Organizational ambidexterity is an increasingly researched topic (e.g., Kang & Snell, 2009; Simsek, Heavey, Veiga, & Souder, 2009) in the organizational sciences because of its evident contributions to innovation outcomes and organizational performance (Raisch, Birkinshaw, Probst, & Tushman, 2009). This literature has long since recognized the criticality of reconciling the conflicting requirements and contradictory demands posed by two forms of learning, viz., exploration and exploitation, and the tensions involved for organizations trying to achieve this balance (March, 1991, 1996). This literature has identified mainly two different types of ambidexterity, viz. structural and contextual (Raisch & Birkinshaw, 2008). Structural ambidexterity is based on an organizational design perspective and has recognized, identified, and expanded on a range of balanced organizational structures required to manage such ambidexterity (e.g., Jansen, Tempelaar, Van Den Bosch, & Volberda, 2009). Contextual ambidexterity, on the other hand, is based on the perspective of shaping the behavioral context in organizations, to enable individuals to engage simultaneously in both exploration and exploitation activities. Although the literature has also suggested two other types of ambidexterity: a) the outsourcing of explorative or exploitative activities (e.g., Baden-Fuller & Volberda, 2007), and b) the temporal cycling through exploration and exploitation (e.g., Raisch & Birkinshaw, 2008), we focus this study on the two dominant types, viz., structural and contextual ambidexterity.

Despite the importance of Human Resource Management (HRM) to managing the behavioral and organizational contexts and structures within which people enable ambidextrous functioning, the HRM perspective on ambidexterity has been relatively non-existent (see Dhifallah, Chanal, & Defelix, 2008 for an exception). The literature has generally described organizational structures, and behavioral contexts as promoters of ambidexterity (e.g., Raisch & Birkinshaw, 2008), both of which are dependent on the quality of the HRM system (e.g., Dhifallah et al., 2008; Kang & Snell, 2009) for effective deployment and execution. Commonly used conceptions of ambidexterity have emphasized the importance of entire HRM systems and practices such as socialization and team-building for fostering shared values (e.g., Andriopoulos & Lewis, 2009); recruitment and selection practices (Bartlett & Ghoshal, 1989); training and career development (Adler et al., 1999); job enrichment schemes, performance management, and rewards system (Gibson & Birkinshaw, 2004). Yet, it is very surprising that very little research has investigated which HRM practices contribute to the building of ambidexterity at the organizational level, especially organizations consisting of multiple businesses. Our unit of analysis is the organizational level, which is inherently multi-level because of multiple business units and/or multiple country locations.

In this paper, we address this issue by providing an empirical investigation of how MNCs manage their human resources at headquarters and across globally spread subsidiaries to build organizational ambidexterity. Specifically we investigate the nature of HRM practices, in specific configurations, that are best suited for ambidexterity in a global organization. This broad objective implies the following research question: What HR practice configurations within broader Intellectual Capital (IC) configurations are best suited for an organization that simultaneously uses different types of ambidexterity? We consider HR practice configurations embedded within intellectual capital configurations because of the compelling logic and broad literature consensus that HRM practices contribute to organizational outcomes via the mediating construct of intellectual capital (Wright, Dunford, & Snell, 2001), especially in contexts involving learning and knowledge management, such as those involving organizational ambidexterity.

As we demonstrate later, firms using contextual ambidexterity have a choice between a pair of intellectual capital configurations (Kang & Snell, 2009), with specific sets of HRM practices aiding each configuration. Firms using both structural and contextual ambidexterity, on the other hand, need to use both of the polar types of intellectual configuration types. This is because structural separation of exploration and exploitation requires different kinds of intellectual capital in these separated units. Thus, not examining intellectual capital configurations in these contexts will make it difficult to identify the relationship between specific sets of HRM practices and ambidexterity. We provide more details on these below.

To examine the above question, we identify and build on previous attempts in the Strategic Human Resource Management (SHRM) literature (e.g., Dhifallah et al., 2008; Kang and Snell, 2009). Despite their valuable contributions, work done by Dhifallah and others (2008), and Kang and Snell (2009) are limited in their scope and our collective understanding as of yet, in this context, is severely limited. First, each of these studies addresses the identification of an HRM architecture for one type of ambidexterity, thereby excluding organizations that may be using both types of ambidexterity simultaneously (e.g., Raisch & Birkinshaw, 2008). It is to be noted here that contextual ambidexterity is normally conceptualized and is applicable at the business unit level (Gibson & Birkinshaw, 2004), but not applicable at the organizational level for multi-business corporations, such as our focal case organization. Second, neither of these studies addresses the challenges of managing ambidexterity in an organization with a global spread of businesses, functions, and operations, which may necessitate the use of a combination of structural and contextual approaches to ambidexterity.

