Effects of functional focus on bounded momentum: Examining firm- and industry-level alliances

Abstract

Building on strategic momentum and alliance studies, we theoretically explain and test the non-monotonicity of the alliance momentum of same- or cross-functional-type alliances. We theorize about critical drivers that generate bounded momentum and further argue whether function-specific momentum occurs sequentially or simultaneously. We examine cross-border alliances of 32 international airlines from 1945 to 1994 and find that the inverted U-shaped pattern of alliance momentum holds for same-type but not for cross-type alliances at both the firm and industry levels. These findings imply that alliance momentum with a specific functional focus evolves sequentially rather than simultaneously.

Keywords

airline industry cross-border alliance formation functional types strategic momentum

Introduction

Most firms endeavor to manage alliance strategies to balance exploration and exploitation in various function domains (Koza and Lewin, 1998; Lavie and Rosenkopf, 2006; March, 1991; Rothaermel and Deeds, 2004). However, previous alliance studies have not paid sufficient attention to the critical question of whether firms take sequential or simultaneous steps in alliance formation within different function domains. That is, does function-specific alliance momentum occur sequentially or simultaneously? Most corporate activities are organized and implemented based on the functional focus, which often defines effective learning boundaries. Functional subunits also drive managers to have “selective perception” (e.g. Ketokivi and Castaner, 2004) and to pay disproportionate attention to their task environment defined by the functional focus (e.g. Kefalas and Schoderbek, 1973; Waller et al., 1995). Indeed, alliance studies have traditionally associated exploration and exploitation with specific value chain functions, such as research and development (R&D), marketing, and production (e.g. Hennart and Reddy, 1997). However, a more refined theoretical and empirical investigation of how the functional foci of previous alliances affect the scope and shape of alliance momentum over time remains necessary.

Repetitive alliance behavior has been an important research topic in the alliance literature (e.g. Gulati, 1995a; Gulati and Gargiulo, 1999; Li and Rowley, 2002). However, previous studies of the relationship between functional foci and the scope of alliance momentum yielded somewhat conflicting results. Several studies have found that most alliance experience is stored within functional departments and/or teams, and it requires significant effort to integrate, share, or diffuse such knowledge across departments or teams (e.g. Doz and Hamel, 1998). Spillovers of alliance experience across functions within a firm, as well as between firms, are limited by multiple factors, such as differential learning and knowledge bases (Cohen and Bacdayan, 1994), the inherent reduction in information exchanged outside of functional areas (March and Olsen, 1989), and the challenges of distance learning and interpretation (Levinthal and March, 1993). These factors may limit the scope of capability development from previous alliances to specific functions. On the other hand, other studies implicitly assumed that alliance capabilities built up through experience are readily transferrable across departments and possibly across organizations, thus downplaying functional foci (Anand and Khanna, 2000; Heimeriks et al., 2007; Zollo and Reuer, 2010). To resolve these conflicting results, it is critical for scholars to understand whether firms establish multiple alliances in different function domains simultaneously or sequentially (i.e. simultaneously exploring multiple alliances in various function domains or sequentially exploiting multiple alliances in a single-function domain), rather than studying alliance momentum at an aggregated level.

Previous research regarding the relationship between functional boundaries and the shape of alliance momentum also contained some conflicting assumptions, results, and interpretations at both the firm and industry levels. First, some studies found that the shape of alliance momentum is monotonic (e.g. Kelly and Amburgey, 1991). However, other studies found that alliance momentum is usually bounded beyond some point (e.g. Gomes-Casseres, 1996). Second, previous studies utilized two distinctly different empirical approaches: Gulati (1995b) combined multiple types of alliances, implying that the specificity of alliance type would not independently affect momentum. Alternatively, Chung et al. (2000) focused on a single type of alliance. Third, functional boundaries are also critical when investigating the scope and shape of alliance momentum at the industry level. The alliance literature has found monotonic industry-level momentum, whereby alliances by other industry participants are positively associated with future alliance formation by a focal firm (Gomes-Casseres, 1996; Gulati, 1995b). However, the role of functional focus in determining the scope and shape of industry-level alliance momentum requires both theoretical development and empirical examination. Furthermore, especially in international contexts, these contrasting assumptions and divergent approaches suggest a need to revisit the relationship between functional focus and alliance momentum.

