Population Level Learning and Industry Change

Edited by Anne S. Miner and Philip C. Anderson
Advances in Strategic Management
, Volume 16

INDUSTRY AND POPULATION LEVEL LEARNING:
ORGANIZATIONAL, INTERORGANIZATIONAL AND COLLECTIVE LEARNING PROCESSES

Anne S. Miner and Philip C. Anderson

OVERVIEW AND IMPLICATIONS

One important school of strategic management theory has long adopted an implicit learning perspective in describing how strategies are formed (e.g., Burgelman, 1988; Mintzberg & McHugh, 1985; Noda & Bower, 1996; Quinn, 1980). These authors have focused on the ways organizational strategies emerge as individuals and groups come to learn about their own situation and to converge on patterns of behavior that appear to work. Mintzberg, Ahlstrand and Lampel (1998, p. 208-9) argue that the role of leadership is not to preconceive deliberate strategies but to manage a process of strategic learning in which novel strategies can emerge, noting that strategies often first appear as patterns from the past and only later become plans for the future.

Even strategy perspectives that emphasize formal analysis, planning and formal strategic choice increasingly emphasize learning and knowledge deployment strategies. Early research on the resource-based view of the firm (e.g., Wernerfelt, 1984) directed attention to knowledge-based resources that are imperfectly tradable in factor markets. Those focusing on the strategic management of technology argued that firm prosperity depends on the proper assessment and deployment of knowledge assets, including both information and competencies (Teece, 1986; Winter, 1987). These concerns naturally stimulated research on how such resources are created (e.g., Leonard-Barton, 1995; Nonaka & Takeuchi, 1995), and empirical studies of knowledge transfer between firms (Darr, Argote & Epple, 1995; Hamel, 1991; Ingram, 1997; Ingram & Baum, 1997), and in broader networks (Powell, Koput, & Smith-Doerr, 1996). Indeed, the study of dynamic capabilities lies at the heart of cutting-edge strategy research today (Iansiti & Clark, 1994; Teece, Pisano & Shuen, 1997), focusing on organizational systems for creating and harvesting knowledge that is difficult to imitate and offers sustainable advantage. Links between learning and competition also inform important work on the dynamics of international competition and industry evolution as well as research on the co-evolution of technologies and industries (Kogut, 1993; Teece, 1986).

At the same time, learning research has also moved center stage in organization theory. Dissertations and major academic journals increasingly highlight organizational learning (e.g., Argote & Epple, 1990; Lant & Mezias, 1992; Levitt & March, 1988; March, 1991; Miner & Mezias, 1996; Walsh & Ungson, 1991). The growing body of related empirical research builds on two identifiable traditions. The first tradition emphasizes learning through the selective retention of standard operating procedures or routines. In the traditional behavioral theory of the firm, search for new routines is triggered by periodic performance gaps (Cyert & March, 1963, 1992; Lant, 1992), while other models emphasize less purposeful search and varied selection processes (Burgelman, 1983; Miner, 1990, 1994). Consistent with very broad models of adaptive systems (Holland, 1996; Huberman, 1989) the key learning outcome in this class of models is change in behaviors of the learning system.

Another tradition envisions learning as a cognitive process, involving inference, computation, and the development of understanding (Glynn, Lant & Milliken, 1994; Weick, 1979). Here, the key learning outcome is change in knowledge structures or mental models. Recent research has supplemented previous simulation (Levinthal & March, 1981; Louanama & March, 1987) and qualitative studies or conceptual studies (Cangelosi & Dill, 1965) with systematic empirical research. Contemporary research addresses learning subprocesses such as knowledge creation, memory formation and access, search, forgetting and knowledge transfer, and has provided evidence of distinct individual, group and organizational level learning processes (Argote, 1999; Epple, Argote & Devadas, 1991; Levinthal & March, 1993).

In this volume of Advances in Strategic Management, we bring together a set of papers that link strategy and learning models, with a special emphasis on how learning at several levels of action produces change in collectivities of organizations. We define a population level learning outcome as systematic change in the nature and mix of routines, strategies or practices enacted in a population of organizations, arising from experience (Miner & Haunschild, 1995). The population level learning framework focuses on how repeated learning processes affect the distribution of routines and competencies in an organizational population. These iterated learning processes include recurrent (1) independent organizational, (2) interorganizational, and (3) collective learning. The framework explores how iterated learning at several levels of analysis will generate systematic population level change. We believe many of the patterns discussed will have important implications for organizational fields, communities and regional clusters, but emphasize industry transformation in this volume and use population and industry interchangeably.

Although each paper contributes to its own distinct nexus of theory and practice, taken as a whole they illuminate three central issues in strategic management:

  1. Industry structure and learning: Does recurrent learning at different levels of action tend to produce convergence of practices and organizational forms in industries and organizational populations? Or will it produce emergent substructures – which may later influence further organizational action? These questions address a central puzzle of strategic learning: how can organizations differentiate themselves and develop distinctive capabilities at the same time they learn from others? They illuminate how nonlinear industry transformation processes affect important strategic choice considerations for individual organizations.
  2. Industry evolution and learning: What factors moderate how recurrent organizational, interorganizational and collective learning influence population/industry transformation? If recurrent learning at varied levels of analysis can produce different population level transformation patterns, what are some common moderators of these outcomes? These concerns address factors that strategists would need to consider in attempting to anticipate how their own and others’ repeated learning processes may produce unanticipated population/industry transformation patterns over time.
  3. Industry attractiveness and learning robustness: How do different learning processes influence population/industry survival and prosperity? Does recurrent learning at one level of analysis help learning at higher levels? For example, can firm and population level learning processes conflict, creating tension between adaptive learning at different levels of analysis? These concerns focus attention on whether organizations may benefit from considering both their own and industry-level survival, and on factors making industries more or less attractive to learning organizations. They highlight issues relevant to strategists seeking to enhance the fate of industry or regional groups.

In Table 1, we indicate links between specific chapters and these issues, based on our perceptions of the papers’ explicit and implicit implications.

TABLE 1

Insight into these three broad areas takes on increased urgency as organizations and industries or organizational populations actively seek to use learning strategies and manage knowledge in pursuit of survival and prosperity. These issues also resonate with important theoretical questions regarding organization learning more broadly and its links to strategic management. The framework may also have special interest as organizational boundaries become more fluid, because it focuses on practices and guiding recipes, rather than on numbers of organizations in the population.

In this chapter, we first clarify the population learning framework and definitions that inform our approach, to lay the foundation for considering specific papers. Next, we highlight selected ways specific papers provide insight into the three questions above, noting implications for strategic management. To conclude, we note selected additional themes raised by the papers, highlight areas we see as ripe for further research, and sketch key contributions of the population level learning framework to selected schools of strategic management and organization theory.

LEARNING OUTCOMES, PROCESSES, AND LEVELS

In this section, we introduce our approach to learning outcomes, processes and levels, to clarify the overall population level learning framework. We then build on these ideas in the following section to argue that strategy development and deployment unfold in the context of population level learning. After summarizing several illustrative links between three key strategic issues listed above and population level learning concepts and themes, we consider each strategic issue’s links to the papers in more detail.

Learning as process and outcome. One reason "learning" papers usually start by defining the concept is that many plausible learning types and levels have been subsumed under this construct in varied literatures. Despite – or perhaps because of – the burgeoning interest in learning, the term itself has been somewhat contentious. Is cognition and understanding essential for learning? Does knowledge creation represent learning? At what level(s) of analysis does learning take place? Early writers were frequently charged with committing the ecological fallacy, inappropriately transporting concepts from one level of analysis to another and engaging in reification of groups and organizations. Do organizations learn, or do only individuals learn? Do organizations have memories?

