594847

research-article2015

JOMXXX10.1177/0149206315594847Journal of ManagementWang, Chen

Journal of Management Vol. XX No. X, Month XXXX 1­–27 DOI: 10.1177/0149206315594847 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav

Capability Stretching in Product Innovation Tang Wang University of Central Florida

Yan Chen Stevens Institute of Technology

Product innovation is conventionally treated as a mechanism for organizations to renew their product portfolios. In this paper, we suggest that product innovation not only enables organizations to introduce new products to the market but also challenges organizations to renew their technological capabilities. Capability stretching is the degree to which an organization extends its technological capabilities to bridge the gap between what it has already known and what the development of a new product requires it to know. Capability stretching can be challenging because it involves the acquisition and assimilation of new and distant knowledge. Drawing on a longitudinal study of product introductions in the workstation industry, we find that capability stretching reduces the chance of new product survival. Furthermore, we also find that organizational boundaries moderate the negative relationship between capability stretching and product survival: Vertical integration exacerbates this negative relationship, whereas horizontal boundary mitigates this negative relationship. However, capability stretching can also be rewarding, as it renews technological capabilities and therefore facilitates adaptation to technological changes. We draw implications for the linkages between product innovation and capability development. Keywords: dynamic capabilities; product innovation; product survival; organizational boundary; vertical integration; horizontal boundary

Acknowledgments: The authors have contributed equally. We would like to thank Catherine A. Maritan, the senior associate editor, and two anonymous reviewers for their extremely helpful comments throughout the development of this paper. We are also grateful to Olav Sorenson for sharing the workstation data set through the Firm and Industry Evolution and Entrepreneurship (FIVE) project. We also would like to thank Constance Helfat, Murad Mithani, Olav Sorenson, and Sunny Li Sun for their comments on earlier drafts of this paper. An earlier version of this paper was presented at the 2013 annual meeting of the Academy of Management, the 2013 Darden and Judge Entrepreneurship and Innovation Research Conference, and the 2013 Industry Studies Conference, and we are grateful to conference participants for their comments and suggestions. Certain data included herein are derived from the Sorenson Workstation FIVE data file; any opinions, findings, and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of the FIVE project. Corresponding author: Tang Wang, Department of Management, College of Business Administration, University of Central Florida, 12744 Pegasus Drive, Orlando, FL 32826, USA. E-mail: [email protected] 1

2   Journal of Management / Month XXXX

Over time, organizational capabilities change. . . . But this fact in no way diminishes the significance of the limits on what particular firms are capable of doing at any time, and the constraints on the range of new things that they can learn to do in a reasonable period of time. (Dosi & Nelson, 2010: 82)

Dynamic capabilities enable firms to build, extend, and rebuild organizational resources and capabilities (Eisenhardt & Martin, 2000; Helfat et al., 2007; Teece, 2007; Teece, Pisano, & Shuen, 1997). New product development capability is a dynamic capability that enables firms to introduce new products to the market and reconfigure their product portfolios, with the aim of improving their product market performance (Brown & Eisenhardt, 1995, 1997; Eisenhardt & Tabrizi, 1995; Schilke, 2014). In the process of developing new products, firms may also acquire new knowledge and assimilate new technologies, contributing to the renewal of technological capabilities (Cohen & Levinthal, 1994; Danneels, 2002; Helfat & Raubitschek, 2000). New product development, therefore, may be a critical organizational process through which firms renew product portfolios and extend technological capabilities. However, these two important roles of new product development may sometimes conflict with each other. When the development of a new product requires a firm to extend its technological capabilities, for example, the resulting product may sometimes not perform very well in the product market because the firm may lack certain prior knowledge and technological capabilities to make the product successful (Cohen & Levinthal, 1990; Denrell & March, 2001). In this article, we define capability stretching as the degree to which a firm extends its technological capabilities to bridge the gap between what it has already known and what the development of a new product requires it to know. On one hand, capability stretching may negatively affect a firm’s product market performance in the short run; on the other hand, it may help the firm renew its knowledge base, extend its technological capabilities, and become better prepared for future product innovation (Cohen & Levinthal, 1994; Denrell & March; Wernerfelt, 1984). In this study, we focus on the impact of capability stretching on new product survival. Such a focus can help better capture the tension between capability stretching and product market performance (March, 2008, 2010). For example, a product can be of strategic importance to a firm because it extends the firm’s technological capabilities, even if it may not be able to achieve immediate market success (Schilling, 2012). A product that is not a market success, therefore, can still provide a firm with important opportunities to acquire new knowledge and develop new capabilities (Wernerfelt, 1984). Although new product survival has been examined from many theoretical perspectives (Carroll, Khessina, & McKendrick, 2010), we still know very little about how capability stretching affects new product survival. We believe that examining the impact of capability stretching on new product survival may help us better understand the tension between product market performance and capability development and may also allow us to identify mechanisms through which firms manage this tension (Denrell & March, 2001; March, 1991; Wernerfelt). In this study, we examine how organizational boundaries moderate the relationship between capability stretching and new product survival. Organizational boundaries reflect not only the boundaries of governance but also the boundaries of knowledge (Argote, 2012;

Wang, Chen / Capability Stretching in Product Innovation   3

Brusoni, Prencipe, & Pavitt, 2001; Carlile, 2004; Kapoor & Adner, 2012). Organizational boundaries reflect the extent to which product knowledge is controlled by a firm and therefore may influence the difficulty of capability stretching (Afuah & Tucci, 2012). We examine how the vertical and horizontal boundaries of a firm affect the relationship between capability stretching and new product survival, with the aim of uncovering the roles of organizational boundaries in resolving the tension between improving product market performance and extending technological capabilities (Argyres, 1996, 2011; Argyres, Felin, Foss, & Zenger, 2012; Argyres & Zenger, 2012). We draw on a longitudinal sample of firms in the computer workstation industry to examine the nature and consequences of capability stretching in product innovation. The workstation industry was a vibrant, high-tech industry in the 1980s and 1990s; in those days, developing a new workstation required substantial technological capabilities (Kidder, 1981). The workstation industry is appropriate for this study because product features and technological capabilities can be explicitly measured in this industry and because products entered and exited the market regularly. Empirical results show that capability stretching reduces the chance of new product survival. Furthermore, we find that this relationship is moderated by organizational boundaries. In particular, vertical integration exacerbates the negative relationship between capability stretching and new product survival, whereas horizontal boundary mitigates this negative relationship. We make two unique contributions to the literature. First, we contribute to the understanding of the tension between extending technological capabilities and improving immediate market performance in product innovation. This important tension has not been clearly shown in prior studies; this is probably because prior studies have focused primarily on product portfolios without distinguishing the varying degrees of capability stretching among products or because prior studies have focused primarily on organizational performance rather than product performance. In our study, we theorize and test a model examining the impact of capability stretching on new product survival, contributing new insights to the management of product innovation. Second, we contribute to our understanding of the roles of organizational boundaries in resolving this tension. We argue and find that the vertical and horizontal boundaries of a firm play different roles in resolving this tension, contributing new insights to the understanding of how organizational boundaries affect the development and evolution of organizational capabilities.

Theory Development and Hypothesis As an important aspect of dynamic capability, the new product development process is fundamentally an organizational search process (Katila, 2002; Katila & Ahuja, 2002; Nelson & Winter, 1982). Organizational search is a process through which organizations construct alternatives and evaluate alternatives (Cyert & March, 1992; Knudsen & Levinthal, 2007; Simon, 1955). Organizational search can be local or distant, and distant search is generally much more difficult than local search (Afuah & Tucci, 2012; Cyert & March; Levinthal & March, 1993; Levitt & March, 1988; March, 1991, 2010; Nelson & Winter). Numerous studies have documented that firms are good at local search but poor at distant search (Afuah & Tucci; Cyert & March; Nelson & Winter), good at exploitation but poor at exploration (Levinthal & March; Levitt & March; March, 1991, 2006, 2010),