Next, although Dhifallah and colleagues (2008) comparatively examined the nature of HR practices in two R&D units, each of these two organizational units were limited to either structural- or contextual ambidexterity, respectively. This work (Dhifallah et al., 2008) is therefore limited by its narrow and exclusive focus on R&D units as opposed to entire organizations. A complete understanding of ambidexterity requires a comprehensive examination of other functional areas such as marketing, and manufacturing and should encompass the entire business unit (Gibson & Birkinshaw, 2004). Additionally, Dhifallah and colleagues (2008) expected to find different HR practice configurations in R&D units using structural ambidexterity from those using contextual ambidexterity. In contrast to their expectations, however, they found that R&D organizations used ‘mixed HR configurations’ in pursuit of ambidexterity, regardless of the type of ambidexterity favored in the larger organization where they were housed. An examination of the associated intellectual capital could have resolved the ‘surprising’ finding by revealing the possibility of different HR practices contributing to the same outcome, viz., equifinality (Wright et al., 2001). Thus, an examination of HRM practices embedded in intellectual capital configurations is required to obtain a good understanding of the HRM practices contributing to ambidexterity.

Overall, we know very little about what configurations of HRM practices are best suited for an organization that uses a) structural ambidexterity, or b) simultaneously uses structural and contextual ambidexterity, in a global setting. We contribute to the literature by studying a French MNC, which simultaneously uses structural and contextual approaches to ambidexterity. Thus, in contrast to the HRM architectures for individual types of ambidexterity (e.g., Dhifallah et al., 2008; Kang & Snell, 2009), we contribute to the literature by providing an empirical understanding of HRM practices at a large MNC encompassing structural and contextual ambidexterity. We provide one of the first comprehensive qualitative empirical investigations of HRM practices in the ambidexterity literature, involving a French MNC with global (180 countries) operations.

We first provide a brief overview of the relevant literature. Next, we provide a brief review of the sparse SHRM literature on ambidexterity issues and identify the separate and combined roles of human-, social-, and organizational- capital in managing the tensions between explorative and exploitative learning. We then turn our attention to the design and methodology we use for the study. We then present the findings of our case study. Our assessment of this MNC reveals a processual view on ambidexterity, characterized by initial structural separation of exploration and exploitation activities, followed by a progressive integration. We present the findings of our investigation of HR practices within this context of processual ambidexterity consisting of both structural and contextual approaches. We characterize the configuration of HRM practices we identified at this firm and place it in the above context. Finally, we discuss contributions, limitations, and directions for future research in this area.

Following early work on organizational learning by March (1991) and his identification of exploration and exploitation learning mechanisms, the literature has defined ambidexterity as the balancing of these two distinctive and sometimes conflicting processes of learning. This literature associates exploratory learning with activities such as search, variation, experimentation, and discovery (quest of new knowledge), while exploitation is associated with activities such as refinement, efficiency, selection, and implementation (deepening existing knowledge; March, 1991: Raisch & Birkinshaw, 2008). The literature describes incremental innovations (minor adaptations) as exploitative and radical innovations (fundamental changes) as explorative (e.g., Simon & Tellier, 2008; Smith & Tushman, 2005). We focus on two different means identified in the literature to simultaneously manage and balance these two learning processes constituting ambidexterity (Raisch & Birkinshaw, 2008).

The first is structural ambidexterity where these two learning activities are separated into distinct, specialized organizational units. Exploration-oriented units are typically characterized by an organic design, incentives for learning and exploration, and an error-embracing management style. Exploitation-oriented units are characterized by a mechanistic design, incentives for efficiency and quality, and an error-avoiding management approach (O’Reilly &Tushman, 2008). The integration between these different units is achieved by managers who have the ability to reconcile different and opposing threads into a coherent whole. The literature suggests that complex organizational designs reconciling mechanistic and organic features are necessary for resolving the paradox involved in balancing exploration and exploitation (e.g., Adler et al., 1999; Jansen et al., 2009).

Contextual ambidexterity, on the other hand, refers to the situation where exploration and exploitation need not be nested in different units, but on the contrary can be carried out in a single business unit (Gibson & Birkinshaw, 2004). This is possible only if management sets an organizational context that encourages individuals to divide their working time between exploration and exploitation. This organizational context is underlain by four attributes, support, discipline, stretch and trust. The combination of these four attributes leads to a contextual balance between exploration and exploitation (Gibson & Birkinshaw, 2004). Comprehensive reviews (e.g., Raisch & Birkinshaw, 2008) have identified two broad gaps for future research to address: a) The simultaneous presence and interaction among structural and contextual approaches to ambidexterity (the extant approach considers them as mutually exclusive), and b) The impact of international context on ambidexterity, involving multiple national business systems and cultures (the extant approach has been ‘context-free’). Examining ambidexterity in a global organization, as we do here, provides a comprehensive understanding of it in manner different from the literature’s (e.g., Kang & Snell, 1999) conclusions. We address these gaps and contribute by examining the intellectual capital and HR practices in a firm where the two hitherto mutually exclusive approaches (viz., structural and contextual) coexist with each other.

