (Anon., 1976–1993). Datapro, a subsidiary of McGraw-Hill, Inc., is a firm that collects information and publishes reports on many segments of the information technology industrial sector, including software, computers, and data communications equipment. One series of publications provides detailed information about modem manufacturers on a yearly basis. Each of these overview reports provides a discussion of the technical operation of modems, new technical advances in the industry, activity by leading firms in the industry, and most importantly for this study, a comprehensive listing of modem products offered for sale during the year of the report. A set of reports published from 1976 through 1993 was purchased from Datapro. These reports listed products that had been introduced in the years 1974 through 1993. The information for each product included the company selling the product, the model name or number, technical specifications, the year the product was first sold, and other information. The technical specification of data transmission rate in bits per second and the year of introduction were the two items that were used in this study to assess product performance and ultimately technologi-

542

G.N. Stock et al. / Technovation 22 (2002) 537–549

cal innovation by the firm. Data for firm size were obtained from a variety of archival sources, including Standard and Poor’s Compustat data base of information on publicly traded firms and industry directories such as Standard and Poor’s Register of Corporations, Corporate Technology Directory, and Ward’s Directory of Corporations (Anon. 1986–1994; 1987–1994; 1971–1994; 1994). Both the product data set and the firm variable data set were constructed on a year-by-year basis and included firms in the industry in each year. Table 1 lists the variables and data sources used in the analysis. Ultimately, our objective is to relate dynamic technological innovation to the variable of firm size. Above we argue that this measure should be based on the output of the innovation process. Moreover, that output should be directly related to the commercial activity of the firm. Therefore, a measure of technological innovation that is especially appropriate would reflect the technical performance of the products developed by a firm and introduced to the market. Firms whose products are more technologically advanced (products showing higher technical performance) are considered in this study to be more innovative than firms whose products are less technologically advanced. Therefore, the extent to which the technical performance of a firm’s products is higher or lower will reflect the effectiveness (or performance) of the firm’s innovation process. The first step is to determine a measure of product performance. In our study, the product of interest is the computer telephone modem. We have chosen two measures of performance. The first measure of performance is relatively straightforward, namely, the data transmission rate in bits per second. The transmission rate indicates number of bits of information that the modem can communicate per second over a telephone line. We selected this measure because it is the performance of the technical attribute reflecting the fundamental function of the product, namely the transmission and reception of data over telephone lines. Although a modem

may have other technical features, the data transmission rate is the key capability. Whatever other capabilities two modems may have, a 14,400 bit per second modem will likely be considered to have higher performance than a 9600 bit per second modem. The use of a single “fundamental” measure of performance in this way is found in a good deal of research in technology and innovation, particularly when considering technological change. For example, in computer disk drives, Christensen (1992a,b) uses megabytes of storage per square inch. Khanna (1995), in a study of the mainframe computer industry, employs computer operation cycle time. Foster (1986) specifies two characteristics of an appropriate performance parameter. The first is that it be of value to customers; the second is that it be expressed in terms that make sense to engineers and scientists in the field. Both attributes are embodied in the measure of transmission rate in bits per second. The second measure we have chosen is the transmission rate divided by the product price, which yields the measure of bits per second per dollar. We include this measure of product performance in our analysis for a number of reasons. The inclusion of price in this measure reflects an economic parameter that may provide a different set of performance indicators for the product and for firms. For example, this measure may allow two firms that compete in different market segments to be compared on a more or less equal basis. A firm that develops “low-end” products, but does it in such a way that it can offer these products for a very low price might be considered to be as technologically innovative as a firm that develops high-performance, high-priced products. In addition, product price has been included in measures of technological performance in a number of other studies (Dodson, 1985; Peterson et al., 1987; Esposito, 1993). This measure is also consistent with Foster’s (1986) criteria, discussed above, for the appropriateness of a performance parameter. We also emphasize that a product is included in the

Table 1 Variables, measures, and data sources Variable

Units of measure

Data sources

Bits per second

Datapro, Inc. industry reports

Bits/second/dollar

Calculated from data in Datapro reports

Size of firm

Log2(Number of employees)

COMPUSTAT, CorpTech, Ward’s Directory

(SIZEit) Year (YEAR)

Year (1983=14 to 1993=24)

Dependent variables Yearly average transmission rate of new modem products introduced (AVGRATEit) Yearly average transmission rate divided by price (AVGRATE/$it) Independent variables