4   Journal of Management / Month XXXX

and good at competence-enhancing innovation but poor at competence-destroying innovation (Anderson & Tushman, 1990; Gatignon, Tushman, Smith, & Anderson, 2002; Tushman & Anderson, 1986). In order to understand the challenges that firms may face in technological search, it is important to recognize that firms are boundedly rational entities (Cyert & March, 1992; March & Simon, 1993; Simon, 1997). Because of bounded rationality, firms can carry out search effectively in technological areas in which they have prior knowledge (Cohen & Levinthal, 1989, 1990); however, they may encounter substantial difficulties when they carry out search in areas in which they lack prior knowledge (Afuah & Tucci, 2012; Cohen & Levinthal, 1990; Denrell & March, 2001). Drawing on these insights, we are interested in exploring the challenges that firms face when they engage in distant search and the mechanisms through which they cope with such challenges. Capability stretching in new product development requires firms to carry out organizational search in technological areas with “effective distance of knowledge.” Afuah and Tucci define the effective distance of knowledge as “the effective distance between a focal agent’s existing knowledge and the knowledge required to solve a problem” (2012: 364). In other words, it reflects the gap between a firm’s current technological knowledge and the technological knowledge that is required to develop a new product (Afuah & Tucci; Schilling, 2012). The effective distance of knowledge is also referred to as capability gap in the literature (Capron & Mitchell, 2009; Day, 2011; Schilling). Facing effective distance of knowledge (or capability gap), firms need to acquire new knowledge and extend their technological capabilities in order to successfully develop new products (Coen & Maritan, 2011; Danneels, 2002; Helfat & Raubitschek, 2000). As the effective distance of knowledge increases, however, firms may face difficulties in assimilating and exploiting new knowledge (Cohen & Levinthal, 1989, 1990, 1994; Denrell & March, 2001). Consequently, a firm’s existing absorptive capacity may limit the extent to which it can acquire new knowledge in the short run; in the long run, however, capability stretching expands the firm’s absorptive capacity, which in turn may enhance the firm’s capacity to acquire, assimilate, and exploit new knowledge.1 Firms therefore face a dilemma in extending their technological know-how. In the short run, capability stretching may result in poor performance. When firms stretch their technological know-how, they need to search in areas in which they lack prior knowledge (Afuah & Tucci, 2012) and work with unfamiliar knowledge elements (Fleming, 2001). Distant search and unfamiliarity give rise to challenges associated with searching, acquiring, and integrating new knowledge and may result in a temporary performance decline (Afuah & Tucci; Fleming; Katila & Ahuja, 2002). In the long run, however, if firms fail to stretch their technological know-how, their technological capabilities will quickly fall behind the industry frontiers, and they may not be able to introduce competitive new products to the market (Danneels, 2002; Leonard-Barton, 1992; Levinthal & March, 1993; March, 1991; Maritan & Brush, 2003).

Capability Stretching and Product Survival Capability stretching is the degree to which a firm extends its technological know-how in order to develop a new product. The larger the degree of capability stretching, the larger the

Wang, Chen / Capability Stretching in Product Innovation   5

effective distance of knowledge (Afuah & Tucci, 2012). In other words, firms with large effective distance of knowledge have low prior knowledge and will need to stretch their technological capabilities in order to develop new products (Afuah & Tucci; Coen & Maritan, 2011; Nelson & Winter, 1982). As a result, the larger the effective distance of knowledge is, the more difficult it is for a firm to carry out new combinations. As Nelson and Winter point out, “It is time-consuming and costly for a firm to learn about, and learn to use, technology significantly different from that with which it is familiar” (237). Such difficulty may result in poor product market performance of a new product, as reflected by the low chance of product survival in the market (Helfat et al., 2007). We therefore expect capability stretching to be negatively associated with new product survival. The negative impact of capability stretching on product survival arises mainly from the unfamiliarity with new knowledge (Dierickx & Cool, 1989; Fleming, 2001; Katila & Ahuja, 2002). The unfamiliarity with new technological knowledge results in technological uncertainty and unreliability (Fleming; Katila & Ahuja). When firms try to extend their product features and technological know-how by adding new knowledge elements, they are experimenting with unfamiliar knowledge elements and unfamiliar combinations (Fleming). In other words, they engage in distant search in areas in which they lack prior knowledge (Afuah & Tucci, 2012). When firms engage in distant search and work with unfamiliar knowledge elements in new product development, the resulting products tend to contain bugs and defects, reducing the reliability and usefulness of the new products (Katila & Ahuja). In addition, unfamiliarity may lead to the mismatch between the technical features and customer needs. Therefore, when firms try to stretch their technological capabilities by incorporating a large amount of new knowledge into a single new product, the product may perform poorly in the market as a result of either technology failure or market mismatch. It is important to note that products involving high degrees of capability stretching are not necessarily radical innovations in the marketplace.2 Truly radical innovations must be new to the market (Sorescu, Chandy, & Prabhu, 2003), while products involving capability stretching can just be new to the firms producing them. Prior research has documented that radical innovations can sometimes contribute to firm performance by creating product differentiation, brand image, or first-mover advantages (Schilling, 2012; Sorescu et al.). However, products involving high degrees of capability stretching are not necessarily radical innovations in the marketplace, although they may be radically new to the firms producing them. We therefore hypothesize the following: Hypothesis 1: Capability stretching reduces the chance of product survival.

Vertical Integration Organizational boundaries are not only boundaries of governance but also boundaries of knowledge (Argote, 2012; Brusoni et al., 2001; Carlile, 2004; Kapoor & Adner, 2012). Organizational boundaries reflect the extent to which product knowledge is controlled within a firm and therefore influence the effective distance of knowledge that needs to be overcome in order to develop a new product (Afuah & Tucci, 2012). In this study, we are interested in how the vertical and horizontal boundaries of a firm affect the relationship between capability stretching and product survival.

6   Journal of Management / Month XXXX

The vertical boundary of a firm is “defined by the scope of activities undertaken in the industry value chain” (Santos & Eisenhardt, 2005: 492); it reflects the extent to which a firm has control over the different components of a product (Lafontaine & Slade, 2007; Williamson, 1985). Vertically integrated firms rely on internal development to supply components to a product (Afuah & Tucci, 2012; Sorenson, 2003). Less vertically integrated firms rely more on their external suppliers to provide components to a product (Afuah, 2000, 2001). Vertical integration has both advantages and disadvantages. One major advantage associated with vertical integration is enhanced coordination (Williamson, 1985). If most components of a product are produced and controlled internally, the firm may be able to achieve better coordination among different organizational units and therefore better integration among product components (Afuah, 2000). Enhanced coordination may result in better product performance and therefore enable the new product to survive and succeed in the market. We therefore expect vertical integration to have a direct positive impact on product survival (Sorenson, 2003). In this study, however, we are not interested in its direct effect but its moderating effect on the relationship between capability stretching and product survival. A vertically integrated firm, for example, may develop most components of a product internally and may face severe difficulties in stretching its technological capabilities. One primary reason is that vertical integration affects the firm’s flexibility to switch to new partners when developing new products (Christensen & Raynor, 2003; Ozcan & Eisenhardt, 2009). Highly vertically integrated firms generally have to work with their existing internal units to develop new products. When they try to develop products that require significant capability stretching, they may lack prior knowledge in various areas and therefore need to stretch their technological knowledge in various areas simultaneously.3 The more areas they control internally, the more capability gaps they may face. As a result, vertically integrated firms may face difficulties in acquiring and assimilating new knowledge to close their capability gaps in multiple areas, thus resulting in poor product performance (Afuah & Tucci, 2012; Schilling, 2012). Less vertically integrated firms, however, rely more on external partners to supply needed technologies; they may also have the flexibility to switch to new suppliers who happen to have new technologies that are needed for new products (Afuah, 2000, 2001). In the workstation industry, for example, Afuah (2000, 2001) has also suggested that firms that are stuck with partners with obsolete technologies may become unable to adapt to technological changes. In sum, the vertical boundary of a firm increases the difficulties of capability stretching and, thus, exacerbates the negative relationship between capability stretching and product survival. We therefore hypothesize the following: Hypothesis 2: Vertical integration strengthens the negative association between capability stretching and product survival.

Horizontal Boundary The horizontal boundary of a firm is “defined by the scope of product/markets addressed” (Santos & Eisenhardt, 2005: 492). A large horizontal boundary means that a firm competes in multiple product categories (Santos & Eisenhardt). A large horizontal boundary is associated with both diseconomies and economies. Its diseconomies result mainly from the loss of

Wang, Chen / Capability Stretching in Product Innovation   7

focus (McDermott & Stock, 2011), as products in different categories may compete for organizational resources and attention. As a result, firms with products in many different categories may fail to concentrate resources on certain new products or fail to clarify organizational priorities for new products (McDermott & Stock). Because of diseconomies of scope, we expect that horizontal boundary has a direct negative impact on product survival. In this study, however, we are not interested in its direct effect but, instead, its moderating effect on the relationship between capability stretching and product survival. We argue that horizontal boundary gives rise to certain economies of scope, some of which may facilitate distant search and capability stretching. One major source of economies of scope is internal technology spillovers and transfers (Henderson & Cockburn, 1996). Internal technology spillovers and transfers can reduce the difficulty of capability stretching because technological knowledge accumulated in other product categories may be transferred to the current product category and, thus, reduce the effective distance of knowledge and the difficulty of capability stretching (Henderson & Cockburn). As the difficulty of capability stretching declines, capability stretching may not severely affect product performance and product survival. Another source of economies of scope may arise from a broadened scope for external search (Laursen & Salter, 2006; Leiponen & Helfat, 2010). Firms with products in multiple product categories may have opportunities to work with different external partners, broadening their access to external knowledge. When firms try to stretch their technological capabilities, they may be able to capitalize on their broad access to external knowledge to acquire knowledge and technologies that are needed to close their capability gaps, enabling them to overcome the difficulty in capability stretching (Leiponen & Helfat; Ozcan & Eisenhardt, 2009). As a result, horizontal boundary may mitigate the negative impact of capability stretching on new product survival. We therefore hypothesize the following: Hypothesis 3: Horizontal boundary weakens the negative association between capability stretching and product survival.

