|
Article Excerpt
If there are profitable opportunities for increased production
anywhere in the economy they will provide for some firm an external
inducement to expand. But this alone tells us nothing about their
significance for any given firm. [Opportunities] are external
inducements to expand only for what might be termed "qualified"
firms--firms whose internal resources are of a kind either to give
them a special advantage in the "profitable" areas or a least not to
impose serious obstacles. (Penrose 1959, p. 86)
1. Introduction
According to resource-based theory, a firm's valuable and unique resources are at the root of its competitive advantage (Penrose 1959, Wernerfelt 1984, Barney 1991, Peteraf 1993, Conner and Prahalad 1996). However, identifying which of a firm's resources matter most for competitive advantage is no easy task. Although resource ambiguity may tend to protect competitive advantage, it presents difficult challenges for researchers in testing predictive theory. Consider the problem of predicting the direction of growth of the firm. Resource-based theory suggests that excess capacity in idiosyncratic resources, combined with externally determined opportunities, leads to expansion in directions related to a firm's existing resource stock (Penrose 1959). The predictive challenge arises in at least three ways: (1) identifying which resources are leveraged for growth; (2) determining how those valuable resources relate to competitive strength in potential target industries; and (3) developing some notion of how the firm chooses among the options.
Identifying the resources that are leveraged for growth is a challenge because the resources upon which competitive advantage rests are often bundled, tacit, intangible, or unobservable (Nelson and Winter 1982; Rumelt 1984; Peteraf 1993; Winter 1987, 1995). Second, determining how valuable resources relate to those effective in another industry is a challenge because the type of relatedness that should matter may be unclear. Relatedness is a multidimensional construct (Pehrsson 2006, Stimpert and Duhaime 1997), and different dimensions are likely to apply to different resources and in different contexts. Moreover, knowing which of the resources are the ones for which relatedness should matter in the target market is also problematic. Finally, even if the first two challenges can be resolved, the question of how to choose among viable target markets remains, but generally a firm will expand into those areas in which its resources deliver the greatest advantage (Penrose 1959).
In this paper, we respond to these challenges by developing a general interindustry relatedness index that can be applied across firm and industry contexts but that does not require explicit identification of resource type. Specifically, our approach employs the insight embodied in the survivor principle (Stigler 1968) by presuming that because existing firms are repositories for resources, skills, and knowledge, the activity patterns of going firms are good indicators of how resources and knowledge relate across diverse activities. To sidestep the problem posed by the difficulty of observing the actual resources that are leveraged for growth at the level of the firm, we posit that there is a characteristic basket of these resources for each industry. The question we answer is not what resources rest within any one industry basket in particular, but rather how the resources in a particular basket relate to the resources in other baskets. Knowing which idiosyncratic resources reside in a particular industry basket is not required for predictive success, because once it is known how that basket relates to every other industry basket, one knows in which candidate directions the leveraging of those unobservable idiosyncratic resources is likely to lead. The advantage of this approach is that it acknowledges that the characteristic resource baskets differ from industry to industry without requiring a specification of those differences. As we demonstrate in an illustrative application here, this permits empirical testing of hypotheses about relatedness without requiring the researcher to make a specific prior commitment as to the types of resources that are critical.
To identify the system of relationships among industries, the index harnesses the information embedded in the joint industry participation choices of every diversified firm in the U.S. manufacturing economy for the specific time period upon which the index is based. For this we specify the finest level of detail at which industrial "participation" can be effectively assessed, which we take to be the four-digit level of the Standard Industrial Classification (SIC) system. Our calculations yield a measured "distance" between the two industries in every pair of four-digit industries in the U.S. manufacturing sector, where low distance corresponds to high relatedness. Our methods could be applied to any system that provides an exhaustive classification of activity at whatever is considered to be the micro level, and to any time period for which the requisite data are available.
The index is applicable to a wide range of problems in strategic management, corporate finance, and economics because it provides a plausible measure of the relative strength of association between every pair of manufacturing industries. The index may be particularly applicable to empirical examinations of strategic theory in areas such as the resource-based view (Peteraf 1993, Barney 1991, Wernerfelt 1984), organizational economics (e.g., Teece 1980, 1982), and knowledge and capabilities (e.g., Winter 2003, 1987; Helfat 2000; Dosi et al. 2000; Teece et al. 1997; Grant 1996; Kogut and Zander 1992; Helfat and Eisenhardt 2004), because these perspectives typically require assessment of the degree of overlap, knowledge, or relatedness between one firm activity and another. Similarly, concepts of relatedness are fundamental to discussions of how firms search for new market-entry opportunities that economize on existing resources as they build new capabilities (Bryce 2003; Coff 1999; Silverman 1999; Teece 1980, 1982); how capabilities develop from sequences of decisions, that are made in the context of resources in hand (Helfat and Raubitschek 2000, Helfat and Lieberman 2002, Helfat and Eisenhardt 2004); or how the ability to share firm-specific resources across activities results in higher levels of firm performance (Teece 1982, Peteraf 1993, Mahoney and Pandian 1992, Teece et al. 1997). Applications of the measure to the study of longitudinal patterns of diversification and firm growth are especially promising because the measure allows sequential analysis of the introduction of new industries into a firm's portfolio, one activity at a time.