G.N. Stock et al. / Technovation 22 (2002) 537–549

data set only once, in the year in which it was first introduced. Our view is that focusing on new products provides a better reflection of a firm’s technological capabilities and decisions. It is very likely that an older product that has been on the market for many years would receive very little consideration from a firm’s R& D department. In addition, when price is considered, older products may be affected by learning curve or marketing considerations that have little to do with a firm’s technological development capabilities. Having identified appropriate measures of product technical performance, we can then construct a measure of firm innovation performance. In our case, we have chosen to employ the simple average of the performance of the products introduced by each firm in each year.3 The average product performance provides a composite measure of the performance of the products introduced by the firm. Dynamic innovation performance is then the rate of change in this yearly firm innovation measure. Firm size is measured by the logarithm (base 2) of the number of employees in the firm. Taking the logarithm reduces the skewness in the firm size distribution. We used the logarithm to the base 2 to aid in the interpretation of the results. The coefficient of this variable in a regression model can be thought of as the change in the dependent variable associated with a doubling in firm size. 3.2. Data analysis Our objective is to relate technological innovation over time to firm size. To investigate this relationship, we develop a number of regression models where technological performance is the dependent variable, and firm size and time are the independent variables. In particular, we would like to see whether firm size has a statistically significant effect on the relationship between performance and time. The general approach is described here; detailed descriptions of the specific regression models will be provided below in the section reporting the results of the analysis. The general form of the regression model is shown below: PERFit⫽b0⫹b1SIZEit⫹b2YEAR⫹b3(SIZE∗YEAR) where PERFit=Innovation performance (either average

3 For example, suppose a firm introduces three new modem products in a particular year. One has a performance of 1200 bits per second (bps); one has a performance of 2400 bps; and the third has a performance of 4800 bps. The yearly firm performance measure would be the average of these three product performance measures:

Firm performance⫽(1200⫹2400⫹4800)/3⫽2800 bps The same approach would be used to calculate the yearly firm performance for the bps/$ product performance measure.

543

transmission rate or average transmission rate/dollar) of products introduced by firm i in year t, SIZEit=firm size (log2 of number of employees) for firm i in year t, YEAR=time variable (year), and SIZE∗YEAR=interaction between firm size and time. The regression equation can be rewritten in the following manner: PERFit⫽b0⫹b1SIZEit⫹(b2⫹b3SIZE)YEAR If b3 is statistically significant, then there is evidence that there is a relationship between the rate of change in technological performance and firm size. If b3 is positive, then the slope would be greater, and we would conclude that firm size is positively related to dynamic innovation. Conversely, if b3 is negative, then the slope of the linear equation relating performance and time will be lower with increasing firm size, and we would conclude that firm size is negatively related to dynamic innovation. Therefore, our research hypothesis would be supported if the sign of the b3 coefficient is positive. To provide some additional insight into the analysis, we also tested variations on this basic model. The details of this additional analysis are presented later in the discussion of the results. In addition, we used a fixed effects model to control for the portion of the dependent variable that is “fixed” in time. Otherwise, parameter estimates could be biased. The approach we used to estimate the fixed effects model was to define a dummy variable for all but one firm in the sample, which yields a regression model with a separate intercept for each firm. This type of model assumes that the effects of the variables of interest (the interaction of size and time) affect all firms in the same way and that only the intercepts are different (Johnston and DiNardo, 1997).

4. Results 4.1. Descriptive statistics We initially began with a data set that covered the period of 1974–1993. This sample included 1767 distinct new products. The technical specifications of these individual products were used to calculate a total of 595 observations. However many of these firms are privately held, and therefore in many cases, data for firm size are not available. This restriction reduced the sample to a total of 306 observations. In addition, for some products, price was not reported in the data, so the sample when transmission rate/price was included consisted of 291 observations. As shown in Figs. 1 and 2, the period of time from 1983 to 1993 exhibited rapid change in the technical performance of the products introduced by the firms in this industry. In contrast, it is apparent that firm performance

544

G.N. Stock et al. / Technovation 22 (2002) 537–549

Fig. 1.

Fig. 2.

Average transmission rate vs year.