To answer the above question, we examine how the functional focus of previous alliances affects both the scope and shape of momentum for the same and/or other functional types of alliances at both the firm and industry levels. Relying on various drivers of momentum at both of these levels, we further refine the existing theories highlighting the importance of functional boundaries, both within firms and across industry firms. We integrate inter-firm cooperation and competition literature (e.g. Axelrod, 1984; Gomes-Casseres, 1996) to better explain the mixed motives affecting alliance formation and scope, and we further specify drivers of sequential and/or simultaneous alliance dynamics. We test this enhanced theory by analyzing the cross-border alliances of 32 international passenger airlines from 1945 to 1994.

Our theoretical and empirical approach may advance several research frontiers. First, theoretically focusing on different types of alliances at both the firm and industry levels affords a deeper understanding of how functional boundaries limit these specific drivers of momentum over time. If the benefits of repetitive momentum significantly depend on functional similarity, then we can expect sequential strategic momentum within the same functions rather than simultaneous evolution across different functions. Second, our study may be able to advance the exploration and exploitation literature by highlighting the role of time dimension. That is, if scholars understand alliance momentum as a sequential or a simultaneous phenomenon, it will yield different implications regarding the issue of balancing alliance portfolios or processing alliance portfolio. Third, incorporating functional foci can refine research designs for various strategic issues. For example, if alliance momentum is type-specific and functionally sequential, research designs pooling distinct alliance types may generate confounding errors. These refined findings can better explain the somewhat divergent or conflicting results of previous empirical investigations. Fourth, our study will refine the shape and scope of momentum of many strategic actions, such as innovation, strategic changes, and spinoffs, that may be very sensitive to functional differences within a firm. For example, the partial spinoff decision for production and marketing functions may evolve along different paths due to heterogeneous knowledge and functional characteristics.

Drivers of strategic momentum

Previous studies have investigated both the types and shapes of strategic momentum in various strategic actions. While Amburgey and Miner (1992) theorized about the different types of strategic momentum—such as repetitive, positional, and contextual momentum—repetitive momentum has been an important focus of many empirical studies. Other studies investigated the theoretical drivers generating repetitive momentum at both the firm and industry levels (e.g. Kelly and Amburgey, 1991; Turner et al., 2013). At the firm level, these studies identified various drivers of momentum, such as favorable or unfavorable outcomes of specific actions, organizational routines, competencies, and perceptual or cognitive constraints. For example, the favorable or even unfavorable performance of previous actions often increases the likelihood that firms will repeat the same actions. Advocates of a specific action may not interpret negative outcomes as evidence of a strategy being incorrect, but rather as a strategy not being implemented properly (Levitt and March, 1988). Organizational routines are standardized as durable solutions for typical problems (Cyert and March, 1963). As such, routines can relieve the burden of complex and time-consuming decision making in similar and potentially repeatable situations. Cognitive theorists emphasized how cognitive constraints or maps affect environmental perceptions and limit the perceived range of reasonable alternatives, thus encouraging the repetition of previous actions (Schwenk, 1984). Repetitive actions also refine relevant competencies, further increasing the likelihood of selecting those actions in the future (Nelson and Winter, 1982).

At the industry level, previous studies explained the diffusion of a specific action across industry firms as a result of isomorphism, vicarious or spillover learning, competitive bandwagon, or fads and fashion. Institutional theorists argued that once a threshold number of firms adopt an action, isomorphism drives most industry participants to engage in mimetic behavior to seek legitimacy (DiMaggio and Powell, 1983). Unlike firm-level learning-by-doing, industry participants may also learn by observing others, that is, by vicarious or spillover learning (Baum et al., 2000; Shaver et al., 1997). Vicarious learning helps firms better understand a specific action and, in turn, encourages subsequent imitation (Miner and Haunschild, 1995). Abrahamson and Rosenkopf (1993) highlighted the role of “competitive bandwagon effects”: As the number of firms that have adopted an innovation increases, the competitive pressure on the remaining firms to adopt the same innovation escalates. This fads and fashion perspective argues that most organizations imitate to conform to emergent norms that sanction a specific action through social interactions in a community (Abrahamson, 1991). These studies suggest that such mimetic behavior may also be driven by economic concerns about the possible competitive disadvantage of late adoption.