This paper proceeds as follows. In [section]2, we provide a brief review of how concepts of relatedness and diversification have been used in the literature, supply the theoretical rationale for our particular approach, and propose solutions to some methodological problems that arise in measuring relatedness using a survivor-based approach. In [section]3, we develop the index, and in [section]4, we offer a test of predictive validity. Section 5 concludes with a discussion of potential applications.
2. Background and Theory
2.1. Measures of Relatedness
Measures of relatedness are designed to assess the degree of commonality (of some sort) within pairs of activities. They differ in their logic and uses from diversification measures, which are typically designed to support evaluation of a diversification strategy at the firm or portfolio level. Whereas diversification measures capture the state of a corporate portfolio at a point in time, relatedness measures can be used to characterize the flow or transition from state to state. Diversification measures do rest on underlying relatedness assessments, however. In standard diversification measures, relatedness is typically computed based on hierarchical distance within the SIC structure--a course that implicitly relies on the designers of the SIC system to have already answered the basic question. (1)
Relatedness components in standard diversification measures cannot effectively serve as stand-alone relatedness indicators because the hierarchical structure of the SIC system does not represent an underlying relatedness scale. Much of the SIC system reflects, for historical reasons, a broad logic of vertical structure and primary raw material. Thus, for example, functionally substitutable products made of steel, aluminum, and plastic appear in different two-digit industries because of the underlying difference in primary feed stock. This virtually guarantees that the knowledge about how to produce a functionally similar product lies scattered around the SIC system. For some two-digit SIC categories, and at finer classification levels, end use plays a more significant conceptual role (electrical equipment or apparel, for example). Ultimately, the fact that two four-digit industries share the same three-digit code (and on up the line) supplies no clear message about strategically significant relationships among activities. Relatedness simply cannot be reliably or directly inferred from the hierarchical structure of the SIC system (cf. Davis and Duhaime 1992, Robins and Wiersema 1995). (2)
Perhaps most importantly, the SIC hierarchy does not consistently reflect relationships among valuable resources in the ways that firms actually combine them to create value. Our approach of inferring relatedness from the aggregate activity combinations of firms provides a strong resource-based measure of relatedness between industries because it reflects the unobservable ways that firms share resources among industry activities. A measure rooted in actual resource combinations inside firms has the advantage of reliably merging both demand-side and supply-side considerations in the way firms deploy resources. Peteraf and Bergen (2003), for example, note that resource substitution effects by rivals are an important potential source of erosion of a firm's resource-based competitive advantage. Rivals who aim to compete on the basis of similarity of resource use, not just type, may go unnoticed by firms that regard as competitors only those firms that have similar resources and produce similar products (demand side). The methodology used here is advantageous in this regard because it also captures use relationships (supply side) while avoiding dependence on the relatively arbitrary industrial taxonomy of the SIC.
There are only a few relatedness measures with a non-SIC-hierarchy-based foundation (e.g., Robins and Wiersema 1995, Silverman 1999, Farjoun 1994, Coff 1999). These are activity-to-activity constructs that have been developed by researchers to test propositions suggested by the resource-based view of the firm. The same is true of the measure developed here. However, this paper focuses on the development of the relatedness measure itself. Unlike the extant relatedness measures, it is developed as a general index with a broad range of potential applications rather than to accomplish a specific empirical task. It also has the advantage that it does not presume prior identification of a key resource class such as human resources (Chang 1992, Farjoun 1994, Coff 1999), patents (Silverman 1999), or technology flows (Robins and Wiersema 1995)--the relevance of which may vary with activity--prior to computing relatedness scores. By moving away from a focus on single resource categories, our measure seeks to capture the aggregate patterns of shared know-how or capabilities (Teece 1982) that are at the root of economies of scope and resource combination decisions.
2.2. Theoretical Rationale
Ultimately, the test of the validity of our index will be its predictive power in a variety of significant empirical settings, going well beyond the illustrative test we present later in this paper. At this stage, we can only support the index by setting forth the assumptions and arguments that motivated our approach to its construction. Such an account may provide some theoretical guidance regarding the appropriate use of the index, and it may also suggest how it might be improved or how the empirical validity of its underlying logic might be checked more directly.
We adopt, first, the premise that the resource-based view (Peteraf 1993, Barney 1991, Wernerfelt 1984) is substantially correct in its assessment of the forces affecting the directions of firm growth. We assume that patterns of corporate diversification and expansion are shaped in a fundamental and sustained way by logic of economic efficiency (Teece 1980). Opportunities for profitable diversification moves arise because there is some overlap between the resources and capabilities that support the existing portfolio of activities and those that are required in some new line of activity (Teece 1982). Such overlaps produce "economies of scope"--a term that we use in a broad sense to cover any and all sources of economic gains arising from the combination of disparate activities (e.g., Teece 1980, Panzar and Willig 1981, Lemelin 1982). Scope economies can arise in a short-run context because indivisibilities or other considerations have led the firm to...
|