Method Sample To test these hypotheses, we employed a sample of firms in the computer workstation industry; the workstation industry has been the focus of many prior studies (Afuah, 2000, 2001; Sorenson, 2000, 2003; Sorenson, McEvily, Ren, & Roy, 2006; Wolter & Veloso, 2008). The data set was originally collected by Professor Olav Sorenson and was made publicly available through the Firm and Industry Evolution and Entrepreneurship project (Sorenson, 2000, 2003; Sorenson et al.). In this data set, computer workstations were defined as distributed computing machines for single users. The primary data source was a publication called Data Sources, which cataloged products in the computer industry. Data Sources identified all machines that could be considered as workstations; this primary source of data was supplemented with a few other sources of data, including corporate reports, product advertisements, and product announcements that could be found in the Institute of Electrical and Electronics Engineers Graphical Computing and Applications, the International Data Corporation Processor Survey, and LexisNexis searches.

8   Journal of Management / Month XXXX

This final data set contains all firms in North America producing workstations between 1980 and 1996. Following Sorenson (2000, 2003), this study focuses on the period between 1980 and 1996 when the industry was a vibrant industry offering workstations that were much more powerful and advanced than personal computers (PCs). During this period, workstations had better technical performance than PCs, especially with respect to the central processing unit (CPU) and random access memory (RAM). This study focuses on a workstation’s technical features along two core components: the CPU and RAM. These two critical components determined the requirements for other hardware components and software applications. Changes in CPU and RAM often required the design of new hardware components and the redevelopment of software applications. To stay competitive in the industry, firms needed to constantly improve CPU clock speed and RAM capacity; as a result, firms also needed to make changes to other hardware components and software applications. During this period, CPU clock speed and RAM capacity went through substantial improvements (see Figure 1).

Dependent Variable The performance of a new product is measured by the tenure of the product in the market in terms of years; it is a standard measure for product survival in the product demography literature (Khessina & Carroll, 2008). Product survival reflects the evolutionary fitness of a product in the market (Helfat et al., 2007: 16). In this database, two variables are related to a product’s tenure: a product’s entry year (p_ entry) and its exit year (p_exit). In the workstation industry, prior studies have treated year 1996 as the natural end point for all workstation products (Sorenson, 2000, 2003; Sorenson et al., 2006) because the distinctions between computer workstations and PCs were erased by the arrival of the Windows NT operating system.4 As the differences between the two categories narrowed, the workstation industry began to decline and disappear. With the arrival of the Windows NT, year 1996 became the end point of the industry, and all products entering in 1996 were considered to have very short product tenure. To test the robustness of the results with respect to the industry’s end year, we ran a robustness test in which we eliminated all the products entering the industry in 1996; in this robustness test, the results are qualitatively consistent. In the empirical analysis, we also need to take into account the potential impact of product generation on product tenure. The computer workstation industry went through two major generations: complex instruction set computer (CISC) and reduced instruction set computer (RISC). RISC technology did not appear in the industry until 1988 when Sun Microsystems began to promote the RISC technology. It became the dominant design in 1993. If the average product tenure varies significantly from generation to generation, we may need to account for this effect on product tenure. To address this issue, we included a dummy variable in the model, which indicates whether a computer workstation uses the CISC architecture or the RISC architecture, to directly account for the generational effect.5

Independent Variable Capability stretching is the degree to which a firm extends its technological know-how in order to bridge the effective distance of knowledge between what it has already known and

Wang, Chen / Capability Stretching in Product Innovation   9

Figure 1 Evolution of Random Access Memory Capacity and Central Processing Unit Clock Speed in the Workstation Industry Random Access Memory

Amount of Memory Capacity (in Kilobytes)

600000 500000 400000 300000 200000 100000 0

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Year

Clock Speed of Machine on Cycles (in Megahertz)

120

Clock Speed

100 80 60 40 20 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Year

what it needs to know to develop a new product. To measure capability stretching, we constructed a variable to reflect the difference between the technology incorporated in the newly developed product and the technology used in prior products of the same firm (Agarwal, Echambadi, Franco, & Sarkar, 2004). We also accounted for the industry’s technology frontier in this measure because the pace of technological improvement changes as the technology frontier of the industry shifts (Nelson & Winter, 1982; Rockart & Dutt, 2015; Schilling, 2012). We first measured capability stretching on the basis of two separate dimensions and then combined the two dimensions into a single index. The two dimensions are RAM capacity and CPU clock speed. Drawing on Agarwal et al. (2004), we measured capability stretching on the RAM dimension as follows:

10   Journal of Management / Month XXXX



Stretch (RAM )i , j ,t =

RAM i , j ,t − max {RAM i , ji ,ti } ji≤ j −1, ti≤t

(1)

max {RAM ia , ja ,ta }

ia , ja ,ta≤tk

In this measure, RAM i , j ,t refers to the RAM capacity (in kilobytes) of machine j introduced by firm i, t is a subscript for time, and max {RAM i , ji ,ti } refers to the focal firm’s maximum ji ≤ j −1, ti ≤ t

RAM capacity among all the prior products that firm i has ever produced before machine j. Let us assume that the maximum level is found in machine k, which was introduced to the market in year tk. The denominator, max {RAM ia , ja ,ta } , refers to the industry’s maximum RAM capacity ia , ja , ta ≤ tk

among all products that all firms in the industry have ever produced in or before year tk. There are two reasons for measuring the industry frontier of max {RAM ia , ja ,ta } as ia , ja ,ta ≤ tk

before machine k rather than before machine j. First, this measure allows us to take care of cases when the most advanced machine in the industry enters the market immediately before the focal firm’s machine j. In such cases, the focal firm might not be able to track the technological changes in the industry when it was developing the new product (Nelson & Winter, 1982). Second, it takes time to develop a new product, and the year a product entered the market is not the year the product development process started (Schilling, 2012). Therefore, we measured the technology frontier of the industry when the focal firm’s most advanced machine, machine k, was introduced to the market. The second dimension is the CPU clock speed and was measured in a similar way: Stretch(CL)i , j ,t =



CLi , j ,t − max {CLi , ji ,ti } ji≤ j −1,ti≤t

max {CLia , ja ,ta }



(2)

ia , ja ,ta≤tk

In this measure, CLi , j ,t refers to CPU clock speed (in megahertz) of machine j introduced by firm i, t is a subscript for time, and max {CLi , ji ,ti } refers to the focal firm’s maximum CPU ji ≤ j −1, ti ≤ t

clock speed of all products that firm i has ever produced before machine j. Let us assume that the maximum level is found in machine k, which was introduced to the market in year tk. The denominator, max {CLia , ja ,ta } , refers to the industry’s maximum CPU clock speed among ia , ja , ta ≤ tk

all products that all firms in the industry have ever produced in or before year tk. To better capture capability stretching in product innovation, we combined the two dimensions into a single measure of capability stretching. We first standardized the two variables, Stretch (RAM )i , j ,t and Stretch (CL)i , j ,t , and then added them together to create a combined index for capability stretching. Therefore, the combined index serves as the final measure of capability stretching in product innovation.

Moderating Variables A typical computer workstation has eight component categories: CPU, RAM, operating system, software applications, communications hardware, monitor, hard disk drive, and motherboard (Sorenson, 2003). A firm can have multiple component products in each component category. Vertical integration was measured by the total number of component products that a firm internally produces divided by the number of component categories in a computer workstation.6 A high degree of vertical integration means that a firm has internally produced a large number of component products.

Wang, Chen / Capability Stretching in Product Innovation   11

Horizontal boundary was measured by the number of product categories in which the firm competes (Santos & Eisenhardt, 2005). Some firms, for example, compete not only in the workstation industry but also in some other industries. Similar measures have been used in prior studies (Jacquemin & Berry, 1979; Santos & Eisenhardt). We lagged both variables by one period.