We begin with a basic, but inclusive definition of the process of learning: An agent learns when experience systematically alters its behavior and/or its knowledge (Argote, 1999; Miner & Mezias, 1996). This definition characterizes both simple trial and error repetition of apparently fruitful actions and complex intellectual discovery processes as forms of learning. Consistent with this definition, we see a change in the distribution of routines, practices or behaviors in the learning system as an important learning outcome. This might, but does not necessarily, include cognitive routines such as scripts, mental models or theories. Like March and his colleagues, we define the learning outcome in terms of a change in organizational routines or practices, not through outcomes such as greater productivity (Levitt & March, 1988).

The link between a context and behavioral routines evoked in that context may or may not, then, be mediated by a knowledge structure. For example, an adaptive system that adjusts future behavior through a simple feedback loop can be seen as ‘learning,’ although there is no notion of intellectual activity. On the other hand, sometimes learning involves creating or updating a mental map, theories about cause and effect, or taken-for-granted assumptions about constraints or possibilities (see Walsh, 1995, for a review of managerial cognition and collective memory). In this case, experience alters behavior by re-shaping the connections in a knowledge structure, which can lead to subsequent change in the distribution of behavioral routines, practices or strategies. The process involves interaction with the world – or experience – with action informing understanding and understanding potentially informing action (Daft & Huber, 1987).

Levels of learning. Viewing a learning outcome as change in the distribution of routines and practices in a system puts a different light on concerns about levels of analysis in learning. Rather than asking whether only individuals, or also collections of individuals can learn, we point out that the distribution of routines, or patterned behavior can shift at many levels of analysis. Clearly experience can systematically alter the future behavior and/or knowledge structures of both individuals and collective actors.

In recent years scholars have increasingly emphasized that learning at one level of analysis is not merely the simple aggregation of learning at lower levels of analysis. Even the individual mind may be thought of productively as a complex ecology of interacting ideas (Bateson, 1972). Complex systems of interaction produce emergent properties that cannot necessarily be predicted on understanding the system’s components’ individual propensities (Cowan, 1994; Hutchins, 1991; Simon, 1996). Learning at one level may be unrelated to, or even inconsistent with effective learning at another level. For example, Argote (1999) shows in detail how learning at the individual level can inhibit learning at the level of social groups, and a simulation by Lounamaa & March (1987) suggests simultaneous rapid learning by different parts of organizations can reduce learning effectiveness. These considerations suggest that collective learning is a worthy object of study in its own right. We cannot always reduce a collectivity’s learning to the sum of learning processes at the next lower level of analysis and action.

For this reason, to understand learning as a driver and instrument of strategy, we not only need to investigate lower levels of analysis such as learning by individual strategists or top management teams learn (e.g., Huff, 1990; Reger, 1990). We must also examine the context provided at higher levels of analysis. Organizational learning must be set in the context of learning processes and outcomes at the next higher level of analysis, involving multiple organizations that interact as they learn (March, 1991). The emergent pattern of behavior that can become strategy at the organizational level both influences and is influenced by the distribution of routines, strategies and practices in the collectivity or population to which it belongs. Expanding the definition of Baum (1996, p. 77) we define a population of organizations as a set of organizations that share at least one major semi-stable trait, activity or resource utilization pattern. An industry, for example, would represent an organizational population; a regional collection of organizations might also represent a local organizational population.

Population level learning as outcome. Miner and Haunschild (1995, p. 115) originated the term population level learning to denote the outcome of "systematic change in the nature and mix of routines in a population of organizations as a result of experience." Miner and Haunschild used the term "routine" to refer to a broader class of activities than narrow, repetitive sets of micro-actions in organizations. Their use of the term included systematic bundles of activity such as practices, programs, search rules, strategies, and other patterned behavioral or cognitive activity (Miner & Haunschild, 1995).

These routines can be enacted at many levels of activity; they might include organizational roles, group practices, organizational strategies, or routines enacted at the level of a population itself, such as an industry-level coordination ‘routine’ used in a research consortium or trade association (Aldrich & Sasaki, 1995; Aldrich Zimmer, Staber, & Beggs, 1994). Because it is based on experience, this learning outcome is distinct from shifts in distribution of routines that are the result of differential births and deaths of organizations. That is, the population level learning outcome is not a compositional effect, in which some routines, practices and strategies become more frequent because organizations which implement them become more frequent. Rather, the change in distribution is itself the product of experience of some unit or units in the population.

For example, consider the U.S. radio industry in the 1960s and 1970s. AM radio stations in the United States exhibited a distribution of programming formats skewed heavily toward top-40-singles music. By the 1990s, most music formats were offered by FM radio stations, and the distribution of AM formats was skewed toward syndicated talk-radio programs. How did this shift in the distribution of routines over time occur? In part, the transformation was driven by differential births and deaths – founders of new AM stations tended to adopt talk formats instead of music formats, and stations featuring music formats tended to be sold to new owners who switched to talk formats. However, this transformation is also in part a story of population-level learning. AM radio stations learned from observing other stations that talk radio could attract listeners and advertisers, so they changed routines. Additionally, supra-organizational institutions emerged that made it easy for stations to institute talk radio formats (in this case, widespread syndication and the emergence of recognized talk radio personalities with nation-wide visibility) (Smith, Wright & Ostroff, 1998).

A population level learning outcome, then, represents a sociological fact that characterizes a collective entity, not its individual members. Individual or groups of organizations enact routines; populations exhibit a distribution of routines. To say that a population level learning outcome has occurred is to say that the distribution of routines (practices and strategies) in a population has shifted between two time periods, as a result of experience somewhere within the population.

We emphasize three learning processes that can produce this population level outcome, each at a different conceptual level of analysis. First, parallel independent organizational learning can produce new distributions of routines. As Carley shows (this volume), a set of organizations that are initially similar can learn independently in the same context with the same tasks, yet end up stratified in terms of practices and performance. Variety in practices arises over time as the result of different individual organizational experiences and learning trajectories.

Second, recurrent interorganizational learning can produce new distributions of routines. When an organization learns by observing other organizations, by exchanging knowledge with others, or by generating knowledge through joint interaction, this represents interorganizational learning. This learning may be either strictly vicarious (when one organization observes but does not interact with another) or interactive (when the learning process arises from active contacts between two or more organizations). Considerable research focuses on the impact of interorganizational learning on the individual learning organization. Does an organization gain from attempting to share knowledge in alliances (Aldrich & Sasaki, 1995; Mowery, Oxley & Silverman, 1996; Powell et al, 1996)? Can a second mover learn effectively from observing a first mover (Antonelli, 1990)? Yet it often stands silent on how repeated instances of interorganizational learning influence the shape of an entire industry or organizational population.

Third, whole industries and populations can observe other industries or populations, drawing on their shared experience to enact or inform population-level routines. This represents a collective population learning process at the population or industry level. A collective population-level learning process occurs when at least one learning step – such as knowledge discovery, retention, retrieval from memory, or contemporary experience – is collective. For example, a population/industry may observe its own shared experience and collectively develop an inter-organizational coordination device (a population level routine). It might also develop a formal collective entity after watching another population/industry do so, build norms based on prior shared experience, or evolve shared mental models of population/industry identity, boundaries or performance standards. The United States semiconductor industry’s observation of Japanese research consortia, and the consquent development of its own research consortia (such as SEMATECH; see Corey, 1997) represents a collective learning process. An emerging industry’s development of a trade association group in response to the shared experience of an external threat could also represent collective learning.