Average transmission rate/price vs year.

does not change a great deal for either measure until approximately 1983. Because we are concerned with technological change and dynamic innovation over time, we decided to limit our analysis to the years 1983–1993. In addition, data for firms that appeared in only one year were also deleted, because the idea of innovation over time obviously is not applicable for those firms. The result of limiting the data set in this way was to reduce the sample to a set of 174 observations (167 observations for the performance measure of transmission rate/price). There are a total of 44 companies in this data set. Table 2 shows minimum, maximum, mean, and median values for the data and variables in this final sample. Table 3 shows correlations between variables in the sample. 4.2. Regression results Several regression models were estimated using the approach outlined above. Table 4 shows the results of the basic regression model for both performance measures. In both models, there is a strong positive relation-

ship between firm performance and year, reflecting the upward movement in technical performance over time for modems. There is also a statistically significant interaction between firm size and year, indicating that there is a significant relationship between firm size and technological innovation over time. Moreover, the interaction coefficient in both models is negative, which indicates that the rate of technological change decreases with increasing firm size. However, the coefficient for the main effect of firm size was not statistically significant. To aid in interpreting these regression results, they are shown below in the rewritten form introduced above. Because we are interested in how firm size affects the innovation over time, we ignore the intercept and dummy variables, as well as the non-significant main effect term for firm size. AVGRATE⫽(1978.2⫺50.5 SIZE) YEAR AVGRATE/$⫽(3.088⫺0.110 SIZE) YEAR What these two models show is that the slope of the performance versus year “line” is smaller for larger firms

G.N. Stock et al. / Technovation 22 (2002) 537–549

545

Table 2 Descriptive statistics Variable

Minimum

Maximum

Mean

Standard deviation

Avg. transmission rate (bits/s) Avg. transmission rate/price (bits/s/$) Log2(employees)

300

28,800

5717

4494

0.510

37.399

8.009

6.828

3.322 (10)a 14

18.622 (403,500)a 24

9.484 (716)a 19.046

4.279

Year a

2.774

The numbers in parentheses are the actual number of employees corresponding to the logarithm.

Table 3 Correlations Variable 1 2 3 4

1 Avg. transmission rate Avg. transmission rate/price Log2(employees) Year

a

2 1.0 0.611***a 0.097 0.694***

1.0 ⫺0.132

4

1.0

*p⬍0.05; **p⬍0.01; ***p⬍0.001.

Table 4 Regression results Independent variable

Dependent variable AVGRATE

Firm dummy variables INTERCEPT SIZE YEAR SIZE*YEAR R2

1.0 ⫺0.245** 0.730***

3

AVGRATE/$

3000); the first quartile value for SIZE is 6.322 (corresponding to an actual number of employees of 80). Entering those values into the equations above yields the following linear relationships: SIZE=11.551 (corresponding to 3000 employees) AVGRATE =[1978.2−(50.5)(11.551)] YEAR

⫺33675*** (5975) 635.8 (544.1) 1978.2*** (279.6) ⫺50.5* (23.0) 0.716

a

⫺37.977*** (8.082) 1.054 (0.783) 3.088*** (0.378) ⫺0.110** (0.031) 0.788

a *p⬍0.05; **p⬍0.01; ***p⬍0.001. Standard errors are shown in parentheses.

than for smaller firms. Recall that the SIZE variable is log2 of the number of employees in the firm. Consider two firms, Firm A and Firm B. Firm A has twice as many employees as Firm B, so the value of SIZE for Firm A would be exactly 1.0 unit greater than the value of SIZE for Firm B. From the regression results shown above, the expected rate of change in average transmission rate for Firm A would be 50.5 (bits/second)/year less than for Firm B. Similarly, the expected rate of change in average transmission rate/price would be 0.110 (bits/second/$)/year less for Firm A than for Firm B. Another example presents an even clearer distinction. The third quartile value for SIZE is 11.551 (corresponding to an actual number of employees of

=1394.9 YEAR AVGRATE/$ =[3.088−(0.110)(11.551)] YEAR =1.817 YEAR SIZE=6.322 (corresponding to 80 employees) AVGRATE =[1978.2−(50.5)(6.322)] YEAR =1658.9 YEAR AVGRATE/$ =[3.088−(0.110)(6.322)] YEAR =2.393 YEAR This example shows that the expected rate of change in technical performance (for either measure) is substantially higher in the smaller of the two firms. Additional evidence of the higher levels of dynamic performance over time for smaller firms is provided by another approach to this analysis. Note from Table 3 that there is a very weak, and in fact non-significant, relationship between firm size and time. We took advantage of the relative time-invariance of size to divide the sample into two sub-samples. One group comprised firms whose average size over the time frame of the study was above the median SIZE of 8.274 (corresponding to 310 employees), and the other comprised firms whose average size was less than or equal to the median value. We

546

G.N. Stock et al. / Technovation 22 (2002) 537–549

then estimated separate regressions for each of the two groups, again using a fixed effects model (Kleinbaum et al., 1988). The results of these two regressions are provided in Table 5. Examination of these results shows that the rate of change in firm performance was quite a bit greater in the small firm subgroup for both performance measures.