Control Variables Last capability.  We controlled for a firm’s prior capability stock (in terms of RAM capacity and CPU clock speed). We created indicators for the highest level of capabilities of the firm (in terms of RAM capacity and CPU clock speed, respectively) before it introduced the current product, standardized the indicators, and then added them together to create a combined index for the prior capability stock. We controlled for last capability in order to separate the effect of capability stretching from the effect of capability stock on product performance. Density. We have controlled for the competitive intensity in the industry. Following Sorenson (2000, 2003), we measured density by the number of competitors in the workstation market. As the number of firms in the industry increases, competitive intensity increases. Frequency of new product introduction.  We included a control variable measuring the frequency of product introductions by the focal firm. The frequency was measured by the number of years elapsed between the firm’s first product and its last product, which was divided by the total number of products the firm has introduced during this period. This measure captures a firm’s frequency of introducing new products to the market. That is, the larger this measure is, the less frequently a firm introduces new products to the market. This measure helps control for the impact of frequent product introductions on the dependent variable, product tenure. First mover.  We included a dummy variable indicating whether a firm belongs to the group of firms entering the workstation industry at the industry’s inception in 1980. It has been found that products introduced by late entrants may display higher exit rates (Khessina & Carroll, 2008). Firm size.  We measured firm size by the natural log of a firm’s revenue in the year when the product was introduced. Previous studies have shown that sales revenue may affect product survival (Chisholm & Norman, 2006). Silicon Valley.  We included a dummy variable taking the value of 1 if the firm’s headquarters reside in the primary metropolitan statistical areas of San Francisco, San Jose, or Oakland. Subsidiary.  We included a dummy variable taking the value of 1 if the firm is a subsidiary and 0 otherwise. Prior research suggests that subsidiaries and independent firms may have different organizational strategies, structures, and processes for product introductions and

12   Journal of Management / Month XXXX

product exits, leading to different rates of product exit (Khessina & Carroll, 2008). Therefore, we included this dummy variable to distinguish subsidiaries from independent companies. Product price. We also included a few product-level control variables (Carroll et al., 2010). We included product price, adjusted for inflation rate, as a control variable because prior research has shown that product price affects the market performance of a product. Unix operating systems.  We included a dummy variable indicating whether the product uses the Unix operating systems. RISC.  We included a dummy variable indicating whether the microprocessor of a product uses the RISC architecture. Year dummies.  We included year dummies to control for potential time effects on product tenure.

Models Because our dependent variable measures product tenure, we chose the survival model to analyze the data. Specifically, we chose the semiparametric Cox model because this model does not rely on strong assumptions about the baseline hazards function (Allison, 2010; Cox, 1972). Cox proportional models allow us to estimate beta without specifying or estimating the baseline hazard—λ0(t). The basic model is defined as follows:

Log h i ( t; Xi ) = λ 0 (t) + Xiβ

(3)

In this model, hi(t; Xi) is the hazard rate for product i at time t, the odds that product i might exit at time t given that it has survived until time t – 1. The base hazard rate is λ0(t), and Xi is the vector of covariates. In this study, we included capability stretching, organizational boundaries, and other control variables as covariates. The vector of regression coefficients is β. By default, we used the Breslow method. Since product is our unit of analysis, interdependence may exist among products introduced by the same firm (De Figueiredo & Kyle, 2006; Khessina & Carroll, 2008; Nerkar & Roberts, 2004). In SAS, PROC PHREG includes the COVSANDWICH option to correct for dependence in models with repeated observations (Allison, 2005); COVSANDWICH corrects for interdependence by using the robust variance estimator (also known as the modified sandwich estimator). To implement this option, we included the AGGREGATE suboption and an ID statement (Allison, 2010).

Results Table 1 shows the descriptive statistics. The sample includes 724 product-year observations. Table 1 shows that the average product longevity in the workstation industry is about 2 to 3 years, with the range between 0 and 8 years. Table 2 shows the regression results. Model 1 is the baseline model with control variables. In Model 2, we added the main independent variable, capability stretching. In

13

1.00 .23** –.20**

.31** .38** –.29** .04 .05 –.19** –.13** .21** .31** –.47** 0 1.63

1.00 .19** –.04 –.04

.27** .01 .00 .00 –.21** –.09* –.30** –.16** –.17** –.13** 2.14 1.53

2

Note: RISC = reduced instruction set computer. †p < .10 (two-tailed test). *p < .05 (two-tailed test). **p < .01 (two-tailed test).

  1. Product survival   2. Last capability   3. Density   4. Frequency of new product introduction   5. Adjusted price   6. Natural log sales   7. Silicon Valley   8. Subsidiary    9. Unix operating system 10. First mover 11. RISC architecture 12. Horizontal boundary 13. Vertical integration 14. Capability stretching Mean Standard deviation

1

.04 .14** .05 .07* –.01 .16** .03 .12** .09* –.21** 55.31 11.9

1.00 –.04

3

.08* –.11** .17** .18** –.19** –.25** –.16** –.30** –.27** .30** 0.13 0.13

1.00

4

1.00 .11** –.05 –.05 –.20** –.15** –.20** –.15** –.10** .07† 9.33 0.91

5

1.00 –.16** .14** .22** –.02 .12** .51** .60** –.27** 20.26 4.51

6

1.00 .26** –.09* .09* –.05 –.23** –.24** .16** 0.05 0.22

7

8

1.00 –.31** –.48** –.22** –.17** –.05 .00 0.15 0.36

Table 1 Descriptive Statistics (n = 724)

1.00 .57** .58** .35** .28** –.11** 0.76 0.43

9

1.00 .37** .13** –.03 –.02 0.87 0.34

10

1.00 .18** .29** –.12** 0.55 0.5

11

1.00 .71** –.21** 8.05 6.61

12

1.00 –.36** 3.7 2.96

13

                  1.00 0 1.58

       

14

14

–0.09 –0.09 –0.09 –0.11 0.41* 0.31 –0.25 0.15 0.04 –0.01

Last capability Adjusted price Natural log sales Silicon Valley Subsidiary Unix operating system First mover RISC architecture Density Frequency of new product introduction Horizontal boundary Vertical integration Year dummies Capability stretching Capability Stretching × Vertical Integration Capability Stretching × Horizontal Boundary –2Log likelihood Likelihood ratio (χ2) df N

0.12 0.08 0.06 0.33 0.19 0.28 0.28 0.15 0.06 0.09

SE 0.12 0.08 0.06 0.34 0.20 0.30 0.28 0.15 0.07 0.09

SE

8,309.84 134.23** 26 724

0.13 0.16 0.00 0.19 Included 0.15** 0.03

–0.01 –0.13 –0.08 –0.12 0.38† 0.23 –0.19 0.23 0.05 –0.01

β

Model 2

0.13 0.08 0.06 0.33 0.21 0.28 0.28 0.15 0.06 0.09

SE

8,305.12 138.95** 27 724

0.11 0.16 0.02 0.18 Included 0.17** 0.04 0.10* 0.05

0.01 –0.15† –0.08 –0.08 0.38† 0.17 –0.17 0.24 0.03 0.00

β

Model 3

0.13 0.08 0.06 0.34 0.20 0.30 0.28 0.15 0.07 0.09

SE

8,309.69 134.38** 27 724

–0.01

0.03

0.13 0.16 0.00 0.19 Included 0.15** 0.03

–0.01 –0.13 –0.08 –0.13 0.38† 0.24 –0.20 0.22 0.05 –0.01

β

Model 4

0.13 0.08 0.06 0.34 0.20 0.28 0.27 0.15 0.06 0.09

SE

8,296.28 147.78** 28 724

–0.17**

0.03

0.12 0.13 0.02 0.16 Included 0.19** 0.04 0.27** 0.06

–0.01 –0.16† –0.06 –0.13 0.40* 0.19 –0.23 0.20 0.02 –0.01

β

Model 5

0.13 0.09 0.06 0.34 0.14 0.29 0.21 0.17 0.08 0.08

SE

6,310.76 177.27** 27 579

–0.24*

       

0.03

0.29* 0.12 –0.18† 0.10 Included 0.22** 0.05 0.37* 0.07

0.14 –0.10 –0.06 –0.23 0.73** 0.20 0.02 0.27† 0.17 0.28**

β

Model 6

Note: All the continuous variables are standardized. SAS PROC PHREG is used (Cox regression) with the COVSANDWICH option (robust variance estimator) to correct for dependence when there are repeated observations for each firm. Model 6 is for the robustness test in which all the products that exited the market in 1996 are eliminated from the database. RISC = reduced instruction set computer. †p < .10 (two-tailed test). *p < .05 (two-tailed test). **p < .01 (two-tailed test).