An important goal of this volume is to explore the claim that learning processes at one level may or may not mirror learning processes at another level. At the same time, it is clear that the degree to which learning processes are collective represents a continuous rather than a discrete variable. For example, if an industry association (a collectivity) drew on data about industry level performance (information from the memory of the whole population) to develop a new coordination routine between industry members (new population level routine) and codified it in an industry level rule (new form of population level memory), an entire learning process would be collective at the industry level. At the other extreme, population level learning might involve primarily one collective learning step combined with other processes. For example, organizations might pool their separate experiences with different technological variants to decide on a shared industry standard. The original knowledge acquisition at the firm level would have been organizational, but the codification into industry-level norms to which all must conform would represent a collective process.

While most papers in this volume tend to emphasize one of the three levels of learning processes, many also consider more than one process, and address partially collective learning processes. Carley, for example, presents a simulation study of parallel independent organizational level learning; papers by Baum and Berta, Greve, Ginsberg, Larsen and Lomi, and Mezias and Eisner primarily explicate processes of interorganizational learning while Anderson, Lant and Phelps, Miner et al., and Rura-Polley describe collective learning processes. At the same time, these authors frequently show how learning processes occur at more than one level, and most imply that different learning processes influence population level transformation in non-obvious ways. The power and complexity of learning’s impact on industries and organizational populations can be overlooked when one focuses solely on strategy as organization-level learning.

 

LINKS BETWEEN POPULATION LEVEL LEARNING

AND STRATEGY DEVELOPMENT AND DEPLOYMENT

Why should students of strategic management care about population level learning outcomes and the learning processes that produce them? Industry and competitive analysis (e.g., Oster, 1999; Porter, 1980) represents a set of frameworks that scholars and managers use to understand the context of an organization’s strategic position. It helps us assess which industries are more or less attractive environments, and how a firm’s strategy positions it with respect to other enterprises. Much traditional industry analysis has drawn on fairly static concepts of industry structure. The nature and mix of routines, strategies and practices in a population that result from experience is the context in which the strategy-making process takes place. By considering how learning processes produce such changes, we begin to develop models of dynamic transformation of the industry and population context for firm level strategies. Theories about population level learning outcomes help us assess why some organizational populations learn faster than others do, and how a firm is positioned with respect to other enterprises whose experience influences its own routines, strategies and practices.

In the following three sections we consider specific implications of the papers in this volume for each of the three major themes: (1) industry structure, (2) industry evolution, and (3) industry attractiveness and robustness. To highlight some of the common themes and implications of the chapters, Table 2 summarizes some of the key insights developed in this volume that relate to each factor. Specific rationales and nuances of related papers and arguments are considered in the detailed sections in Table 2.

TABLE 2

Industry structure and learning

If firms are repeatedly learning from their own experience and that of others, one might expect populations and industries to become homogeneous in their routines, strategies and practices. In the well-known evolutionary economics model of Nelson and Winter (1982), for example, it is assumed that each firm can copy any superior technology in the neighborhood of its current techniques, and can imitate the practices of higher-performing firms, subject to some error. If firms learn repeatedly from the same reality, or from each other, how can they create and sustain competitive advantages over time in knowledge-based competition?

Several papers in this volume illuminate aspects of this central puzzle in strategic management. They explicate how repeated learning at several levels can produce inter-firm differences. Firms can become segregated into persistently different performance categories characterized by different learning paces. Patches of collective strategies can emerge that are isolated from each other, and populations can break up into local learning neighborhoods. Importantly, heterogeneity does not simply arise because of barriers to efficient or effective learning; it is generated by the dynamics of learning processes themselves. Furthermore, in some cases, learning produces population/industry level contours that shape further organizational learning and population/industry transformation, so that the trajectory of learning over time is heavily dependent on history. We highlight in this section four specific factors that may influence industry structure: recurrent independent learning, emergent patches of collective strategies through cycles of competition and learning, collective learning, and changing diversity in pools of routines.

Stratified industry/population through repeated independent learning. Carley (this volume) describes results of a simulation study that models internal organizational learning with individual level learning and organizational features that shape individuals’ interactions. Sets of organizations begin with roughly similar features, and face the same tasks. Although the organizations are all learning, they do not converge to a single type of organization or set of organizational features in the population as a whole. Repeated parallel independent organizational learning (by organizations that do not learn by observing each other) can lead to different strata, each containing a set of organizations that is locked into a particular change strategy. Although there is oscillation in specific rankings, the high-performing organizations in one period are also typically the high performers in the next.

Carley observes that high performers in this stratified system do not necessarily share fixed features with each other. Rather, their internal learning processes and pace of internal change tend to be similar. Repeated independent organizational learning creates a shift in the distribution of organizational performance, traits, and behavioral learning routines in the population. The latter appears to be most relevant to the performance outcomes for individual organizations. The important general insight is that quasi-stable industry patterns emerge not because some organizations are learning from their experience while others are not, but because of detailed, local features of learning processes within each organization. This in turn affects each organization’s specific internal networks and procedures, which, in a path dependent process, produces quasi-stable industry/population distributions of routines, strategies and practices – along with persistent performance differences.

This simulation’s results contradict the popular assumptions that repeated independent learning by initially similar firms charged with the same task will necessarily produce convergence to a shared set of practices and routines. It supports predictions that firm level learning mechanisms tend to lock in, which in turn produces population level patterns that are not easily altered. Firms can’t generate superior performance simply by emphasizing the importance of learning, or by imitating other population members whose results are superior. It implies that learning firms need to attend to the fine-grained details of their internal learning pace and processes if their learning is to provide competitive advantage, and that they can expect high and low performance to persist even all firms pursue a learning strategy.

Emergent patches of collective strategies. In Ginsberg, Larsen & Lomi’s simulation (this volume) firms compete in a rounds of a prisoner’s dilemma within neighborhoods of interaction, then adopt the strategy of the winner of the competition in their own neighborhood. This method of determining their competitive strategy for the next round can be seen as interorganizational learning in the form of outcome imitation (Haunschild & Miner, 1997). The simulation explicates how repeated rounds of competition and interorganizational learning, under suitable conditions, produce patches of organizations that follow a collective cooperative strategy: the firms in each patch ‘lock-in’ to cooperation with each other. Their model shows how local decision rules followed over time induce and sustain aggregate regularities in the industry/population as a whole. Ginsberg et. al. find that the initial proportion of cooperative strategists in the population influences whether patches of interfirm cooperation form. Increased scope of interaction produces larger islands of cooperation in this context. They interpret this range of interaction in terms of the cognitive limitations of participating organizations, but the range of interaction could also be determined by such factors as shared sense of identity among organizations or simple geographic proximity (Anderson, this volume; Lant & Phelps, this volume).

Like Carley’s simulation, this paper shows how learning at a lower level of analysis can produce regularities at the population/industry level that then shape future interactions among organizations. In these models, collective learning emerges from interaction. Lower level learning produces stable contours at the population/industry level that constrain the ongoing strategies and behavior of firms in the future, as well as the learning strategies open to these firms. Again, these models imply that firms pursuing a learning strategy may need to consider whether the collective impact of firm learning may be creating broader industry/population patterns that will constrain the focal firm in the future. For example, in Ginsberg, Larsen & Lomi’s set-up, a firm that prefers to maintain a cooperative strategy might intentionally look for contexts that combine a high proportion of firms following a cooperative strategy within one’s neighborhood, a larger scope of interaction and more sensitivity to diffuse competition in the process.