5. Implications and directions for future research The most immediate implication of these findings was the lack of support for our research hypothesis. This hypothesis, following the logic of the evolutionary framework of Nelson and Winter (1982) and the empirical results of Methe (1992), predicted that there would be a positive relationship between firm size and dynamic innovation performance. Not only did our results fail to support this expectation, they in fact found a negative relationship between firm size and dynamic innovation. Smaller firms showed a significantly higher rate of change in product performance, on average, than did larger firms. Our results therefore provide evidence for the argument that smaller firms are more technologically innovative, at least in a dynamic sense. This study provides a number of contributions to the literature in technology and innovation. The relationship between size and innovation has long been an important area of research in the literature on technology and innovation. There is a good deal of literature examining firm size and innovation, and there is a good deal of literature investigating innovation within a dynamic context, but research that empirically considers the explicit relationship between firm size and dynamic technological innovation is scarce. In fact, what research does exist differs from our results, as noted above. Although we cannot provide a definitive explanation for this apparent contradiction, we can provide a reasonable first step in this direction. As we noted above, the traditional economic perspective on the Schumpeterian hypothesis addresses the relationship between firm size and the efficiency, or productivity, of the innovative pro-

cess. Indeed, research in this area is often explicitly characterized as investigating whether there are economies of scale in innovation. The concept of “economies of scale” by definition examines the relationship between cost and size. Our research takes a qualitatively different point of view. We examine the relationship between firm size and the effectiveness of the innovation process over time (as reflected by the rate of change in the technical performance of products developed over time). Therefore, our interpretation of these results is not that they are necessarily a contradiction of the evolutionary framework; rather, they may simply address a different research question. In fact, it may be that larger firms were more efficient over time in developing new products than smaller firms, even if these products were not as technically advanced. The next question, which would be the subject of future research, would be whether effectiveness (as we have conceptualized it here) or efficiency of the dynamic innovation process is more important from a competitive perspective. Another possible explanation for why our results differ from the evolutionary framework suggests a second direction for future research. Nelson and Winter (1982, p. 279) acknowledge that their model (and other models of Schumpterian competition) fail to include the possible effects of “bureaucratic control structures” found in large firms. These and other organizational variables that differ among large and small firms may also provide an explanation for our results. Therefore, it may not be size, per se, that is responsible for the difference in innovation performance; it may be that organizational characteristics often found in small firms are the determining factor. In other words, can a large firm “act small” in the innovation process? One way to address this question would be to study dynamic innovation in entrepreneurial divisions or joint ventures of large firms. One direction for future research would therefore be the development of a conceptual model that explicitly includes organizational variables of this sort. Further, such a model would address the possible causal relationships between firm size and dynamic technological innovation. Research in this area would include the develop-

Table 5 Regressions of performance versus year for large and small firms AVGRATE

Firm dummy variables INTERCEPT YEAR R2 a

AVGRATE/$

Small firms

Large firms

Small firms

Large firms

⫺31213***a (3139.5) 1737.1*** (139.1) 0.811

⫺18048*** (3045.8) 1260.1*** (134.9) 0.638

⫺30.447*** (6.245) 2.367*** (0.277) 0.767

⫺19.000*** (3.196) 1.537*** (0.143) 0.774

*p⬍0.05; **p⬍0.01; ***p⬍0.001. Standard errors are shown in parentheses.

G.N. Stock et al. / Technovation 22 (2002) 537–549

ment of a more detailed conceptual model that would explicitly consider the organizational and managerial constructs underlying the effects of firm size. This avenue of research should also include empirical investigation of these relationships. Both conceptually and empirically, research along this line would require analysis at a lower organizational level than the firm. Moreover, the methodological approach would need to include the collection and analysis of primary data from members of the organizations being studied. The development of a detailed approach in this area is beyond the scope of this discussion, however. This paper provides a starting point for a new direction for research in an enduring topic. Innovation is an ongoing process. It is therefore especially appropriate to examine the relationship between size and innovation over time. Schumpeter (1942) explicitly recognized this dynamic characteristic of innovation in his conceptualization of economic change as successive cycles of creative destruction. One of the objectives achieved in this paper has been to provide a foundation for future empirical research examining innovation in such a context.