The need and rationale to examine a firm’s intellectual capital and its HR practices for ambidexterity objectives is as follows. First, ambidexterity is achieved by resolving the paradox resulting from the simultaneous management of exploration and exploitation, which necessitate fundamentally different organizational structures, strategies and contexts (e.g., Mothe & Brion, 2008; Raisch & Birkinshaw, 2008). The literature identifies a strong connection between a firm’s orientation to learning and its knowledge stocks (e.g., Cohen & Levinthal, 1999). Thus, a real understanding of the ‘trade-off’ between exploration and exploitation necessitates a direct examination of the organization’s knowledge stocks and therefore its intellectual capital (Youndt et al., 2004), which represents the sum of all its knowledge stocks (e.g., Subramaniam & Youndt, 2005). Moreover, learning theorists argue that organizations do not create knowledge in and of themselves but only do so by virtue of the people (or human capital) in them (e.g., Argyris & Schon, 1978; March, 1991). Conversely however, individual learning is a necessary but insufficient condition for organizational learning (Argyris & Schon, 1978), which requires individuals to exchange and diffuse shared insights, knowledge, and mental models (Simon & Tellier, 2008) through the organization’s social capital (Youndt et al., 2004). Ultimately, the knowledge created by human capital, and diffused by its social capital, becomes codified and institutionalized in the organization, thereby turning into organizational capital (e.g., Kang & Snell, 2009). Next, researchers identify HRM as one of the more crucial investments in an organization’s efforts at building its intellectual capital, among others including Information Technology (IT) and R&D (e.g., Youndt et al., 2004). Recent work implicitly notes the importance of human capital (e.g., Dhifallah et al., 2008) and social capital (Simon & Tellier, 2008) within IT and R&D units for achieving broad organizational ambidexterity objectives, without directly examining them. Since IT and R&D functions are also dependent on people and human capital, the HRM function and its practices are among the most important in an organization for creating the context for organizational ambidexterity (Wright et al., 2001). If we are to believe that HRM practices lead to the creation of appropriate knowledge stocks in the form of intellectual capital, as the evolving consensus (Wright et al., 2001) suggests, it becomes imperative to explicitly examine HRM practices embedded within intellectual capital configurations.

The SHRM literature has only recently begun to investigate the impact of HRM practices on organizational ambidexterity (see Dhifallah et al., 2008; Kang and Snell, 2009; Simon & Tellier, 2008). While some in this literature (e.g., Dhifallah et al., 2008; Simon & Tellier, 2008) have not examined intellectual capital, others (Kang & Snell, 2009) have not addressed global context issues. Moreover, all of these studies only focus on one type of ambidexterity at a time and therefore ignore the simultaneous presence of two types of ambidexterity. Thus, our collective understanding of ambidexterity is severely limited. However, as these authors suggest, this literature contains the critical building blocks that can be assembled together to create the foundation for future studies. This body of work identifies the importance of HRM practices for developing a firm’s human capital (e.g., Lepak & Snell, 1999), social capital (e.g., Kang & Snell, 2009), and organizational capital (e.g., Wright et al., 2001), which are constituents of the overall intellectual capital (e.g., Youndt et al., 2004). Further work identifies the importance of human- and social capital for radical innovation (exploitative) capabilities, and the importance of organizational capital for incremental innovation (explorative) capabilities (Subramaniam & Youndt, 2005). Thus, intellectual capital and various profiles of it are critical for ambidextrous learning (Subramaniam & Youndt, 2005; Youndt et al., 2004). Organizations possess unique intellectual capital profiles by virtue of the different levels of human-, social-, and organizational-capital constituents. Working within the confines of a configurational approach (e.g., Huselid, 1995), and invoking the crucial concept of equifinality (Wright et al., 2001), Kang and Snell (2009) develop a pair of theoretically possible intellectual capital architectures, each combining unique types of human-, social-, and organizational capital that would be most instrumental for managing contextual ambidexterity. Additionally, Kang and Snell (2009) also identify configurations of HR practices that help the firm to manage these intellectual capital architectures to foster ambidextrous learning.