Previous momentum research further refined the monotonic shape of repetitive momentum by showing that past actions encourage a future action up to a certain point, but discourage it beyond that point (Chung et al., 2000; Gulati, 1995b). We refer to this inverted U-shaped momentum as bounded momentum. Previous studies showed that the bounded momentum of strategic actions often occurs at the firm level. When repeating the same strategic actions, it may be increasingly difficult for firms to avoid momentum-decelerating factors, such as resource constraints, diseconomies of management (Shaver and Mezias, 2009), loss of strategic flexibility (Gomes-Casseres, 1996), and possible sub-additivity (Vassolo et al., 2004). Similar dynamics of bounded momentum may also occur at the industry level. The benefits of vicarious learning through alliances also accrue at a decreasing rate. Later observations of repeated actions may not yield novel information or facilitate additional learning (Shaver et al., 1997). Additionally, past industry-level alliances beyond a certain point may create inherent constraints, such as increasing competition for the same partners (Axelrod, 1984) and limited carrying capacity (Aldrich, 1979). These previous studies provide a strong theoretical foundation to develop two general predictions about the strategic momentum of repetitive alliances at the firm and industry levels. Therefore, we put forth the following two hypotheses as a starting point of our study rather than an end point:

Hypothesis 1. There will be an inverted U-shaped relationship between the focal firm’s previous number of alliances and the firm’s propensity to form new alliances for all types of alliances.

Hypothesis 2. There will be an inverted U-shaped relationship between the other industry firms’ previous number of alliances and the focal firm’s propensity to form new alliances for all types of alliances.

Functional focus and alliance momentum

Examining the functional focus of strategic actions may help refine repetitive momentum studies. Functional focus is fundamental to strategic management and determines both the boundaries and content of possible learning (Zander and Kogut, 1995). March and Simon (1958) argued that learning would be bounded by “subgoal pursuit” based on functional subunits since subunit managers often have the same motivational structures, incentive alignment, cognitive frameworks, and computational constraints. Others found that managers who developed functional experience and subcultures often exhibited “selective perception” and tended to focus on opportunities and threats associated with their respective functional specialty (Ketokivi and Castaner, 2004; Waller et al., 1995). This may explain why most industries are primarily organized by and known for functional tasks and capabilities. Thus, specifically examining the effects of functional foci affords a more robust investigation of whether repetitive momentum is restricted by function or is transferrable across different functions.

A functional focus is especially important to inter-firm alliances, which often focus on separate functional activities, and different alliance types may generate distinct routines, learning, cognition, and constraints (Amaldoss and Staelin, 2010; Lavie and Rosenkopf, 2006). The existing literature provides a well-developed theoretical basis for classifying the functional focus of alliances by distinguishing among a few essential types: marketing, production, and technology alliances (Anand and Khanna, 2000; Ghemawat et al., 1986). However, previous alliance studies have lagged in conceptualizing and testing the effects of functional focus on repetitive momentum (Anand and Khanna, 2000; Hoehn-Weiss and Karim, 2014; Jiang et al., 2010). These have implicitly assumed either that capabilities from accumulated experience could be readily transferrable across functions or conversely that no connection existed across functions. These have taken either an aggregated approach by pooling alliances regardless of the functional focus or a narrow approach by examining only one functional purpose at a time.

Our study endeavors to build upon repetitive momentum studies by explicitly tracking how the functional focus of an alliance affects momentum. Our research question is whether positive and negative drivers of strategic momentum are limited by the same or cross-functional boundaries. Specifically, we examine to what extent repetitive momentum develops from a given type of alliance and/or across alliance types. We also investigate how the functional focus affects bounded momentum at both the firm and industry levels.