Acknowledgements The authors would like to acknowledge the support of the Cato Center for Applied Business Research at the University of North Carolina, which provided partial funding for this research.

References Abernathy, W.J., Clark, K.B., 1985. Innovation: mapping the winds of creative destruction. Research Policy 14, 3–22. Abernathy, W.J., Utterback, J.M., 1978. Patterns of industrial innovation. Technology Review 80 (June/July), 40–47. Acs, Z.J., Audretsch, D.B., 1987. Innovation, market structure, and firm size. Review of Economics & Statistics 69, 567–575. Acs, Z.J., Audretsch, D.B., 1988. Innovation in large and small firms: an empirical analysis. American Economic Review 78, 678–690. Acs, Z.J., Audretsch, D.B., 1991a. Innovation and technological change: an overview. In: Acs, Z.J., Audretsch, D.B. (Eds.), Innovation and Technological Change: An International Comparison. University of Michigan Press, Ann Arbor, MI. Acs, Z.J., Audretsch, D.B., 1991b. R&D, firm size, and innovative activity. In: Acs, Z.J., Audretsch, D.B. (Eds.), Innovation and Technological Change: An International Comparison. University of Michigan Press, Ann Arbor, MI. Anderson, P., Tushman, M.L., 1990. Technological discontinuities and dominant designs: A cyclical model of technological change. Administrative Science Quarterly 35, 604–633. Anon., 1971–1994. Standard and Poor’s Register of Corporations. McGraw-Hill, New York, NY. Anon., 1976–1993. Datapro Reports on Data Communications. Data-

547

pro Research, Delran, NJ. Anon., 1986–1994. Corporate Technology Directory. Corporate Technology Information Services, Woburn, MA. Anon., 1987–1994. Ward’s Business Directory of U.S. Private and Public Corporations. Gale Research, Detroit, MI. Anon., 1994. Economic Report of the President. US Government Printing Office, Washington, DC. Anon., 1995. Lean — and frail. Wall Street Journal 5 July 1995, 1. Anon., 1998a. Big banks, big problems. Business Week 3575 (27 April), 134. Anon., 1998b. Reshaping the 1,000: The top 40 mergers of the year. Fortune 137(8), 150–154. Audretsch, D.B., Acs, Z.J., 1991. Innovation and size at the firm level. Southern Economic Journal 57, 739–744. Barnett, B.D., Clark, K.B., 1996. Technological newness: an empirical study in the process industries. Journal of Engineering and Technology Management 13, 263–282. Bernstein, A., 1998. Oops, that’s too much downsizing. Business Week 3581 (8 June), 38. Bound, J., Cummins, C., Griliches, Z., Hall, B.H., Jaffe, A., 1984. Who does R&D and who patents? In: Griliches, Z. (Ed.), R&D, Patents, and Productivity. University of Chicago, Chicago, IL. Chakrabarti, A.K., Halperin, M.R., 1991. Technical performance and firm size: analysis of patents and publications of US firms. In: Acs, Z.J., Audretsch, D.B. (Eds.), Innovation and Technological Change: An International Comparison. University of Michigan Press, Ann Arbor, MI. Christensen, C.M., 1992a. Exploring the limits of the technology Scurve, part I: Component technologies. Production and Operations Management 1 (4), 334–357. Christensen, C.M., 1992b. Exploring the limits of the technology Scurve, part II: Architectural technologies. Production and Operations Management 1 (4), 358–366. Cohen, W.M., Klepper, S., 1996. A reprise of firm size and R&D. Economic Journal 106, 925–951. Colvin, G., 1999. The year of the megamerger. Fortune 139 (1), 62–72. Cooper, R.G., 1979. The dimensions of industrial new product success and failure. Journal of Marketing 43 (3), 93–103. Cooper, R.G., 1990. New products: what distinguishes the winners? Research-Technology Management 33 (6), 27–31. Cooper, R.G., Kleinschmidt, E.J., 1986. An investigation into the new product process: steps, deficiencies, and impact. Journal of Product Innovation Management 3 (2), 71–85. Damanpour, F., 1987. The adoption of technological, administrative, and ancillary innovations: Impact of organizational factors. Journal of Management 13, 675–688. Damanpour, F., 1992. Organizational size and innovation. Organization Studies 13, 375–402. Dodson, E.N., 1985. Measurement of the state of the art and technological advance. Technological Forecasting and Social Change 27, 129–146. Dosi, G., 1982. Technological paradigms and technological trajectories: A suggested interpretation of the determinants and directions of technical change. Research Policy 11, 147–162. Esposito, E., 1993. Technology measurement: a composite approach. Technological Forecasting and Social Change 43, 1–17. Ferguson, C.H., 1988. Beyond entrepreneurialism to U.S. competitiveness: From the people who brought you voodoo economics. Harvard Business Review 66 (3), 55–62. Foster, R., 1986. Innovation: The Attacker’s Advantage. Summit Books, New York. Galbraith, J.K., 1956. American Capitalism: The Concept of Countervailing Power. Houghton Mifflin Company, Boston, MA. Garten, J.E., 1998. Daimler has to steer the Chrysler merger. Business Week 3587 (20 July), 20.