8,321.30 122.77** 25 724

0.17 0.16 –0.06 0.19 Included

β

Parameter

Model 1

Table 2 Proportional Hazard Model Predicting Product Exit

Wang, Chen / Capability Stretching in Product Innovation   15

Model 3, the interaction between vertical integration and capability stretching was added; in Model 4, the interaction between horizontal boundary and capability stretching was added. Model 5 is the full model including all the variables and interaction terms. The model difference test suggests that Model 5 achieves the best fit among all models. In Model 6, we performed a robustness test in which we eliminated from the sample all products that exited the market in year 1996. The results are qualitatively consistent with those from Model 5. Hypothesis 1 predicts that capability stretching reduces the chance of product survival; in other words, capability stretching increases the chance of product exit. The results in Model 2 show that the coefficient for capability stretching is positive and significant (β = 0.15, p < .01, hazard ratio = 1.16). Therefore, Hypothesis 1 is supported. A 1 SD increase in capability stretching will result in a 16% (i.e., 1.16 – 1 = 0.16) increase in the hazard of the product exit. We further explored how the survival functions vary with changes in the degree of capability stretching. In the SAS PHREG command we used, we included the BASELINE statement with an OUT option. We set all the continuous variables at mean level and the dummy variables at zero level. The graph is shown in Figure 2. The graph shows that the survival probability is lower for products with higher degrees of capability stretching. Hypothesis 2 predicts that vertical integration exacerbates the negative association between capability stretching and product survival; in other words, vertical integration positively moderates the association between capability stretching and product exit. The results in Model 3 show that the coefficient for the interaction between capability stretching and vertical integration on product exit is positive and significant (β = 0.10, p < .05), supporting Hypothesis 2. This result is also robust across models. For example, we also found that the result is significant in Model 5 (β = 0.27, p < .01). To further illustrate the interaction effect, we graphed the interaction in Figure 3. The vertical axis represents the hazard rate of product exit. We used 1 SD below and above the mean as the values for the moderator (i.e., vertical integration). We used unconditional exit probabilities at a fixed point in time—t = 2; h0(2) = 0.40—for the graphs because the baseline hazard changes over time. Figure 3 shows that, under a low level of vertical integration, capability stretching does not affect product exit. However, under a high level of vertical integration, capability stretching significantly increases the hazard rate of product exit. Hypothesis 3 predicts that horizontal boundary mitigates the negative association between capability stretching and product survival; in other words, horizontal boundary weakens the relationship between capability stretching and product exit. The results in Model 4 show that the coefficient for the interaction term is not significant; however, the results in Model 5 show a negative and significant coefficient for the interaction (β = –0.17, p < .01). Because Model 5 fits the model better than Model 4 (χ2 difference = 13.4, df difference = 1), we conclude that Hypothesis 3 is supported. We created Figure 4 to illustrate the relationship between capability stretching and the hazard rate of product exit under different scopes of the horizontal boundary (e.g., 1 SD below and above the mean). Since the baseline hazard varies over time, we used a fixed point in time—t = 2; h0(2) = 0.40—for the graphs. Figure 4 shows that, under a low level of horizontal boundary, capability stretching increases the hazard rate of product exit. However, under a high level of horizontal boundary, capability stretching reduces the hazard rate of product exit.

16   Journal of Management / Month XXXX

Figure 2 Survival Functions for Products With Varying Degrees of Capability Stretching 1.0 0.9

Survival Function

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0

1

2

3

4

5

6

7

8

Product Age Note: The line with star marks refers to products with a high level of capability stretching, the line with square marks refers to products with an average level of capability stretching, and the line with triangle marks refers to products with a low level of capability stretching. Consistent with the results, the graph suggests that the survival probability is higher for products with a low level of capability stretching. The graph was obtained by specifying the BASELINE statement in the SAS PHREG command.

Figure 3 The Moderating Effect of Vertical Integration

Post Hoc Analysis We carried out four post hoc analyses. In the first post hoc analysis, we used a new proxy for organizational boundary to reexamine the moderating effects of organizational boundaries on the relationship between capability stretching and product survival. In this analysis,

Wang, Chen / Capability Stretching in Product Innovation   17

Figure 4 The Moderating Effect of Horizontal Boundary

organizational boundary was measured not by product market scope but by knowledge base scope. We used the scope of a firm’s patent stock as a proxy for the scope of the firm’s knowledge base (Lerner, 1994). Since product market and knowledge base are two ways to look at a firm (Wernerfelt, 1984), we hope that this post hoc analysis can shed new light on the moderating hypotheses. Drawing on the National Bureau of Economic Research patent database, we measured the scope of a firm’s patent stock by the total number of patent classes into which a firm’s patents were classified during the period between 1980 and 1996 (Lerner). We created two separate indicators. One indicator measures related knowledge scope; that is, it measures the number of patent classes between current patent classification (CCL) 700 and 719 that are related to the workstation industry. The other indicator measures unrelated knowledge scope; that is, it measures the number of the remaining patent classes unrelated to the workstation industry. The results reported in Table 3 Model 7 show that related knowledge scope exacerbates the impact of capability stretching on product exit (β = 0.03, p < .01). Unrelated knowledge scope, however, mitigates the impact of capability stretching on product exit (β = –0.01, p < .01). The results are largely consistent with our arguments that vertical boundary exacerbates the cost of capability stretching, while horizontal boundary mitigates it.7 In the second post hoc analysis, we tried to disentangle two different scenarios for product exit. A product may exit the market when the next product has not yet been introduced to the market or when the next product has already been in the market.8 In this post hoc analysis, we split the original sample into two subsamples and ran the analyses separately. The results are reported in Models 8 and 9 of Table 3. The results suggest that capability stretching increases the chance of product exit, regardless of the scenarios.9 In the third post hoc analysis, we further examined a potential benefit of capability stretching—whether capability stretching enables firms to adapt to technological changes. We examined whether capability stretching reduces the waiting time for firms in the workstation industry to adopt the RISC architecture and introduce RISC workstations to the market. We created a measure, exit-to-adopt, to capture the “waiting time” from a CISC workstation to the first RISC workstation by the focal firm. Because RISC workstations first

18   Journal of Management / Month XXXX

Table 3 Post Hoc Analysis

Model 7: Product exit Parameter Last capability Adjusted price Natural log sales Silicon Valley Subsidiary Unix operating system First mover RISC architecture Density Frequency of new product introduction Horizontal boundary Related horizontal boundary (t – 1) Unrelated horizontal boundary (t – 1) Vertical integration Year dummies Capability stretching Capability Stretching × Related Horizontal Boundary Capability Stretching × Unrelated Horizontal Boundary Capability Stretching × Vertical Integration Capability Stretching × Horizontal Boundary –2Log likelihood Likelihood ratio (χ2) df (N)

β

SE

Model 8: Product exit before the next product was introduced β

SE

Model 9: Product exit after the next product was introduced β

SE

Model 10: Time to adopt a new product architecture β

0.11 –0.10 0.02 1.06** –0.34 0.09 –0.29 0.26 0.49** 0.01

0.26 0.09 0.34 0.34 0.22 0.40 0.49 0.19 0.11 0.24

–0.08†

0.05

0.17  

0.00

0.01

 

0.47 0.44 Included 0.09 0.09 0.03** 0.01

–0.01**

5,243.43 136.88** 29 (492)

0.07 –0.22* –0.02 0.30 0.59† 0.30 0.64† 0.13 –0.16 –0.02

0.17 0.11 0.10 0.30 0.31 0.45 0.33 0.27 0.11 0.12

0.04 –0.08 –0.18** –0.67 0.42* 0.19 –0.15 0.13 0.11 0.12

0.13 0.09 0.06 0.51 0.20 0.20 0.25 0.14 0.13 0.09

0.93** –0.25†+ 0.43† 1.16* 0.27 –0.22 1.06

0.39*

0.18

0.01

0.16

–0.56*

SE

–0.13 0.23 Included 0.25** 0.05

0.15 0.19 Included 0.20** 0.06

0.08 –0.60*

0.25 0.14 0.24 0.45 0.36 0.36 0.71   0.35 0.30

0.89† 0.46 Included 0.12* 0.06  

0.00

 

–0.05

0.21

0.30**

0.07

 

0.24

0.20

–0.19**

0.05

 

1,680.59 53.75** 28 (192)

5,795.68 123.45** 28 (532)

994.66 102.82** 21 (157)

     

Note: RISC = reduced instruction set computer; t = time. †p < .10 (two-tailed test). *p < .05 (two-tailed test). **p < .01 (two-tailed test).

appeared in the market in 1988, the sample for this post hoc analysis was restricted to all CISC workstations introduced to the market after 1988. The results from the Cox regression model are reported in Model 10 of Table 3. Results show that capability stretching has a positive effect on the speed to adopt the RISC architecture (β = 0.12, p < .05). Therefore, by extending a firm’s technological capabilities, capability stretching may help firms adapt to

Wang, Chen / Capability Stretching in Product Innovation   19

Table 4 Post Hoc Analysis Model 11 Parameter

β

Intercept Last capability First mover Silicon Valley Subsidiary Density Frequency of new product introduction Size of horizontal boundary Size of vertical integration Capability stretching Capability Stretching × Size of Vertical Integration Capability Stretching × Size of Horizontal Boundary –2 Res log likelihood N

16.95** –0.42† –3.57 –1.43** 0.70† 0.02† 2.65† –0.05 0.70* –1.01**

456.1 122

Model 12 SE

β

SE

4.49 0.21 4.69 0.35 0.36 0.01 1.46 0.05 0.28 0.26

16.95** –0.36† –3.53 –1.42** 0.62† 0.02† 2.23 –0.09 0.79** –0.92** –0.34† 0.10†

4.49 0.22 4.69 0.34 0.36 0.01 1.47 0.07 0.28 0.33 0.19 0.05

459.1 122

   

Note: Dependent variable = natural log (sales). †p < .10 (two-tailed test). *p < .05 (two-tailed test). **p < .01 (two-tailed test).

technological changes; such a benefit may be extremely important for firms in dynamic markets where they need to keep pace with rapid technological changes. In the fourth post hoc analysis, we extended the model from the product level to the firm level. We used revenue (in natural log) as the dependent variable. The independent variable, capability stretching, is calculated as the average degree of capability stretching across all products introduced by the same company in the same year. In this way, both the dependent and independent variables are aggregated from the product level to the firm level. Consistent with the product-level models, Model 11 of Table 4 shows that capability stretching has a negative impact on firm revenue (β = –1.01, p < .01). In addition, Model 12 of Table 4 shows that the interaction between capability stretching and vertical integration on firm revenue is negative (β = –0.34, p < .10) and that the interaction between capability stretching and horizontal boundary on firm revenue is positive (β = 0.10, p < .10). Therefore, the firm-level results are consistent with the product-level results. Therefore, we believe that the tension between extending technological capabilities and improving product market performance is real and strong and that this tension should be carefully managed.