Human-, Social-, and Organizational Capital: Intellectual capital consists of three separate sub-dimensions viz., human capital (i.e., knowledge, skills, and experience of employees), social capital (knowledge residing in groups and networks), and organizational capital (i.e., institutionalized knowledge residing in systems, databases, routines etc.) (Youndt et al., 2004). Empirical research notes that high investment in HRM is associated with an intellectual capital profile with high levels of all three constituents (Youndt et al., 2004). Subramaniam and Youndt (2005) find that social capital is related to both types of innovation, as a result of resolving the paradox between the two different types of learning mechanisms described in the literature. Although Simon and Tellier (2008) do not examine social capital in their study of R&D projects, they find that the types of social networks (proxy for social capital here) are different in projects focusing on exploration versus exploitation, but are critical for both types. Specifically, they find that while a connection between dense, homogenous networks, composed of strong ties exists in exploitation projects, the connection between sparse, heterogeneous networks, composed of weak ties exists in exploration projects (Simon & Tellier, 2008). This body of work generally suggests the critical role of HRM practices and HRM investments in building intellectual capital profiles contributing to managing ambidexterity.

Intellectual Capital Architecture: Focusing exclusively on contextual ambidexterity, Kang and Snell (2009) identify two distinct architectures, viz., refined interpolation and disciplined extrapolation as alternative mechanisms to manage ambidexterity. Refined interpolation consists of a specialist human capital, cooperative social capital, and an organic organizational capital (Kang & Snell, 2009; see Table 1). Specialist HC involves highly specialized and in-depth knowledge of a narrow domain and is thus better suited for exploitative learning mechanisms. Cooperative SC is characterized by dense network connections and institutional trust mechanisms and is inherently suited for exploitative learning mechanisms (see also Simon & Tellier, 2008). To balance the above two exploitation oriented components, Kang and Snell (2009) argue for an organic OC characterized by decentralized structures and higher levels of autonomy in groups and networks that foster more explorative learning mechanisms than the other type (viz., mechanistic OC).

The disciplined extrapolation architecture, on the other hand, consists of a generalist human capital, entrepreneurial social capital, and a mechanistic organizational capital (Kang & Snell, 2009). Generalist HC involves knowledge from across domains, inherently more suitable for explorative learning. Entrepreneurial SC consists of non-generalized, dyadic trust between individuals and loosely connected, weak and non-redundant networks generally favoring exploratory learning (see Simon & Tellier, 2008). Again, to balance the above two components’ natural inclinations towards exploration, Kang and Snell (2009) argue for mechanistic OC characterized by rigid rules, procedures, and highly formalized structures for enhancing exploitative learning.

In addition to these two types of intellectual capital architectures, Kang and Snell (2009) identify configurations of HR practices that aid these two architectures. To support refined interpolation, Kang and Snell (2009) theorize and utilize an HR configuration consisting of job or function-based development for specialist HC, internal labor market (ILM) based employee relations system for cooperative SC, and error embracing performance/control systems for organic OC. Job or function-based development is characterized by narrow job design, focused career development, and a person-job fit approach to recruitment. Internal labor market based employee relations system is characterized by internal staffing/promotion, seniority based compensation, and socialization practices (Osterman, 1984). Error embracing performance/control systems acknowledge mistakes as a natural by-product of learning, allow individuals to make decisions, to set their own performance goals, and to make changes in the way they perform their jobs (i.e., empowerment).

To support disciplined extrapolation on the other hand, these authors (Kang & Snell, 2009) argue for the use of a configuration consisting of skill-based HC development for generalists, market-based employee relations system for entrepreneurial SC, and error-avoiding performance/control system for mechanistic OC. Skill-based development is characterized by broad (multi-dimensional) job design, job rotation, and a person-organization fit approach to recruitment (Lepak & Snell, 1999). A market-based employee relations system is characterized by extensive external recruitment, performance-based compensation, and general training and development. Error avoiding performance/control systems uphold specific provisions regarding work protocols, use behavior (versus result)-based evaluation and rewards, and performance programs imposed top-down (Lepak & Snell, 1999). This work builds on the foundations built by earlier scholars and provides a theoretical framework that enables subsequent work in this area.

In summary, the only available theoretical framework in HRM (Kang & Snell, 2009) focuses exclusively on contextual ambidexterity. The rare empirical studies focus on specific organizational units (e.g., R&D) or projects within them, while focusing on one or the other type of ambidexterity (e.g., Dhifallah et al., 2008; Simon & Tellier, 2008). Thus, we do not know what kinds of HR practices, within broader intellectual capital architectures are suitable for simultaneous use of two types of ambidexterity in a global setting. We contribute to the literature by addressing this issue.