548

G.N. Stock et al. / Technovation 22 (2002) 537–549

Gilder, G., 1988. The revitalization of everything: The law of the microcosm. Harvard Business Review 66 (2), 49–61. Graves, S.B., Langowitz, N.S., 1993. Innovative productivity and returns to scale in the pharmaceutical industry. Strategic Management Journal 14, 593–605. Greising, D., 1998. It’s the best of times—or is it? Business Week 3560 (12 January), 36–38. Greising, D., Galuszka, P., Morris, K., Osterland, A., Smith, G., 1998. $1,000,000,000,000 Banks. Business Week 3575 (27 April), 32–39. Griliches, Z., 1990. Patent statistics as economic indicators: a survey. Journal of Economic Literature 28, 1661–1707. Halperin, M.R., Chakrabarti, A.K., 1987. Firm and industry characteristics influencing publications of scientists in large American companies. R&D Management 17, 167–173. Harrington, A., Rao, R.M., Urresta, L., Maroney, T., 1998. Mergers: Why this historic boom will keep making noise. Fortune 137 (8), 148–156. Harrison, B., 1994. Lean and Mean: The Changing Landscape of Corporate Power in the Age of Flexibility. Basic Books, New York, NY. Henderson, R., 1995. Of life cycles real and imaginary: The unexpectedly long old age of optical lithography. Research Policy 24, 631–643. Henderson, R.M., Clark, K.B., 1990. Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly 35 (1), 9–30. Henderson, R., Cockburn, I., 1996. Scale, scope, and spillovers: the determinants of research productivity in drug discovery. Rand Journal of Economics 27, 32–59. Jensen, E.J., 1987. Research expenditures and the discovery of new drugs. Journal of Industrial Economics 36, 83–95. Johnston, J., DiNardo, J., 1997. Econometric Methods. McGraw-Hill, New York. Kamien, M.I., Schwartz, N.L., 1975. Market structure and innovation — a survey. Journal of Economic Literature 13, 1–37. Kamien, M.I., Schwarz, N.L., 1982. Market Structure and Innovation. Cambridge University Press, Cambridge. Khanna, T., 1995. Racing behavior: Technological evolution in the high-end computer industry. Research Policy 24, 933–958. Kleinbaum, D.G., Kupper, L.L., Muller, K.E., 1988. Applied Regression and Other Multivariable Methods. PWS-Kent Publishing Company, Boston, MA. Koretz, G., 1998. Downsizing’s painful effects. Business Week 3573 (13 April), 23. Kupfer, A., 1998. MCI WorldCom: It’s the biggest merger ever. Fortune 137 (8), 118–128. Lichtenberg, F.R., Siegel, D., 1991. The impact of R&D investment on productivity—new evidence using linked R&D–LRD data. Economic Inquiry 29, 203–228. Loomis, C.J., 1999. Citigroup: Scenes from a merger. Fortune 139 (1), 76–88. Mansfield, E., 1980. Basic research and productivity increase in manufacturing. American Economic Review 70, 863–873. Methe, D.T., 1992. The influence of technology and demand factors on firm size and industrial structure in the DRAM market — 1973– 1988. Research Policy 21, 13–25. Narin, F., Noma, E., 1987. Patents as indicators of corporate technological strength. Research Policy 16, 143–155. Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory of Economic Change. Belknap-Harvard Press, Cambridge, MA. Pavitt, K., Robson, M., Townsend, J., 1987. The size distribution of innovating firms in the UK: 1945–1983. Journal of Industrial Economics 35, 297–316. Peterson, D.K., Miller, P.E., Fischer, W.A., Zmud, R.W., 1987. Technology measurement and the appraisal of information technology. Technological Forecasting and Social Change 42, 251–259.