Collective learning and population/industry heterogeneity. Anderson (this volume) tracks actual changes in the distribution of two competing production routines in the cement industry. He describes how the U.S. cement industry moved from a period in which dry mixing dominated, through a period in which wet mixing of raw materials was most common, and then returned to a preponderance of dry mixing. Early in the period studied, that U.S. cement industry developed a shared emphasis on quality as the key competitive dimension. Customers had no reliable technical way to assess cement quality directly, so they used social cues, reputation or high price as indicators of quality.

After World War I, a new wave of foreign competition led to a shared belief that thorough mixing of ingredients produced better cement that achieved load-bearing strength more quickly. Although this view was not grounded in direct scientific evidence, it became a shared, taken-for-granted belief – derived through interorganizational learning – that helped generate a shift to the use of water to gain more thorough mixing at the cost of fuel efficiency and throughput. PCA-sponsored research eventually generated ways to actually test quality (e.g., durability) of cement directly, which in turn permitted technical standard levels to be adopted by the industry. It also vastly improved dust control, eliminating a key advantage of the wet-mixing process. Eventually, through further innovation, dry mixing re-emerged as the dominant routine in the industry.

Both simple imitative learning and a change in knowledge structures occurred. Indeed, collectively sponsored research actually produced more accurate shared theories about what produced more durable and reliable cement. More interestingly, the original shared performance standards shaped collective and firm level learning. But the results of this learning then transformed the industry’s performance standard, which provided the new learning landscape for both firms and collective learning. Which routine prevails as the distribution of routines shifts as a result of experience-and which firms win and lose as their routines fall into and out of favor--is a function of collective action, not economic superiority alone. Firms can affect the outcome of competition among clusters of routines by influencing how a population collectively defines which dimensions of merit matter, what standards collective bodies set, and whaT research trajectories such bodies pursue.

In another approach emphasizing collective learning, Lant and Phelps (this volume) argue that crucial learning occurs in the interactions between the organizations (rather than only within the firms themselves). They build on emerging sociocognitive theories of strategic groups and on situated learning theory to consider this process of collective population learning, drawing upon observations of the emerging media industry of Silicon Alley in New York City. Lant and Phelps go beyond the idea of shared mental models by conceptualizing population/industry level learning as actually occurring within the ongoing interaction of the firms over time. Lant and Phelps’ approach implies that the development of cognitive groups does not always result in the strategic myopia described in Scottish knitwear makers (Porac, Thomas & Baden-Fuller, 1989). They suggest that variations in perceived boundaries and strength of identity can generate heterogeneity within cognitive strategic groups.

Changing pools of diversity of routines. Rura-Polley (this volume) tackles potential origins of population/industry heterogeneity among practices by studying directly the nature and degree of variation in proposed childcare practices in two distinct organizational subpopulations: Catholic childcare institutions in Germany and the United States over a 75-year period. She argues these two national contexts produced very different patterns of variation among proposed practices, even though the overarching institution – the Catholic Church – maintained its underlying values and policies related to children’s institutions. Rura-Polley’s qualitative analyses indicate variation in these candidate routines and strategies – at least in the German case – reflects broader social trends. The degree and nature of variation in candidate practices was clearly different in the two countries. Temporal patterns were not the same, and the two countries differed in which topic areas showed most variation. Rura-Polley’s work implies that the candidate pool of routines/strategies available varies between subpopulations. This suggests industry structure may arise in part from the degree of available diversity of practices and strategists might consider the nature and pool of available practices in choosing industries to enter.

Taken together, these papers demonstrate ways in which all three levels of learning can, --separately and in combination – produce population/industry heterogeneity, which in some cases may be sustained through lock-in processes. One of the most important features of this insight, we think, is the notion that while individual firms learn, they may be creating unintended contours within their own industries that constrain their own fate later on. Insight into the dynamics of such processes provides an additional strategic factor for organizations whose fate will depend on the survival and prosperity of their industries as a whole.

Industry Evolution and Learning: Moderators of the Impact of Recurrent

Organizational, Interorganizational and Collective Learning

Population-level learning occurs when the distribution of routines – broadly defined – in an industry changes as a result of experience. Industry evolution occurs when the distribution of organizations themselves, or organizational forms in an industry changes over time. For example, when a change in the proportion of generalists versus specialists, M-form versus U-form enterprises, or independent firms versus joint ventures persists over time, the industry has evolved, with important strategic consequences. Since an organization’s routines help govern its form and indeed some forms can be seen as complex bundles of routines, it is clear that these two dynamic processes are interrelated. Traditional industry-level competitive analysis ascribes industry transformation to exogenous factors, such as technological or regulatory change, and to the winnowing effects of competition. The articles in this volume suggests that who organizations learn from and why they adopt routines from each other is an endogenous factor that also contributes to the transformation of a population over time.

The chapters highlight features of learning processes that may shape industry evolution: learning from near vs. learning from distal neighbors, within-group versus across group learning; and modes of organizational learning.

Learning from near vs. learning from distal neighbors Several papers explicate important ways in which distance between organizations – whether in terms of space or other dimensions – moderates the impact of learning processes on industry evolution.

For example, Greve studies organizations with multiple branches. The evolution of the radio broadcasting industry in recent years has been marked by a rise in the number of stations belonging to such multi-station, multi-geography companies (Smith, Wright & Ostroff, 1998). Is this because branch systems facilitate the transfer of routines from one region to another? In his empirical study of format changes in radio stations in varied regions, Greve finds that greater experience of a branch system’s units outside a given geographic market causes lower performance of each branch in a focal market. Greve also finds that the more experience a branch system has outside a given local market, the greater the variation in performance in that market. The results imply routines or format strategies transferred from distant neighborhoods actually harm rather then help the imitating stations. He argues that the difference in competitors between different regions creates this effect. These findings imply that if firms repeatedly imitate practices from distant neighborhoods, interorganizational imitation may decrease the immediate performance of imitators. However, Greve points out bringing in routines from distant organizational neighborhoods could help the local neighborhood as a whole, if other organizations learn from the failed routines that local branches borrowed from their sister stations.

Distal learning also occurs among organizations that do not share common ownership. Baum and Berta (this volume) study a group of firms competing in a behavioral simulation. They find that organizations learn from other organizations when their members interact with each other as individuals more frequently, and when the imitating organizations have similar market shares. Interestingly, they find that firms do not tend to imitate the most successful competitors in their local populations. Rather, they look further afield, imitating the most successful firms in other populations. Ginsberg, Larsen, and Lomi’s study suggests that the further away from their neighborhoods firms search for successful organizations to imitate, the less fragmented an industry becomes.

These studies imply the boundaries of neighborhoods of interaction will powerfully influence industry evolution, and that choice of neighborhoods of interaction as well as imitation may represent an important strategic choice for individual organizations.’

Within-group versus across-group learning: Neighborhood boundaries and their permeability. The distance between two organizations that may learn from each other is not simply a continuous measure in social or physical space. Industries can fragment into subgroups; firms that belong to different subgroups have to cross a boundary to learn from one another, even if they are close demographically or geographically. A change in the distribution of firms across such groups constitutes industry evolution, if group membership is relatively distinct and persistent. How are such changes influenced by learning processes?