Research setting. In the tradition of studies designed to generate theoretical insights, we used a single revelatory case design (Yin, 1994) to examine HR practices facilitating ambidexterity. To develop a refined theoretical view, we conducted a case study at a MNC (listed on the Paris Stock Exchange) operating in the electrical appliances sector having global leadership positions. The company holds strong leadership positions in Europe due to the rapid development based on a dynamic policy of strategic acquisitions. For more than twenty years now, the MNC has maintained the number-one place for wiring devices and cable management worldwide, as well as number-one rankings in many key national markets.

In 2009, annual turnover of the company employing 30,000 people worldwide was around € 4 billion and its net income was almost € 300 million. The company holds from 14% to 20% of the world market share in certain products and services. It has manufacturing and/or distribution subsidiaries and offices in more than 70 countries, and sells its products in about 180 countries. Almost 60% of the business revenues are drawn from Europe, another 15% from the American continent, and the rest from all over the world. Its key markets are France, Italy and the United States, however new/emerging economies currently account for one third of its revenue.

The company is organized around three industrial divisions in charge of R&D and production and local front offices in countries, responsible for sales. The company offers integrated solutions and a complete range of specific services customized to different types of clients: residential apartments and industrial constructions, commercial buildings and hospitals, hotels and schools, etc. Sensing a need to speed up the new product development process the company, in recent years, has launched an ‘innovation challenge’ and reduced projects cycle time from four to two years. The catalogue of the MNC contains more than 170,000 offerings in the 80 categories of products.

We chose this particular company for its demonstration of consistent success at exploitative and explorative (usual proxies for ambidexterity) innovation respectively (e.g., Andriopoulos & Lewis, 2009). This company is judged by observers as being excellent, both at conducting routinized operations, and at innovating. For more than 30 years now, innovation, espoused as one of the company’s four major values, has been guiding the company’s development. In 2010, 38% of sales turnover was generated from new products; and almost 60% of the company’s investments were allocated to new products. A research and development team of almost 2000 engineers is one of the strengths of the company. The MNC possesses more than 4000 patents and spends close to 4.5% of the sales turnover on research and development. An acceleration of new product launches and a continuous search of opportunities for external growth are the two major characteristics of this company. Due to the firm’s multinational nature, balancing exploration and exploitation takes place in the context of the presence of multiple national business systems. Finally, the authors have had a longstanding relationship with this MNC, thus facilitating access to various parts of the organization as well as open conversations with informants.

Nature of Ambidexterity at this firm. At this company, the innovation process requires a constant balance between exploration and exploitation. Therefore, ambidexterity takes place at multiple levels and locations. At each level and location, specific actors or groups of actors make decisions and take action to maintain the balance between exploration and exploitation. In this section, we present the different activities in a sequential manner, but remind readers that exploration and exploitation are managed simultaneously and coexist at any point in time throughout the organization. Additionally, although we trace structural and contextual ambidexterity locations in this sequential description, we remind readers that they also coexist at any point in time throughout the organization.

First, innovative ideas originate everywhere in the organization, within subsidiaries as well as at the Headquarters. At the beginning, these ideas are generated and voiced by individuals (facilitated by suitable organizational contexts), more often than not from operational levels. In their everyday work, individuals simultaneously carry out their routine tasks and look for new ideas. The dominant mode at this stage is thus contextual ambidexterity (Gibson and Birkinshaw, 2004). The best ideas are then examined for developing potential projects. The project portfolio is made of innovative, new-to-the-firm (exploration) projects and more incremental developments of existing lines of products (exploitation). The balance among these different kind of projects is made at the tactical level by middle managers located at the HQ. Because of the structural separation of exploration and exploitation by project structure, the dominant mode in this phase is structural ambidexterity (O’Reilly and Tushman, 2008). Once the detailed study and prototyping of projects is completed, the top management team then decides on the definitive portfolio mix for the mid-term. At this stage, the dominant mode is contextual ambidexterity. Once the projects are completed, the new products are progressively transferred to production. The knowledge transformation implied is made by specific teams that work in close coordination with R&D teams while remaining structurally separate. The dominant mode here is structural ambidexterity. Heads of each country are free to adopt a new product or not and are in charge of deciding the appropriate mix of new and old products in its portfolio. Lastly, if a head of a country decides to produce a new product, local production and sales contexts must be adapted by local employees who then manage old and new products in their daily activity in the contextual ambidexterity mode. Broadly, therefore, the two forms of ambidexterity are present at different stages of the innovation cycle at this MNC. Instead of being purely structural or contextual in form, the ambidexterity at this firm encompasses a process whereby structural ambidexterity at some levels is complemented by contextual ambidexterity at other levels.