Porter, M., 1985. Technology and competitive advantage. In: Porter, M. (Ed.), Competitive Advantage. The Free Press, New York, NY. Prahalad, C.K., Hamel, G., 1990. The core competence of the corporation. Harvard Business Review 68 (3), 79–91. Scherer, F.M., 1965. Firm size, market structure, opportunity, and the output of patented inventions. American Economic Review 55, 1097–1125. Schumpeter, J.A., 1934. The Theory of Economic Development. Harvard University Press, Cambridge, MA. Schumpeter, J.A., 1942. Capitalism, Socialism, and Democracy. Harper, New York. Schwalbach, J., Zimmerman, K.F., 1991. A Poisson model of patenting and firm structure in Germany. In: Acs, Z.J., Audretsch, D.B. (Eds.), Innovation and Technological Change: An International Comparison. University of Michigan Press, Ann Arbor, MI. Solow, R., 1957. Technical change and the aggregate production function. Review of Economics and Statistics 39, 312–320. Taylor, A., 1999. The automakers: More mergers. Fortune 139 (3), 26–27. Tushman, M.L., Anderson, P., 1986. Technological discontinuities and organizational environments. Administrative Science Quarterly 31, 439–465. Tushman, M.L., Rosenkopf, L., 1992. Organizational determinants of technological change: Toward a sociology of technological evolution. Research in Organizational Behavior 14, 311–347. Utterback, J.M., Abernathy, W.J., 1975. A dynamic model of process and product innovation. Omega 3 (6), 639–656.

Gregory N. Stock is Assistant Professor in the College of Business at Northern Illinois University. His research has focused on technology and supply chain management. His recent articles have examined technology transfer, manufacturing technology implementation, product development, and new organizational approaches to supply chain management and have been published in journals such as IEEE Transactions on Engineering Management, the Journal of Operations Management, the Journal of High Technology Management Research, Production and Inventory Management Journal, and the International Journal of Operations and Production Management. Prior to beginning his academic career, Dr. Stock spent several years in industry as a design engineer in high technology industries such as computer graphics and data communications. He has B.S. and M.S. degrees in electrical engineering from Duke University and a Ph.D. degree in operations management from the University of North Carolina. Dr. Stock has taught undergraduate and graduate courses in operations management, supply chain management, and technology management at a variety of institutions, including Northern Illinois University, Arizona State University, and the China–Europe International Business School. Noel P. Greis received an A.B. degree in mathematics from Brown University, and M.S., M.A., and Ph.D. degrees in engineering from Princeton University. Dr. Greis is Director of the Center for Logistics and Global Strategy in the Kenan Institute of Private Enterprise and Adjunct Assistant Professor of Operations, Technology, and Innovation Management in the Kenan–Flagler Business School at the University of North Carolina at Chapel Hill. Dr. Greis was previously a Member of the Technical Staff at AT&T Bell Laboratories and Bell Communications Research. Her research interests focus on global logistics and the management of technology, and her work has been published in IEEE Transactions on Engineering Management, Research Policy, Decision Sciences, California Management Review, and the International Journal of Operations and Production Management. William A. Fischer is a Professor of Technology Management at IMD. He received a D.B.A. degree from George Washington University. His principal teaching and research interests involve the management of technology, including the management of the creative processes within R&D; the creation and coordination of an international technology presence; and technology transfer. Dr. Fischer has worked in the steel industry, the US

G.N. Stock et al. / Technovation 22 (2002) 537–549 Army Corps of Engineers, and has also worked as a consultant on R&D/technology issues in such industries as: pharmaceuticals, telecommunications, textiles and apparel, and packaging. Additionally, he has served as a consultant to a number of government and international-aid

549

agencies such as the World Health Organization. In addition to IMD, he has been on the faculties of Clarkson University and the University of North Carolina, and was the Executive President and Dean of the China– Europe International Business School.