Discussion Capability Stretching in Product Innovation Technological capabilities enable firms to develop new products effectively (Dosi, 1988; Nelson & Winter, 1982). Product development, in turn, also contributes to the development

20   Journal of Management / Month XXXX

of technological capabilities (Danneels, 2002; Helfat & Raubitschek, 2000). Capability stretching is therefore rewarding because it enables firms to acquire new knowledge, assimilate new technologies, and therefore become better prepared for future product innovation (Afuah & Tucci, 2012; Day, 2011; Schilling, 2012). However, capability stretching can also be challenging, as capability stretching may start a new learning curve (Liberman, Benton, Dershaw, Abramson, LaTrenta, & Morris, 2001) and may reduce the chances of product survival. The first products using newly acquired knowledge and technologies may have glitches and defects and may thus not perform very well in the market (Fleming, 2001; Katila & Ahuja, 2002). Firms therefore face a dilemma in managing capability stretching. If they stretch their capabilities in product development, the immediate performance of the new product may suffer; if they do not stretch their capabilities, however, they may gradually lag behind industry frontiers. In the short run, capability stretching reduces new product performance. In the long run, however, capability stretching in prior product innovations may help firms accumulate technological knowledge and reduce the effective distance of knowledge, enabling them to carry out effective technological searches and new combinations in the future (LeonardBarton, 1992; March, Sproull, & Tamuz, 1991; Nelson & Winter, 1982; Wernerfelt, 1984). In this study, we drew on prior studies to develop an objective measure for capability stretching and examine its impact on new product survival (Afuah & Tucci, 2012; Agarwal et al., 2004; Rockart & Dutt, 2015). We found that capability stretching indeed negatively affects the survival chances of new products using new, distant technological knowledge; however, we also found that capability stretching facilitates adaptation to major technological changes (e.g., the transition from CISC architecture to RISC architecture in the workstation industry). Our study seems to suggest that product innovation may be considered as a mechanism for not only improving product market performance but also stretching technological capabilities. These two faces of product innovation may sometimes conflict with each other, and this conflict may fundamentally reflect the tension between exploitation and exploration (Denrell & March, 2001; Levinthal & March, 1993; March, 1991, 2006, 2010). Facing this tension, firms may sometimes need to sacrifice short-term rewards in order to look for new possibilities (Holland, 1992; Kauffman, 1993; Schumpeter, 1934; Spall, 2003; Sutton & Barto, 1998). However, firms may find it difficult to justify such short-term sacrifices because the returns to capability stretching may seem to be distant and uncertain (Levinthal & March; March, 1999, 2008, 2010). Therefore, it is imperative to understand how firms can effectively cope with this tension in product innovation. In this study, we found that organizational boundaries affect the relationship between capability stretching and new product survival. Vertically integrated firms may face severe effective distance of knowledge in capability stretching. At a high level of vertical integration, firms may face substantial difficulties in capability stretching, further exacerbating the negative association between capability stretching and new product survival. Our results therefore help uncover the complicated interaction between vertical boundary and capability stretching on product survival, providing an important foundation for future research. We also found that a large horizontal boundary may help firms mitigate the negative impact of capability stretching on product survival. Firms with a large horizontal boundary enjoy some economies of scope because knowledge accumulated in other product categories can be internally transferred to the focal product category (Argote & Ingram, 2000; Henderson

Wang, Chen / Capability Stretching in Product Innovation   21

& Cockburn, 1996; Miller, Fern, & Cardinal, 2007). Internal technology spillovers and transfers may reduce the effective distance of knowledge facing innovating firms and help them cope with distant search and capability stretching (Afuah & Tucci, 2012). In addition, firms with a large horizontal boundary may enjoy the benefits of a broadened scope for external search (Laursen & Salter, 2006; Leiponen & Helfat, 2010). Such broad access to external knowledge may help firms acquire and assimilate external knowledge that is helpful in bridging capability gaps. Our results therefore help uncover a mechanism through which horizontal boundary facilitates capability stretching, suggesting new directions for future research.

Capability Stretching as a Dimension of Dynamic Capabilities According to Helfat et al., “A dynamic capability is the capacity of an organization to purposefully create, extend, or modify its resource base” (2007: 4). In this study, we conceptualized capability stretching as a dimension of dynamic capabilities, the dimension that is most closely related to the extension of capabilities. We drew on prior studies to develop an objective measure for capability stretching (Afuah & Tucci, 2012; Agarwal et al., 2004; Rockart & Dutt, 2015) and examine its impact on product survival. We believe that the conceptualization and operationalization of capability stretching can facilitate subsequent research in dynamic capabilities and advance our understanding of capability development and evolution. Prior studies have categorized organizational capabilities into operational capabilities and dynamic capabilities (Helfat et al., 2007; Helfat & Winter, 2011; Teece et al., 1997; Zollo & Winter, 2002), organizational learning into exploration and exploitation (Levinthal & March, 1993; March, 1991, 2010), and organizational search into local search and distant search (Afuah & Tucci, 2012; Cyert & March, 1992; Nelson & Winter, 1982). However, capabilities, learning, and search may vary not only by type but also by degree. Recent developments, for example, have led researchers to recognize the need to further conceptualize and measure the “effective distance of knowledge” in organizational search (Afuah & Tucci) and the “degree of exploration” in organizational learning (Walter, Lechner, & Kellermanns, in press). To the best of our knowledge, however, no prior studies have explicitly examined the degree of capability stretching in product development. We believe that our measure for capability stretching is a small step in this important direction. This measure is doubtless crude and imperfect, and we hope that more researchers can join our efforts in perfecting this concept and its related measures. Furthermore, we believe that future research along this direction can help mitigate the tautological tendency of defining dynamic capabilities by their impacts on organizational performance. We believe that the degrees and types of capability change can be, and should be, explicitly measured independent of firm performance (Barreto, 2010; King & Tucci, 2002). We believe that future research along this direction will not only strengthen the theoretical foundations of dynamic capabilities but also enrich the practical implications of this important literature.

The Role of Organizational Boundaries in Capability Stretching In this study, we also enrich the conversation between the organizational economics literature and the organizational capabilities literature (Argyres & Zenger, 2012; Williamson,

22   Journal of Management / Month XXXX

1999). We noticed that organizational boundaries moderate the relationship between capability stretching and new product survival. Vertical boundary has been found to affect the relationship between capability stretching and new product survival differently from horizontal boundary. We found that highly vertically integrated firms face considerable difficulties in extending their capabilities in product innovation, therefore exacerbating the negative impact of capability stretching on product survival. Conversely, firms with a large horizontal boundary may benefit from internal knowledge spillovers and transfers and a broadened scope for external search, mitigating the negative impact of capability stretching on product survival. Organizational boundaries therefore affect the challenges that firms may face in capability stretching. Accordingly, the way that a firm structures its organizational boundaries may affect the difficulties it may encounter in capability stretching. When a firm chooses its organizational boundaries, it may also implicitly choose the evolutionary trajectories for its organizational capabilities. Therefore, firms may need to pay more attention to the structuring of organizational boundaries because organizational boundaries may affect the dynamics of organizational capabilities (Argyres, 1996, 2011; Argyres et al., 2012; Argyres & Zenger, 2012).

Limitations One limitation of this research is that it is conducted within a single industry. Therefore, it remains unclear whether the conclusions of this study can be generalized to other industries. We hope that other researchers will join us in generalizing the results to other industries. Another limitation is that, as a result of data availability, we do not have information about the technical performance of the products. Subsequent studies can combine this data set with other data sets to study the impact of capability stretching on the technical performance of new products. A third limitation is that this study focuses on capability stretching along existing performance trajectories; subsequent studies can extend our model to examine how firms deal with capability stretching when firms move to new performance trajectories (Christensen, 1997; Christensen & Raynor, 2003). Finally, we were not able to identify the exact reasons for product exits in this study because of data limitation. A product can exit the market as a result of either product improvement or product discontinuation. However, the reason for each product exit cannot be exactly identified in this study. We have carried out some post hoc analyses to address this limitation; however, we believe that much more can be done. One promising future research direction, for example, is to identify the exact reason for each product exit, classify product exits into different types according to the reasons for product exits, and further examine the role of capability stretching in each type of product exit.