Wade and Porac’s chapter in this volume builds a model suggesting that the movement of personnel among firms in an industry is an important conduit for interorganizational learning that influences subsequent industry structures. Industry evolution in turn creates opportunities that encourage or discourage certain types of managerial migration among firms. In this way, interorganizational learning (through the movement of people between organizations) and industry evolution are linked in a mutually causal loop.

Early in an industry’s history, these authors suggest, key managers from different firms tend to migrate from the same industries, bringing with them a common set of routines. For example, German experts greatly influenced the early practice of both the American beer brewing industry and the American cement manufacturing industry. Consequently, routines tend to be fairly homogenous across firms, and collective structures that promote collective population-level learning emerge fairly easily.

As an industry evolves, Wade and Porac suggest, it tends to split into two subsectors, a core and a periphery, differentiated by status, size, and/or performance. Organizations belonging to the core tend to monitor and learn from each other. Population-level learning takes place as routines diffuse through movement of managers from core to peripheral organizations, and through experimentation among peripheral firms, leading to the emergence of novel routines. When an industry is jolted by a shock, personnel flows tend to reverse; core firms begin to recruit from the periphery in order to import novel routines.

In considering implications of their observations, Wade and Porac argue the degree of permeability of an industry’s core may have important implications for the industry/population’s long term adaptation. If core and periphery boundaries are nearly impermeable, the core may fail to learn new routines, strategies and practices, and over time fade. In contrast, if the core/periphery boundary is permeable, personnel movement from the periphery may introduce routines and practices that permit the to core to adapt through the adoption of these new practices. Lant and Phelps make a similar argument, pointing out that permeable boundaries between cognitive strategic groups may permit new routines and strategies to move into established groups, providing variance permits adaptive learning that can enhance long term survival of the strategic group.

These theories imply industry evolution will be influenced by permeability of boundaries around subpopulations, and that important strategic action may include choices about the strength and type of boundaries maintained.

Taken together, these different moderating effects of neighborhoods on the population impact of interorganizational imitation have important strategic implications that flow from understanding who organizations target for learning. Greve’s work implies that learning from organizations too distant from oneself often reduces performance, yet imitating nearby competitors can produce escalation in which all firms ‘improve’ in absolute terms but not in relative position (the so-called Red Queen effect in evolutionary theory) (Kauffman, 1993; March, 1991; Van Valen, 1973). This points to the significance of carefully selecting interorganizational learning targets. These localized neighborhood effects on repeated interorganizational learning also suggest that strategists may benefit from considering not only their own targets for interorganizational learning, but also the general pattern of interorganizational knowledge flow in their own industry or strategic group. Superior selection of individual interorganizational learning targets could provide some firm level competitive advantage within an industry or strategic group. At the same time, inadequate permeability of neighborhood boundaries may reduce the viability of the whole industry/population.

Modes of interorganizational learning. Imitating other organizations constitutes one form of interorganizational learning: imitating another organization means using its experience as a basis for one’s own actions. The process may or may not involve forming inferences or causal theories about the imitated practice. On what basis do organizations choose which routines to imitate? Three different modes of interorganizational learning are addressed by various papers in this volume: (1) frequency based learning (when organizations adopt the practice of a large number of other organizations), (2) trait based learning (when organizations adopt the practice favored by organizations with a distinct trait such as size or prestige), and (3) outcome based learning (when organizations observe the apparent results of a practice and adopt it or not based on those results) (Haunschild & Miner, 1997; Miner & Raghavan, 1999).

Baum and Berta’s behavioral simulation presents further evidence for the use of more than one mode of interorganizational learning and imitation: their groups appear to choose learning targets both on the basis of traits and outcomes. Mezias and Eisner theorize that each learning/imitation mode produces a different pattern of industry transformation, due to differences in their relative speed and cost. They contend that outcome-based learning leads to the most rapid evolution when ambiguity is low, because successful new organization forms diffuse rapidly, and firms quickly exit highly competitive niches. However, they believe the costs of searching for a new form and adopting a new form are highest when firms employ outcome-based learning. Combining intuitions about costs and speed, Mezias and Eisner predict that both frequency-based and trait-based learning will lead to long periods of stability in the distribution of organizational forms, punctuated by short periods of rapid, large-scale change. In contrast, outcome-based learning will result in smaller, more frequent changes in an industry’s mix of organizational forms.

One implication of work on modes of imitation and interorganizational learning is that firms may profit from considering the broader pattern of interorganizational learning within which they function, so they can anticipate where their industry/population may be moving, given that repeated interorganizational imitation by multiple organizations can produce unexpected effects. Firms may need to pay attention to the interorganizational learning modes of other organizations – not to understand what specific knowledge a competitor may be gaining, but in order to gain some insight into how the collective pattern of imitation might transform the industry’s practices over time.

Population and industry attractiveness/robustness:

Learning’s impact on population survival and prosperity

It is often assumed if the individual organizations are all learning in sensible ways, the aggregate result will be adaptive for the population or industry as a whole. Yet the research in this volume shows this assumption to be false, and explicates how different learning processes may interact to promote or detract from population/industry survival and prosperity. We describe two important factors the chapters imply will shape how learning influences survival and prosperity of the whole population/industry: the degree and nature of tensions between organizational and population level learning, and temporal learning patterns.

Tensions between organizational and population level learning outcomes. Several papers imply that lower-level learning may not only have unintended impact on population/industry transformation, but may in some cases conflict with fruitful population level learning outcomes. These studies raise important questions about the impact that learning processes have on organizational and population/level performance, including both survival and prosperity.

Miner, Kim, Holzinger and Haunschild draw on historical reports of organizational populations and industry transformations to examine ways failure and near-failure by individual organizations may spur different degrees of collective learning processes. The authors describe learning processes in which failing organizations stimulate other organizations to avoid specific routines/strategies/practices, to search or imitate nonfailing organizations or to reinforce their own current practices. They also describe collective learning processes in which the population/industry develops new collective routines such as research consortia or federated political groups or spur compensatory actions.

Miner et al. argue these processes produce shifts in the distribution of routines/strategies in a population in several nonobvious ways. For example, they observe how dominant firms may imitate challengers in the periphery of an industry and fail in their own revival while legitimating the routines they imitate. This does not imply that all learning from the failure of others is valuable. However it suggests that the failure or near failure of some organizations in many cases does spur fruitful learning by other organizations and the population/industry as a whole. In fact the authors argue the population/industry may learn more from failures or near failures than from apparent successes, because of their greater visibility and potential for developing more accurate causal models.

Similarly, Greve shows in his radio study how organizations that imitate practices from distant regions reduce the level and increase the variance in their own performance. He suggests this helps the population/industry to which they belong by introducing new routines/strategies and allowing others observe the impact of these new routines/strategies. He argues such firms engage – in effect – in altruistic learning: their experience detracts from their own performance but may enhance that of the industry/population to which they belong.

Three other papers point to the complementary situation in which fruitful firm-level learning detracts from learning at the population/industry level. Lant and Phelps, Wade and Porac, and Carley all describe organizational learning processes that reduce variance within the firms themselves. They point out that this useful firm-level learning may restrict population/industry learning and adaptation by reducing the variation in routines/strategies available in the future. These situations can be seen as the reverse of the failure argument, which hinges on the idea that contrasts between failure, near-failure and success provide the variance in both behavior and outcome that represents the most valuable learning context.