Data Sources and Collection. We, as an institution, have had a long standing relationship (since the late 1990s) with this MNC encompassing collaborative work on case study development, development of modules on innovation for the MNC, several collective research projects, regular involvement of MNC managers in our courses, supervised student internships at this company, and social interactions. By virtue of this long relationship, we had access to and utilized a variety of archival data sources such as company videos, internal brochures, newsletters, intranet, annual reports of the company, newspaper articles, published case studies (joint copyright), and previous research project documents on knowledge management processes, in addition to specifically focused interviews (between 2010 and 2011) for this particular study. We used an interview schedule specifically developed for this study, focused directly on the HRM practices associated with the exploration and exploitation processes of organizational learning. Specifically, we asked the interviewees about the company’s overall approach to exploration and exploitation respectively, their approach to managing them simultaneously, the interviewee’s role in managing these two learning processes, how these processes unfolded, who or what they saw as major contributors or critical elements of this process, HRM’s role in the process, the role of organizational structure/design in this process, and their overall evaluation of the success of these efforts. We triangulated data from the interviews with our database of different archival sources (to ensure credibility and reliability) in seeking converging evidence and inductively theorizing the intellectual capital architecture and HRM practices we develop in this study. Table 2 below summarizes functions and positions of our respondents.

We used the notion of purposive sampling (Lincoln & Guba, 1985) and selected interviewees according to two main criteria. First, we ensured that all the main functional areas where organizational learning and knowledge management were likely to play a key role were covered. Second, we selected respondents from different hierarchical levels to have an overall view of functional areas, as well as of the interactions and interdependencies with other functions. Incidentally, all of our interviewees have held different positions at headquarters and in several subsidiaries at different points in time. Utilizing these two criteria allowed us to have a global vision of the way this firm handles exploration and exploitation at the organizational level. This variety of viewpoints gave us a global overview of the firm’s functioning. These interviews helped us, both to deepen the emergent themes of exploration/exploitation activities, and to understand the managerial sensemaking efforts in critical events. We continued the data collection process until we stopped learning anything new from the additional informants and sources (Lincoln & Guba 1985).

Interviews typically lasted 110 minutes, although a few ran as long as three hours. We took detailed notes regarding our observations and interpretations, which we included in our analysis. In addition, we received permission from all respondents to audio-record their interviews, which were then professionally transcribed and validated. This corpus, presented in table 3 below, combined with our database of eighty archival documents, provided us with extensive information on the events, the innovation initiatives, and the balancing of exploration and exploitation activities.

Data Analysis. We present a description of the coding procedures, following construct definitions, in tables 3a and 3b. We used existing theoretical categories shown in these tables for all our coding, following theory generation principles identified by Eisenhardt (1989), with the aim of discovering new links between them, not established in previous literature. For all the coding in this study, each of the three authors independently coded all of the relevant data sources. Differences were then discussed and reconciled in all instances. In the end, we had complete agreement for all the instances and each of the theoretical constructs used in the analysis. We analyzed the interviews for themes related to the use of human-, social-, and organization capital for explorative and exploitative learning respectively, using the guide shown in table 4b. We also analyzed the interviews for use of different HR practices for these two modes of learning respectively. In this step of the coding the relevant match of the intellectual capital constituents with the corresponding learning type emerged from the data, consistent with our qualitative/inductive approach to theory building. Overall, the process of data collection, data representation and data verification were interrelated in our study. Our theory building in this study is recursive in that it cycles among the case data, the emerging theory, and extant theory from the literature.

Our findings provide new insights to the extant knowledge on organizational ambidexterity, specifically as it pertains to the most appropriate HR practice configuration for a firm that uses structural and contextual approaches. We sought to identify the intellectual capital architecture and the associated HR configuration of practices that best support ambidexterity (of multiple types) initiatives in one MNC which is a world leader in its sector. First, our qualitative investigation at this French MNC confirms theoretical expectations (Raisch & Birkinshaw, 2008) that the two approaches to ambidexterity are not mutually exclusive and these co-exist within the same firm. Instead of being purely structural or contextual in form, the ambidexterity at this firm encompasses a process whereby structural ambidexterity at some stage of the overall process is complemented by contextual ambidexterity at other stages. Therefore, our findings about the nature of the intellectual capital profile and HR practices at this form are in contrast to the theoretical work by Kang and Snell (2009) which was limited to one type, viz., contextual ambidexterity. Our findings vis-à-vis social capital, in particular, show convergence with the findings of Simon and Tellier (2008) in the domain of social networks corresponding to exploration and exploitation. Additionally, our findings about the nature of HR practices are different from the work of Dhifallah and colleagues (2008) which was limited to structural ambidexterity in one case and contextual ambidexterity in another.