Conclusion Product innovation is conventionally treated as a mechanism for renewing product portfolios. In this paper, we have suggested that product innovation not only enables organizations to introduce new products to the market but also challenges organizations to renew their technological capabilities. Once we recognize these two faces of product innovation, we may start to realize that these two roles may sometimes conflict with each other. In the long run, capability stretching enables organizations to acquire new knowledge, assimilate new

Wang, Chen / Capability Stretching in Product Innovation   23

technologies, and become better prepared for product innovation. In the short run, however, capability stretching can be unattractive because it may result in a temporary decline in product market performance. The short-run unattractiveness of capability stretching, however, may vary across firms depending on the structuring of organizational boundaries. Organizational boundaries reflect the extent to which product knowledge is controlled within a firm and therefore affect the capability gaps that need to be overcome in capability stretching. Consequently, the structuring of organizational boundaries may affect the difficulty of capability stretching and therefore the subsequent evolution of organizational capabilities. We hope that this study emphasizes the usefulness of exploring the role of organizational boundaries in capability stretching and organizational renewal and the importance of studying capability stretching as a dimension of dynamic capabilities. We believe that studies along these lines will shed new light on the linkages among product innovation, organizational boundaries, and organizational capabilities.

Notes 1. We thank the anonymous reviewers for identifying the links among the effective distance of knowledge, absorptive capacity, and capability stretching. We also would like to thank them for their suggestions on interpreting these relationships. 2. We appreciate the comments from the anonymous reviewers. 3. Admittedly, vertically integrated firms may enjoy the advantage of knowledge spillovers among different vertically controlled stages when the knowledge bases of different stages are very similar. However, when the knowledge bases of different stages are quite different, vertical integration may not be helpful in transferring knowledge among these different stages. Therefore, we argue that in highly vertically integrated firms, the benefits of knowledge spillovers among different stages may be outweighed by the difficulties of capability stretching in multiple internally controlled stages. 4. The Windows NT operating system was first released in 1993, but it was not widely adopted until Version 4.0 became available in 1996. Because of Windows NT, workstation computers shifted from being qualitatively different from PCs to being quantitatively different from PCs. 5. An instruction set is the repertoire of operations that a microprocessor can carry out. Two general instruction set architectures are CISC and RISC. Within the CISC architecture, increases in computing power are associated with increases in the size of the instruction set and in the power of individual instruction. As more and more instructions are added to the instruction sets, many complex instructions become useless. RISC is an architectural innovation for microprocessors (Afuah, 2001: 1217). The adoption of the RISC architecture marked a key architectural change in computer workstations. The RISC architecture follows the philosophy that only instructions that can improve performance will be added to the instruction sets. As an architectural innovation, the RISC architecture affects not only the CPU but also its linkages to other components (Afuah; Henderson & Clark, 1990). It changes the way microprocessors and workstations are designed (Afuah). While CISC architectures build complex instructions into hardware, RISC computers place a high demand on software, putting a higher demand on memory and software (Afuah). The transition from the CISC architecture to RISC architecture was also not an easy task because a lot of changes needed to be made. For example, Afuah notes that “designing a RISC workstation required a lot more than replacing a CISC microprocessor with a RISC one. RISC processors, with their simpler instructions, required the memory and software to do a lot more than their predecessors had done” (1218). 6. We appreciate the comments from the anonymous reviewers. 7. We would like to thank the anonymous reviewers for pointing out the direction for this analysis. In this analysis, it is important to note that related knowledge scope may be related to both vertical integration and related diversification; knowledge in patent classes related to the workstation industry can be used to produce different components of a workstation as well as other products in related industries. Unrelated knowledge scope, however, may be related to unrelated diversification, and unrelated diversification seems to be better able to mitigate the negative impact of capability stretching on product survival. The results of this post hoc analysis are interesting but inconclusive; they may require further studies to clarify their implications.

24   Journal of Management / Month XXXX 8. We thank an anonymous reviewer for suggesting this post hoc analysis and for identifying important insights implied by this analysis. 9. In this post hoc analysis, we also examined the moderating effects of organizational boundaries on the relationships between capability stretching and product exit. We noticed that the moderating effects manifest only in the scenario under which a product exits the market when the next product has already been in the market. This seems to suggest that organizational boundaries play a more important role in capability stretching under the condition when a newer product has been introduced to the market before the focal product is withdrawn from the market.

References Afuah, A. 2000. How much do your co-opetitors’ capabilities matter in the face of technological change? Strategic Management Journal, 21: 397-404. Afuah, A. 2001. Dynamic boundaries of the firm: Are firms better off being vertically integrated in the face of a technological change? Academy of Management Journal, 44: 1211-1228. Afuah, A., & Tucci, C. L. 2012. Crowdsourcing as a solution to distant search. Academy of Management Review, 37: 355-375. Agarwal, R., Echambadi, R., Franco, A. M., & Sarkar, M. 2004. Knowledge transfer through inheritance: Spin-out generation, development, and survival. Academy of Management Journal, 47: 501-522. Allison, P. D. 2005. Fixed effects regression methods for longitudinal data using SAS. Cary, NC: SAS. Allison, P. D. 2010. Survival analysis using SAS: A practical guide. Cary, NC: SAS. Anderson, P., & Tushman, M. L. 1990. Technological discontinuities and dominant designs: A cyclical model of technological change. Administrative Science Quarterly, 35: 604-633. Argote, L. 2012. Organizational learning: Creating, retaining and transferring knowledge (2nd ed.). New York: Springer. Argote, L., & Ingram, P. 2000. Knowledge transfer: A basis for competitive advantage in firms. Organizational Behavior and Human Decision Processes, 82: 150-169. Argyres, N. 1996. Evidence on the role of firm capabilities in vertical integration decisions. Strategic Management Journal, 17: 129-150. Argyres, N. 2011. Using organizational economics to study organizational capability development and strategy. Organization Science, 22: 1138-1143. Argyres, N. S., Felin, T., Foss, N., & Zenger, T. 2012. Organizational economics of capability and heterogeneity. Organization Science, 23: 1213-1226. Argyres, N. S., & Zenger, T. R. 2012. Capabilities, transaction costs, and firm boundaries. Organization Science, 23: 1643-1657. Barreto, I. 2010. Dynamic capabilities: A review of past research and an agenda for the future. Journal of Management, 36: 256-280. Brown, S. L., & Eisenhardt, K. M. 1995. Product development: Past research, present findings, and future directions. Academy of Management Review, 20: 343-378. Brown, S. L., & Eisenhardt, K. M. 1997. The art of continuous change: Linking complexity theory and time-paced evolution in relentlessly shifting organizations. Administrative Science Quarterly, 42: 1-34. Brusoni, S., Prencipe, A., & Pavitt, K. 2001. Knowledge specialization, organizational coupling, and the boundaries of the firm: Why do firms know more than they make? Administrative Science Quarterly, 46: 597-621. Capron, L., & Mitchell, W. 2009. Selection capability: How capability gaps and internal social frictions affect internal and external strategic renewal. Organization Science, 20: 294-312. Carlile, P. R. 2004. Transferring, translating, and transforming: An integrative framework for managing knowledge across boundaries. Organization Science, 15: 555-568. Carroll, G. R., Khessina, O. M., & McKendrick, D. G. 2010. The social lives of products: Analyzing product demography for management theory and practice. The Academy of Management Annals, 4: 157-203. Chisholm, D. C., & Norman, G. 2006. When to exit a product: Evidence from the US motion-picture exhibition market. American Economic Review, 96: 57-61. Christensen, C. M. 1997. The innovator’s dilemma: When new technologies cause great firms to fail. Boston: Harvard Business School Press. Christensen, C. M., & Raynor, M. E. 2003. The innovator’s solution: Creating and sustaining successful growth. Boston: Harvard Business School Press.