At the heart of all these studies is the idea that individual organizations face trade-offs between potentially valuable organizational and population-level learning processes. The general dilemma is familiar to firms in the area of knowledge deployment in research consortia, where firms try to balance supporting collective discovery of knowledge useful to the entire industry with safeguarding their own knowledge and avoiding free-riding by others (Mowery, Oxley & Silverman, 1996). The papers in this volume suggest the dilemma may apply more broadly than in the obvious context of research and development. Efficient solutions to this puzzle are not apparent. It is hard to devise appropriate systems of side payments that reward failing firms for enhancing the knowledge of others or deviant, low-performing firms for preserving variance potentially useful to the whole population. It suggests that collective consortia and industry associations could deliberately facilitate diversity in the populations/industries they represent, rather than relentlessly promoting ‘best practices’ in all members. At the same time, structural features of the population’s competitive system can crucially influence the best strategy for the firm’s learning, consistent with March’s (1991) description of ecologies of learning.

Population/Industry performance impact of temporal learning patterns. March (Lounamaa & March, 1987; March, 1991) has illustrated potential dangers of overly-rapid learning in simulations of organizations learning from their own experience, especially in the presence of noise and multiple learning entities. The papers here underscore potential tradeoffs between fast and slow experiential learning at the population/industry level.

Baum and Berta’s paper explores the important question of how speed of population level learning affects industry performance and survival. Speed is an important issue in all forms of learning (Argote, 1999). Baum and Berta provide the first systematic empirical evidence supporting the prediction that populations will face windows of opportunity in which learning is most fruitful at the population/industry level (Miner & Haunschild, 1995). Baum and Berta find their behaviorally simulated population’s speed of overall learning (using numbers of learning events in specific time periods) affected their population average performance as predicted. Population performance suffered in very speedy and very slow learning populations. The authors argue that some populations learn too slowly to harvest learning while others learn too rapidly to discover effective routines.

Mezias and Eisner’s discussion of varied modes of interorganizational learning also explores the impact of speed in population level transformation. Mezias and Eisner argue outcome based imitation will result in the most rapid diffusion of a new high performance organizational form, while outcome imitation will produce more frequent change in the mix of forms, but the magnitude of such changes will be more modest. Since these learning dynamics may themselves affect population/industry survival in the face of competing populations/industries, imitation mode may affect population survival or prosperity indirectly through its impact on the pace of population level learning.

These and other papers taken together imply the relative importance of behavioral versus cognitive learning may moderate the impact of both speed and variance (including contrasts between failure and success) in the impact of different learning levels on population level performance. If other organizations can easily generate valid insights and strategies by observing failing organizations, including those that fail from overly rapid organizational learning, the population/industry may thrive in the presence of high failure rates or overly-rapid learning by individual organizations. To the degree such population/industry such inferential learning is difficult, dangers of overly rapid organizational learning increase.

THE FUTURE OF POPULATION LEVEL LEARNING RESEARCH

The population level learning framework represents work in progress. Although learning models are attractive because they embrace dynamic processes and fit contemporary emphasis on the role of knowledge, their very richness and complexity is something of a disadvantage in focusing further research. By raising several common themes in different contexts, the studies in this volume help offer a focus for further investigation, four of which we emphasize here.

More systematic empirical research on how features of neighborhoods of interaction moderate learning’s impact on population/industry heterogeneity, survival and prosperity shows promise. It may make sense to contrast learning models here with related work examining regional populations and more traditional agglomeration processes (Krugman, 1995). Similarly, empirical research on the impact of different interorganizational learning modes will be important. Careful attention to existing formal models of interorganizational imitation unarguably implies that interorganizational learning processes can produce heterogeneity (Miner & Raghavan, 1999). Because most prior empirical research has emphasized only one learning mode, we have little systematic evidence about their actual relative impact on industry and population outcomes and implications for strategic management.

One important notion presented in this volume is that certain population level contours, such as shared performance standards or even identities, can arise from learning processes, but then also shape future behavior, including future learning at lower levels. This important insight implies that potentially unanticipated collective outcomes may limit future strategic choice for learning organizations in unexpected ways. Systematic empirical research on such processes may reveal that they are much more than simple moderators influencing the impact of repeated organizational or interorganizational learning. Research on the competitive structures in which learning outcomes are judged may also have important strategic implications (March, 1991).

Deeper exploration of interactive and situated learning, as well as the notion of networks as the memory banks for populations or industries deserve continued attention (Koput, Smitth-Doerr & Powell, 1997; Suchman, 1995). In fact, the broad definition of learning used in the population level learning framework mandates careful empirical work to tease out what specific learning processes apply under what conditions This emphasis is consistent with growing empirical work on specific learning processes (Argote, 1999). The notion that networks may represent some form of memory – a learning element – for example, represents an important focus for continued empirical work. It also makes sense to examine related analytic work done by evolutionary economists and historians who have considered learning models in some detail (England, 1994).

Other important issues for continuing work were not highlighted in this volume. For example, related prior research indicates the level of collective experience can affect organizational survival chances, which in turn influence population/industry survival and prosperity. (Baum & Ingram, 1998). Research on ship-building and hotels suggested individual organizations may not be able to continue learning from others after their founding (Argote,Beckman, & Epple, 1990; Baum & Ingram, 1998). However, the potential impact of what can be learned after start-up is relatively weak compared to irreversible initial design and strategy choices because of the large capital investments needed in these industries. Future work should examine conditions under which ongoing population experience may enhance the survival and growth of current firms.

Additionally, clearly, some organizations are better at learning than others are (Carley, this volume; Leonard-Barton, 1995; Nonaka & Takeuchi, 1995). Improving the organization’s ability to learn, from its own experience and that of others has been a principal focus of the burgeoning literature on organizational learning and knowledge management (e.g., Grant, 1996). Further work exploring firm level differences in capabilities in learning from others represents an obvious focus for future research, going beyond the notion of absorptive capacity to a more comprehensive vision of strategic choices in targets, modes and capabilities for interorganizational learning.

The population-level learning framework suggests that in addition to studying the internal mechanisms through which firms generate or absorb routines, we should understand the context within which they learn. For example, Wade and Porac describe how transferring knowledge from managers who bring core-firm experience to a firm operating in the periphery can be helpful or not, depending on how industry evolution alters the opportunity structure, opening or blocking career paths within core enterprises. Similarly, Carpenter and Westpal (1999) showed how the influence of board interlocks on firm strategic decision making depended on the external, social-structural context in which the directors were embedded. More precise understanding of these areas would have value not only for firm level strategists, but also for collectivities of firms and groups such as policy makers charged with the well-being of organizational populations or industries rather than individual organizations.

International and historical research by Kogut (1993) and others (Nelson, 1993) pointed to a final area we think promising for continued research: the dynamics of populations, regions and industries learning from each other. In many cases, such learning may involve the transfer of population-level routines including not only technological standards, for example, but also routines or strategies for coordinating members of a population. Additionally, while we have emphasized industries as important populations, populations defined by geography and culture represent an important frontier. Waller, Gibson and Carpenter (1999) for example, noted that cultural differences were likely to influence the type, amount and pace of knowledge transfer at both firm and population levels. Insight into these processes can inform strategists concerned with the survival or growth of their own industries or regions, and appear increasingly relevant in contemporary economic settings.

THEORETICAL VALUE

We think that a focus on how recurrent learning influences population and industry transformation makes a distinct contribution to scholarship in both strategic management and organization theory.