Our findings contribute to the literature on ambidexterity by providing empirical confirmation of the strong connection between intellectual capital configurations and the management of knowledge stocks and knowledge flows. Within each of the three components of intellectual capital, where one approach (e.g., generalist HC, cooperative SC, Organic OC) leads to more effectiveness in the building of new knowledge stocks (i.e., exploration), the opposite approach (e.g., specialist HC, cooperative SC, mechanistic OC) leads to better results for managing optimal utilization of existing knowledge stocks (i.e., exploitation) by efficiently managing knowledge flows. Thus, overall, each of the polar types for each of the three components of intellectual capital are important for building ambidexterity. Thus, our findings reveal that resolving the paradoxical requirements of exploration and exploitation not only require different strategies, structures, and contexts, but it also requires different intellectual capital configurations.

Proposition 1: Organizational ambidexterity constituting structural- and contextual-ambidexterity requires an intellectual capital configuration consisting of both polar types of each component (viz., generalist- and specialist- HC; entrepreneurial- and cooperative- SC; mechanistic- and organic- OC).

Importantly, our empirical findings reveal more complexity in these practices in comparison to the simplistic theoretical framework available in the literature (Kang & Snell, 2009). Where Kang and Snell (2009) theoretically identify one polar type for each component of intellectual capital, our empirical findings suggest that both polar types within each component are important in an organization that uses two approaches to ambidexterity. First in terms of human capital, we find that it is important to have generalists for exploration and specialists for exploitation. This seems to be especially critical for globally spread organizations that use a mixture of structural and contextual ambidexterity. We suggest that both specialist and generalist HC types may be necessary in such a context. For instance, if the firm used only contextual ambidexterity, it could use either generalists or specialists as part of their intellectual capital configuration (Kang & Snell, 2009). This use of either generalists or specialists in the human capital category can be counterbalanced by appropriately choosing social capital and organizational capital components (Kang & Snell, 2009), as noted earlier. Firms that also use structural ambidexterity, however, do not have this option of choosing between specialists and generalists because structurally separated exploration units would require generalists and corresponding exploitation units would require specialists. We confirm this empirically through identification of the simultaneous presence of job descriptions and hiring practices associated with each at this firm, thereby also linking HR practices to intellectual capital configurations.

Proposition 1a: Organizational ambidexterity constituting structural- and contextual-ambidexterity requires a human capital configuration consisting of both polar types (viz., generalist- and specialist- HC).

Next, we find that a mixture of entrepreneurial SC and cooperative SC best serves the balancing of exploration and exploitation learning processes respectively. This is in contrast to the work of Kang and Snell (2009) for reasons that are similar to those provided above for human capital. However, we see a fair degree of convergence in our findings with the work of Simon and Tellier (2008) who focused on the nature of social networks in R&D projects dedicated to either exploration or exploitation. The connection between dense, homogenous networks, composed of strong ties to exploitation projects in Simon and Tellier’s (2008) work converges with our findings of a cooperative SC (a la Kang & Snell, 2009) for such projects in the entire organization (not just R&D). Similarly, the connection between sparse, heterogeneous networks, composed of weak ties with exploration projects (Simon & Tellier, 2008) converges with our finding of entrepreneurial SC (a la Kang & Snell, 2009) for the entire organization. Additionally, early findings in the literature pertaining to personnel transfers and job rotation (see Dhifallah et al., 2008; O’Reilly & Tushman, 2008) for managing the tensions between exploration and exploitation find support within our focus on social capital in this study. Specifically, we show that internal mobility within and across countries, use of expatriates and inpatriates, and job rotation across countries and divisions, in our study, serves to link dense networks in subsidiaries with sparse networks in other locations.

Social capital is perhaps the most affected component of the overall IC configuration, as a direct function of increase in global dispersion. One common problem with the nascent literature on ambidexterity is that it tends to be context-free (see Raisch & Birkinshaw, 2008) and therefore ignores MNCs and their global and cultural spread. This characterization is true of Kang and Snell’s (2009) work as well in that they do not necessarily address the global dimension. As SC primarily pertains to social exchange mechanisms, the nature of trust in these exchanges, and the degree of autonomy different parties to the exchange may enjoy, this is the dimension that is most likely to be affected by global dispersion. Here again, our findings suggest that both entrepreneurial and cooperative types would best serve the needs of ambidexterity objectives in such contexts. Thus, our findings reveal more complexity on this dimension as well.

Proposition 1b: Organizational ambidexterity constituting structural- and contextual-ambidexterity requires a social capital configuration consisting of both polar types (viz., entrepreneurial- and cooperative- SC).