Wang, Chen / Capability Stretching in Product Innovation   25 Coen, C. A., & Maritan, C. A. 2011. Investing in capabilities: The dynamics of resource allocation. Organization Science, 22: 99-117. Cohen, W. M., & Levinthal, D. A. 1989. Innovation and learning: The two faces of R&D. Economic Journal, 99: 569-596. Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35: 128-152. Cohen, W. M., & Levinthal, D. A. 1994. Fortune favors the prepared firm. Management Science, 40: 227-251. Cox, D. R. 1972. Regression models and life-tables. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 34: 187-220. Cyert, R. M., & March, J. G. 1992. A behavioral theory of the firm (2nd ed.). Oxford, England: Blackwell. Danneels, E. 2002. The dynamics of product innovation and firm competences. Strategic Management Journal, 23: 1095-1121. Day, G. S. 2011. Closing the marketing capabilities gap. Journal of Marketing, 75(4): 183-195. De Figueiredo, J. M., & Kyle, M. K. 2006. Surviving the gales of creative destruction: The determinants of product turnover. Strategic Management Journal, 27: 241-264. Denrell, J., & March, J. G. 2001. Adaptation as information restriction: The hot stove effect. Organization Science, 12: 523-538. Dierickx, I., & Cool, K. 1989. Asset stock accumulation and sustainability of competitive advantage. Management Science, 35: 1504-1511. Dosi, G. 1988. Sources, procedures, and microeconomic effects of innovation. Journal of Economic Literature, 26: 1120-1171. Dosi, G., & Nelson, R. R. 2010. Technical change and industrial dynamics as evolutionary processes. In B. H. Hall & N. Rosenberg (Eds.), Handbook of the economics of innovation, vol. 1: 51-127. Amsterdam: North-Holland. Eisenhardt, K. M., & Martin, J. A. 2000. Dynamic capabilities: What are they? Strategic Management Journal, 21: 1105-1121. Eisenhardt, K. M., & Tabrizi, B. N. 1995. Accelerating adaptive processes: Product innovation in the global computer industry. Administrative Science Quarterly, 40: 84-110. Fleming, L. 2001. Recombinant uncertainty in technological search. Management Science, 47: 117-132. Gatignon, H., Tushman, M. L., Smith, W., & Anderson, P. 2002. A structural approach to assessing innovation: Construct development of innovation locus, type, and characteristics. Management Science, 48: 1103-1122. Helfat, C. E., Finkelstein, S., Mitchell, W., Peteraf, M., Singh, H., Teece, D., & Winter, S. G. 2007. Dynamic capabilities: Understanding strategic change in organizations. Malden, MA: Blackwell. Helfat, C. E., & Raubitschek, R. S. 2000. Product sequencing: Co-evolution of knowledge, capabilities and products. Strategic Management Journal, 21: 961-979. Helfat, C. E., & Winter, S. G. 2011. Untangling dynamic and operational capabilities: Strategy for the (n)everchanging world. Strategic Management Journal, 32: 1243-1250. 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: 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. Holland, J. 1992. Adaptation in natural and artificial systems (2nd ed.). Cambridge, MA: MIT Press. Jacquemin, A. P., & Berry, C. H. 1979. Entropy measure of diversification and corporate growth. Journal of Industrial Economics, 27: 359-369. Kapoor, R., & Adner, R. 2012. What firms make vs. what they know: How firms’ production and knowledge boundaries affect competitive advantage in the face of technological change. Organization Science, 23: 1227-1248. Katila, R. 2002. New product search over time: Past ideas in their prime? Academy of Management Journal, 45: 995-1010. Katila, R., & Ahuja, G. 2002. Something old, something new: A longitudinal study of search behavior and new product introduction. Academy of Management Journal, 45: 1183-1194. Kauffman, S. A. 1993. The origins of order: Self-organization and selection in evolution. New York: Oxford University Press. Khessina, O. M., & Carroll, G. R. 2008. Product demography of de novo and de alio firms in the optical disk drive industry, 1983-1999. Organization Science, 19: 25-38. Kidder, T. 1981. The soul of a new machine. Boston: Little, Brown.

26   Journal of Management / Month XXXX King, A. A., & Tucci, C. L. 2002. Incumbent entry into new market niches: The role of experience and managerial choice in the creation of dynamic capabilities. Management Science, 48: 171-186. Knudsen, T., & Levinthal, D. A. 2007. Two faces of search: Alternative generation and alternative evaluation. Organization Science, 18: 39-54. Lafontaine, F., & Slade, M. 2007. Vertical integration and firm boundaries: The evidence. Journal of Economic Literature, 45: 629-685. Laursen, K., & Salter, A. 2006. Open for innovation: The role of openness in explaining innovation performance among U.K. manufacturing firms. Strategic Management Journal, 27: 131-150. Leiponen, A., & Helfat, C. E. 2010. Innovation objectives, knowledge sources, and the benefits of breadth. Strategic Management Journal, 31: 224-236. Leonard-Barton, D. 1992. Core capabilities and core rigidities: A paradox in managing new product development. Strategic Management Journal, 13: 111-125. Lerner, J. 1994. The importance of patent scope: An empirical analysis. RAND Journal of Economics, 25: 319-333. Levinthal, D. A., & March, J. G. 1993. The myopia of learning. Strategic Management Journal, 14: 95-112. Levitt, B., & March, J. G. 1988. Organizational learning. Annual Review of Sociology, 14: 319-340. Liberman, L., Benton, C. L., Dershaw, D. D., Abramson, A. F., LaTrenta, L. R., & Morris, E. A. 2001. Learning curve for stereotactic breast biopsy. American Journal of Roentgenology, 176: 721-727. March, J. G. 1991. Exploration and exploitation in organizational learning. Organization Science, 2: 71-87. March, J. G. 1999. The pursuit of organizational intelligence: Decisions and learning in organizations. Malden, MA: Blackwell. March, J. G. 2006. Rationality, foolishness, and adaptive intelligence. Strategic Management Journal, 27: 201-214. March, J. G. 2008. Explorations in organizations. Stanford, CA: Stanford University Press. March, J. G. 2010. The ambiguities of experience. Ithaca, NY: Cornell University Press. March, J. G., & Simon, H. A. 1993. Organizations (2nd ed.). Cambridge, MA: Blackwell. March, J. G., Sproull, L. S., & Tamuz, M. 1991. Learning from samples of one or fewer. Organization Science, 2: 1-13. Maritan, C. A., & Brush, T. H. 2003. Heterogeneity and transferring practices: Implementing flow manufacturing in multiple plants. Strategic Management Journal, 24: 945-959. McDermott, C. M., & Stock, G. N. 2011. Focus as emphasis: Conceptual and performance implications for hospitals. Journal of Operations Management, 29: 616-626. Miller, D. J., Fern, M. J., & Cardinal, L. B. 2007. The use of knowledge for technological innovation within diversified firms. Academy of Management Journal, 50: 307-325. Nelson, R. R., & Winter, S. G. 1982. An evolutionary theory of economic change. Cambridge, MA: Belknap Press. Nerkar, A., & Roberts, P. W. 2004. Technological and product-market experience and the success of new product introductions in the pharmaceutical industry. Strategic Management Journal, 25: 779-799. Ozcan, P., & Eisenhardt, K. M. 2009. Origin of alliance portfolios: Entrepreneurs, network strategies, and firm performance. Academy of Management Journal, 52: 246-279. Rockart, S. F., & Dutt, N. 2015. The rate and potential of capability development trajectories. Strategic Management Journal, 36: 53-75. Santos, F. M., & Eisenhardt, K. M. 2005. Organizational boundaries and theories of organization. Organization Science, 16: 491-508. Schilke, O. 2014. On the contingent value of dynamic capabilities for competitive advantage: The nonlinear moderating effect of environmental dynamism. Strategic Management Journal, 35: 179-203. Schilling, M. A. 2012. Strategic management of technological innovation (4th ed.). New York: McGraw-Hill. Schumpeter, J. A. 1934. The theory of economic development: An inquiry into profits, capital, credit, interest, and the business cycle. Cambridge, MA: Harvard University Press. Simon, H. A. 1955. A behavioral model of rational choice. Quarterly Journal of Economics, 69: 99-118. Simon, H. A. 1997. Administrative behavior: A study of decision-making processes in administrative organization (4th ed.). New York: Free Press. Sorenson, O. 2000. Letting the market work for you: An evolutionary perspective on product strategy. Strategic Management Journal, 21: 577-592. Sorenson, O. 2003. Interdependence and adaptability: Organizational learning and the long-term effect of integration. Management Science, 49: 446-463.

Wang, Chen / Capability Stretching in Product Innovation   27 Sorenson, O., McEvily, S., Ren, C. R., & Roy, R. 2006. Niche width revisited: Organizational scope, behavior and performance. Strategic Management Journal, 27: 915-936. Sorescu, A. B., Chandy, R. K., & Prabhu, J. C. 2003. Sources and financial consequences of radical innovation: Insights from pharmaceuticals. Journal of Marketing, 67(4): 82-102. Spall, J. C. 2003. Introduction to stochastic search and optimization: Estimation, simulation, and control. Hoboken, NJ: John Wiley & Sons. Sutton, R. S., & Barto, A. G. 1998. Reinforcement learning: An introduction. Cambridge, MA: MIT Press. Teece, D. J. 2007. Explicating dynamic capabilities: The nature and microfoundations of (sustainable) enterprise performance. Strategic Management Journal, 28: 1319-1350. Teece, D. J., Pisano, G., & Shuen, A. 1997. Dynamic capabilities and strategic management. Strategic Management Journal, 18: 509-533. Tushman, M. L., & Anderson, P. 1986. Technological discontinuities and organizational environments. Administrative Science Quarterly, 31: 439-465. Walter, J., Lechner, C., & Kellermanns, F. W. in press. Learning activities, exploration, and the performance of strategic initiatives. Journal of Management. doi:10.1177/0149206313506463 Wernerfelt, B. 1984. A resource-based view of the firm. Strategic Management Journal, 5: 171-180. Williamson, O. E. 1985. The economic institutions of capitalism. New York: Free Press. Williamson, O. E. 1999. Strategy research: Governance and competence resources. Strategic Management Journal, 20: 1087-1108. Wolter, C., & Veloso, F. M. 2008. The effects of innovation on vertical structure: Perspectives on transaction costs and competences. Academy of Management Review, 33: 586-605. Zollo, M., & Winter, S. G. 2002. Deliberate learning and the evolution of dynamic capabilities. Organization Science, 13: 339-351.