For strategic management theory, this work – as we note in several places above – supplements the existing focus on interactions between learning and competition from the firm’s point of view. It highlights the potential importance of firms trying to understand the nonlinear dynamics of the way their own learning may alter the population/industry context in which they act and continue to learn. It is consistent with efforts to think more carefully about how regions, nations or industries can promote their own survival and prosperity in the face of knowledge creation and learning by competing regions, nations and industries (Kogut, 1997; Porter, 1990; Teece, 1986;).

This work also continues to develop themes raised by the resource-based view of the firm, through emphasizing the importance of understanding mixes resources in the form of routines, strategies and practices at a higher level of analysis. The framework also opens the door to conceptions of industry development that do not take the most crucial unit of analysis as the firm itself. To the degree that it emphasizes organizations, it focuses attention on learning differences among firms rather than structural regularities. This slight adjustment of focus may offer insights we could not easily achieve when we assume the nature and distribution of firms represent the most sensible unit for thinking about population/industry prosperity. This emphasis may also be timely. Attention to changes in the mix of routines enacted in a population directs attention to changes in what is actually getting done rather than to changes in organizational size or boundaries. To the degree that organizations are increasingly transient or maintain very cloudy borders, this focus may be increasingly relevant.

In organization theory, the population level learning framework seems a natural extension of learning theories, which have moved from individuals, to groups, to whole organizations but stopped short of any higher level. By examining the impact of learning on higher-level collectivity the population level learning framework responds to Stern and Barley’s call to address organization theory’s ‘neglected mandate:’ of examining the impact of organizations on the broad social systems in which they [are] embedded (Stern & Barley, 1996).

While overlapping in some areas, our framework differs in emphasis from several related perspectives, however. Research on the diffusion of innovations, for example, tends to study the fate of individual innovations over time rather than the overall mix of practices. It could not easily account for Anderson’s case of a single routine dominating, receding, and returning to prominence in a population, nor can it easily consider how routines interact with each other.

The population learning framework also provides a different account of many processes than does neoinstitutional theory. The population level learning perspective predicts the presence of other organizations early in an emerging industry may enhance survival of new firms and industry growth because real discovery of useful knowledge is occurring and advancing through interorganizational learning. This is a different claim for why increased density may enhance survival than arguments based on social legitimacy (Aldrich, 1999; Delacroix & Rao, 1994; Miner & Haunschild, 1995).

This framework also implies that organizations may converge on some traits while diverging on others, even without the influence of social legitimacy questions, due to variation in learning modes targets, and neighborhoods. The framework also differs from organizational ecology approaches to industry transformation by shifting the focus of attention to changes in distributions of routines, strategies and practices in a population rather than to distributions of organizational forms.

Learning processes that involve pure behavioral trial and error learning can often be conceptualized as evolutionary processes at a lower level of analysis and can be a matter of taste which metaphor is used. Some conceptualizations of nested evolutionary systems, for example (e.g., Singh & Baum, 1994) can be seen as a type of learning at a higher level of analysis, but the population learning terminology may not be needed for additional insight. To the degree transformations involve aspiration levels, deliberate experimentation, knowledge structures and cognition, however, learning models contain elements not easily incorporated in traditional evolutionary models of change. These models often emphasize blind variation rather than search, and do not envision a separate system of representations of the world as change goes on. Cultural evolutionary theorists (Boyd & Richerson, 1985) offer approaches closer to our framework, but have tended to emphasize individual level processes. Our perspective predicts organizations and industries may sometimes achieve useful and valid mental representations of the world. Yet by assuming barriers, stages and limits to learning the framework also makes different predictions than do economic models that assume efficient knowledge spillover.

CONCLUSION

Vital strategy research has long focused on how firms can balance knowledge creation and deployment with their efforts to find sustainable competitive advantage (Boisot, Griffiths & Moles, 1997; Teece, 1986; Teece, Pisano & Shuen, 1997). Organizations seek to gain knowledge efficiently and quickly from others, yet be able to harvest the value of their own knowledge and competencies without their appropriation by others. Difficulties in accomplishing this, combined with increasing rates of change in contexts of competition and technology have persuaded some that learning itself represents a crucial organizational competency (e.g., Lei, Hitt & Bettis, 1996; Teece, Pisano & Shuen, 1997). At the same time, international competition and the pace of technological change have brought to light ways in which entire populations or industries compete with each other over time (Kogut, 1993; Porter, 1990).

The papers in this volume explore how parallel independent organizational learning, recurrent interorganizational learning and collective population learning processes can change in the nature and mix of routines, strategies and practices in a population of organizations (or industry). The population level learning perspective supplements existing work with two distinct emphases. First, it directs attention to the impact of recurrent learning at different levels on population level transformation, rather than only on the firms directly involved in the learning. Second, it emphasizes change in the mixes of routines/ practices in the population, rather than change in the mix of organizations themselves. This focuses attention on what the organizations in a population are actually doing rather than on how many organizations the population contains, a focus that may be more relevant if boundaries become more plastic. These emphases lead to models that often underscore multiple interactions between competition and learning, and suggest important issues pertinent to strategic managers and policy makers. We invite the reader to consider the work here, and to join the conversation represented by this rich set of papers.

 

REFERENCES

Aldrich, H.E., Organizations evolving. London: Sage Publications.

Aldrich, H.E., Zimmer, C. R., Staber, U. H., Beggs, J. J. (1994). Minimalism, mutualism, and maturity: The evolution of the American trade association population in the twentieth century. In J. Buam & J. Singh (Eds.), Evolutionary dynamics of organizations (pp. 223-39). New York: Oxford University Press.

Aldrich, H. E., & Sasaki, T. (1995). R&D consortia in the United States and Japan. Research Policy¸ 24, 301-316.

Antonelli, C. (1990). Profitability and imitation in the diffusion of process innovations. Rivista Internazionale di Scienze Economiche e Commerciali, 37, 109-126

Argote, L. (1999). Organizational Learning: Creating, Retaining & Transferring Knowledge. Boston: Kluwer.

Argote, L., Beckman, S. L., & Epple, D. (1990). The persistence and transfer of learning in industrial settings. Management Science, 36, 140-154.

Argote, L., & Epple, D. (1990). Learning curves in manufacturing. Science, 247, 920-924.

Bateson, G. (1972). Steps to an ecology of mind; collected essays in anthropology, psychiatry, evolution, and epistemology. San Francisco: Chandler Publishing Company.

Baum, J.A.C. (1996). Organizational ecology. In S. R. Clegg, C. Hardy, & W. R. Nord (Eds.), Handbook of Organization Studies (pp. 77-114). London: Sage.

Baum, J.A.C., & Ingram, P. (1998). Survival-enhancing learning in the Manhattan Hotel Industry, 1898-1980. Management Science, 44, 996-1016.

Boisot, M., Griffiths, D., & Moles, V. (1997). The dilemma of competence: differentiation versus integration in the pursuit of learning. In R. Sanchez & A. Heene (Eds.), Strategic Learning and Knowledge Management (pp. 65-82). Chichester: John Wiley & Sons.

Burgelman, R. A. (1988). Strategy making as a social learning process: the case of internal corporate venturing. Interfaces, 18, 74-85.

Cangelosi, V. E., & Dill, W. R. (1965). Organizational learning: Observations toward a theory. Administrative Science Quarterly, 10, 175-203.

Carpenter, M., & Westphal, M. J. (1999). A network perspective on how outside directors impact strategic decision making. Academy of Management Best Papers Proceedings.

Corey, E. R. (1997). Technology fountainheads: The management challenge of R&D consortia. Boston. Harvard Business School Press.