Finally, we find that a mixture of organic and mechanistic OC is best suited for balancing exploration and exploitation. Simultaneously managing exploration and exploitation requires a balance of the above two types of OC. Much of the balance between centralization and decentralization comes primarily in the form of the matrix structure, which serves to thicken the communication network in a global organization, thereby easing the flow of knowledge within and across units.

Proposition 1c: Organizational ambidexterity constituting structural- and contextual-ambidexterity requires an organizational capital configuration consisting of both polar types (viz., mechanistic- and organic- OC).

In terms of the required overall IC configuration, our findings indicate that a mixture of ideal types (e.g., entrepreneurial and cooperative for social capital) for each of the constituent dimensions are perhaps what is best suited for building ambidexterity in a global organization, that uses both contextual and structural types.

Our findings in terms of underlying HR practice configurations also reveal marked differences from the theoretical work on contextual ambidexterity (Kang & Snell, 2009). However, our findings are consistent with the findings of Dhifallah and colleagues (2008) which suggest there is no one best way to manage R&D scientists and that a hybrid approach to HRM is effective regardless of the approach to ambidexterity. Our findings essentially identify a hybrid approach to practically every component of an overall HRM system of practices (see table 5), across all functions of a global organization. This is essentially because one type of HR practice is best suited for exploration and another for exploitation. Since all ambidextrous organizations balance between these two learning processes, they require a hybrid set of HRM practices consisting of polar opposites in terms of the individual components of the overall HRM system of practices.

Consistent with the work of McClean and Collins (2011) which argued that high performance work systems may not be required for entire organizations but would rather add more value when selectively applied, we find that these high performance practices are best suited for exploratory learning processes and that administrative HR systems are best suited for exploitative learning. Our findings suggest that a) both job- or function-based development practices and skill-based development practices; b) a mixture of market-based and internal labor market-based employee relations systems; and c) a mixture of error-embracing and error-avoiding performance management systems, are critical at this firm for facilitating a combination of structural and contextual ambidexterity. Thus, not only are hybrid practices important in managing human resources in R&D units (Dhifallah et al., 2008) but also entire organizations. We add to the literature with a detailed set of hybrid practices, in this study.

Proposition 2: Organizational ambidexterity constituting structural- and contextual-ambidexterity requires a HRM practice configuration consisting of both polar types (viz., high performance work systems and administrative HR systems).

Our study is limited in that this is one unique observation, in an industrial sector, which can be characterized as traditional, at best. However, our study is an in-depth and qualitative investigation, and we believe that these revelatory case findings are rich and likely significant for further theory development in this regard. Most importantly, we find the need for broader theory development for developing HRM practice configurations suitable for broad and comprehensive ambidexterity approaches of relatively large MNCs. Additionally, broader consensus is needed in the operational conceptualizations of intellectual capital and its dimensions. Alternative conceptualizations of human, social, and organizational capital as either continuous dimensions (varying from low to high) or as types in an overall typology, have been used in the nascent literature to date, making it complex to measure and analyze (see Youndt et al., 2004). Whereas Youndt and colleagues (2004) use overall measures of the three components of IC, varying from low to high, and then developing profiles based on cluster analysis, Kang and Snell’s (2009) theoretical formulation for ambidexterity uses a typological approach to each of these dimensions.

Finally, our findings on HR practices, also argue for further investigation of universal versus contingent approaches (e.g., Youndt, Snell, Dean, & Lepak, 1996) to ambidexterity. First, consistent with both old (e.g., Youndt et al., 1996) and new evidence (e.g., McClean & Collins 2011) we find evidence of the simultaneous presence of high-performance work systems and administrative HR systems within the same organization but geared to different objectives (exploration vs. exploitation; and structural vs. contextual ambidexterity, in this case). Simultaneously we find the existence and importance of each pair of typological types of the individual components of intellectual capital for ambidexterity that is both structural and contextual. These suggest to us that contingent approaches to ambidexterity and therefore each of the constituent learning processes may be more effective than universal approaches as has been hitherto suggested (e.g., Kang & Snell, 2009).

We make several contributions to the SHRM literature on ambidexterity through this study. We provide one of the first empirical examinations directly focused on ambidexterity, in the SHRM literature. We identify the need for a broader range of HR practices, ranging from stability-oriented to flexibility-oriented, as a very important requirement for building ambidexterity. This work also contributes to the body of literature that examines the interplay of people (human capital) and structures (social and organizational capital) in deriving ambidexterity benefits. Further, we also contribute by examining the role and interplay of HR systems of practices with the different components of intellectual capital, thereby highlighting the need for further examination of this relationship between these constructs for ambidexterity. In the process, we note the theoretical refinements that are necessary for the advancement of efforts in this crucial area of research and practice. Finally, therefore, we hope that future research will heed our calls for the specific theoretical refinements outlined here.