Cowan, G. A. (1994). Conference opening remarks. In G. A. Cowan, D. Pines, & D. Meltzer (Eds.), Complexity: Metaphors, models, reality (pp. 1-4). Reading, MA: Addison-Wesley.

Cyert, R. M., & March, J. G. (1963).[1992] A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall.

Daft, R. L., & Huber G. P. (1987). How organizations learn: A communication framework. Research in the sociology of organizations, 5, 1-36. Greenwich, CT: JAI Press.

Darr, E. D., Argote, L., & Epple, D. (1995). The acquisition, transfer, and depreciation of knowledge in service organizations: Productivity in franchises. Management Science, 41, 1750-1762.

Delacroix, J., & Hayagreeva R. (1994). Externalities and ecological theory: Unbundling density dependence. In J. A. C. Baum & J. V. Singh (Eds.), Evolutionary dynamics of organizations (pp. 255-268). New York: Oxford University Press.

England, R. W. (1994). Evolutionary concepts in economics. Ann Arbor: University of Michigan Press.

Epple, D., Argote, L., & Devadas, R. (1991). Organizational learning curves: A method for investigating intra-plant transfer of knowledge acquired through learning by doing. Organization Science, 2, 58-70.

Grant, R. M. (1996). Toward a knowledge-based theory of the firm. Strategic Management Journal, 17, 109-122

Hamel, G. (1991). Competition for competence and interpartner learning within international strategic alliances. Strategic Management Journal 12, 83-103.

Haunschild, P. R., & Miner, A. S. (1997). Modes of imitation: The effects of outcome salience and uncertainty. Administrative Science Quarterly, 42, 472-500.

Holland, J. (1996). Hidden order: How adaptation builds complexity. Reading, MA: Addison-Wesley.

Huberman, B. (1989). The adaptation of complex systems. In B. Goodwin & P. Saunders (Eds.), Theoretical biology (pp. 124-33). Edinburgh: Edinburgh University Press.

Huff, A. S. (1990). Mapping strategic thought. Chichester: John Wiley.

Iansiti, M., & Clark, K. B. (1994). Integration and dynamic capability: Evidence from product development in automobiles and mainframe computers. Industrial and Corporate Change, 3, 557-605.

Ingram, P. (1997). Opportunity and constraint: organizations’ learning from the operating and competitive experience of industries. Strategic Management Journal, 18, 75-98.

Ingram, P., & Baum, J. A. C. (1997). Chain affiliation and the failure of Manhattan hotels, 1898-1980. Administrative Science Quarterly, 42, 68-102.

Kauffman, S. A. (1993). The origins of order: Self organization and selection in evolution. New York: Oxford University Press.

Powell, W. W., Koput, K. W., & Smith-Doerr, L. (1996). Interorganizational collaboration and the locus of innovation: Networks of learning in biotechnology. Administrative Science Quarterly, 41, 116-145.

Krugman, P. (1995). Development, geography, and economic theory. Cambridge, MA: MIT Press.

Lant, T. K. (1992). Aspiration level adaptation: An empirical exploration. Management Science, 38, 623-644.

Lant, T. K., & Mezias, S. J. (1992). An organizational learning model of convergence and reorientation. Organization Science, 3, 47-71.

Levinthal, D. A., & March, J. G. (1981). A model of adaptive organizational search. Journal of Economic Behavior and Organization 2, 307-333.

Levinthal, D. A., & March, J. G. (1993). The myopia of learning. Strategic Management Journal, 14, 95-112.

Lei, D., Hitt M., & Bettis, R. (1996). Dynamic core competences through meta-learning and strategic context. Journal of Management, 22, 549-569

Leonard-Barton, D. (1995). Wellsprings of knowledge. Boston: Harvard Business School Press.

Levitt, B., & March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14, 319-340.

Lounamaa, P. H., & March, J. G. (1987). Adaptive coordination of a learning team. Management Science, 33, 107-123.

March, J. G. (1981). Footnotes to organizational change. Administrative Science Quarterly, 26, 563-577.

Miner, A. S. (1990). Structural evolution through idiosyncratic jobs: The potential for unplanned learning. Organization Science, 1, 195-210.

Miner, A. S., & Haunschild, P. R. (1995). Population level learning. Research in Organizational Behavior, 17, 115-166.

Miner, A. S., & Mezias, S. J. (1996). Ugly duckling no more: Pasts and futures of organizational learning research. Organization Science, 7, 88-99.

Mintzberg, H., Ahlstrand, B., & Lampel, J. (1998). Strategy safari : A guided tour through the wilds of strategic management. New York: Free Press.

Mintzberg, H., & McHugh, A. (1985). Strategy formation in an adhocracy. Administrative Science Quarterly, 30, 160-197.

Mowery, D. C., Oxley, J. E., & Silverman, B. S. (1996). Strategic alliances and interfirm knowledge transfer. Strategic Management Journal, 17, 77-91.

Nelson, R. R. (1993). National innovation systems. New York:Oxford University Press.

Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Cambridge, MA: Belknap Press of Harvard University Press.

Noda, T., & Bower, J. L. (1996). Strategy making as iterated processes of resource allocation. Strategic Management Journal, 17, 159-192.

Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company. New York: Oxford University Press.

Oster, S. (1999). Modern competitive analysis (3rd edition). New York: Oxford University Press.

Porac, J. F., Thomas, H., & Baden-Fuller, C., (1989). Competitive groups as cognitive communities: The case of Scottish knitwear manufacturers. Journal of Management Studies, 26, 397-416.

Porter, M. E. (1980). Competitive strategy : techniques for analyzing industries and competitors. New York: Free Press.

Porter, M. E. (1990). The Competitive Advantage of Nations and Their Firms. New York: Free Press.

Quinn, J. B. (1980). Strategies for change: logical incrementalism. Homewood, IL: Richard D. Irwin.

Reger, R. K. (1990). Managerial thought structures and competitive positioning. In A. S. Huff (Ed.), Mapping Strategic Thought (pp. 71-88). Chichester: John Wiley.

Simon, H. A. (1996). The sciences of the artificial (3rd edition). Cambridge, MA: MIT Press.

Smith, F. L., Wright, J. W., & Ostroff, D. H. (1998). Perspectives on radio and television. Mahwah, NJ: Lawrence Erlbaum.

Spender, J. C. (1989). Industry recipes:The nature and sources of managerial judgement. Oxford: Blackwell.

Stern, R. N., & Barley, S. R. Organizations and social systems: Organiztion theory’s neglected mandate. Administrative Science Quarterly, 41, 141-146.

Teece, D. J. (1986). Profiting from technological innovation: implications for integration, collaboration, licensing and public policy. Research Policy, 15, 285-305

Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18, 509-533.

Van Valen, L. (1973). A new evolutionary law. Evolutionary Theory, 1, 1-30.

Waller, G., & Carpenter, M. (1999). Time’s arrow: The impact of differences in the time perspective of knowledge management in a multicultural context. Presentation, Academy of Management, Chicago, IL.

Walsh, J. P. (1995). Managerial and organizational cognition: Notes from a trip down memory lane. Organization Science, 6, 280-321.

Walsh, J. P., & Ungson, G. R. (1991). Organizational memory. Academy of Management Review, 16, 57-91.

Weick, K. E. (1979). The social psychology of organizing (2nd ed.). Reading, MA: Addison-Wesley.

Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5, 171-180.

Winter, S. G. (1987). Knowledge and competence as strategic assets. In D. J. Teece (Ed.), The competitive challenge: strategies for industrial innovation and renewal (pp. 159-183). Cambridge, MA: Ballinger.