Same-type versus cross-type effects on alliance momentum at the firm level

Prior experience may be readily applicable to subsequent same-type alliances rather than cross-type alliances at the firm level for several reasons. “Same-Type effects” refer to the impact of a given type of past alliances on a future alliance of the same type—for example, the effect of past operation alliances on future operation alliances. “Cross-Type effects” refer to the impact of a given type of past alliance on a future alliance of a different type—for example, the effect of past operation alliances on future marketing alliances. First, for same-type alliances, the applicability of existing knowledge is more likely than for cross-type alliances due to its path dependency to exploit existing routines and the high relevance of stored experience (Cohen and Bacdayan, 1994). Second, the degree and/or amount of learning from previous experiences also depend on the level and/or the degree of involvement in prior events (Cohen and Levinthal, 1990). Compared to same-type alliances, the degree of involvement for cross-type alliances tends to be significantly limited due to its different functional roles, responsibilities, routines, required expertise, and task assignments. Third, applying prior knowledge is also influenced by its identification and accessibility. Identifying and retrieving function-specific knowledge residing within the same functional areas are often easier due to common expertise and shared cognitive frameworks. Therefore, within the same functional boundaries, it is easier to exploit existing routines and to make sequential arrangements to form additional alliances (Levinthal and March, 1993).

Beyond a certain number, repetitive alliances cannot avoid several momentum-reducing factors, which tend to offset the momentum-generating effects mentioned above. The momentum-reducing effects of repetitive alliances may also be strong for same-type alliances but not for cross-type alliances, specifically due to the resource constraints for replicating the same routines, the rapid loss of strategic flexibility, and the possible sub-additivity of additional alliances (Vassolo et al., 2004). First, firms continuously forming alliances within the same functional boundaries will ultimately exhaust their finite supply of resources (Das and Teng, 2000; Radner, 1993). Moreover, exploiting finite resources with multiple partners within the same functional boundaries can complicate resource allocation decisions. However, constraints of finite resources may be alleviated when firms engage in cross-type alliances which utilize relatively more diverse resources and knowledge across different functional departments. Second, accumulation of same-type alliances will significantly reduce the strategic flexibility (e.g. decisions on new partner selection or resource allocation) of firms, providing less motivation for forming a new alliance than those for cross-type alliances (Axelrod, 1984). However, while cross-type alliances cannot completely eliminate the constraints on potential competition, information leakage (Hamel et al., 1989), and operational constraints (Reuer and Koza, 2000), these may at least delay possible information leakage and/or attenuate potential conflicts across different function domains. Third, in terms of managing an alliance portfolio, the possible sub-additivity effect (Vassolo et al., 2004) will be especially strong for same-type alliances but not for cross-type alliances. For example, the high correlation between an ongoing alliance portfolio and an additional alliance often diminishes its marginal gain. Additionally, beyond a certain number, same-type alliances may require more complex coordination and negotiations than cross-type alliances because the alliance activities of the former are more over-lapped and/or interdependent than those of the latter (Singh and Mitchell, 1996). Resource allocation decisions for an additional cross-type alliance are also less interdependent than those for same-type alliances. Therefore, we expect that the effects of cross-type alliances will be less significant than those of same-type alliances. This suggests the following hypothesis:

Hypothesis 3. There will be an inverted U-shaped relationship between the focal firm’s previous number of alliances and the firm’s propensity to form new alliances for same-type alliances but not for cross-type alliances.

Same-type versus cross-type effects on alliance momentum at the industry level

Previous alliance experience may also be more easily applicable to subsequent same-type alliances than to cross-type alliances at the industry level. Vicarious learning by industry firms may be more effective for same-type alliances than for cross-type alliances. When trying to learn from others’ experiences, the high similarity of same-type alliances may enhance the ability to infer and imitate routines. This may encourage more imitation by the focal firm (Miner and Haunschild, 1995). Causal ambiguity also affects vicarious learning from others’ alliance experiences (Simonin, 1999). Since cross-type alliances involve different functional tasks and systems, it will be more difficult to identify critical tasks, skill sets, and constituencies. Such ambiguity may ultimately limit the imitation of cross-type alliances by other firms. Conversely, same-type alliance networks in an industry may facilitate information flows more than cross-type alliance networks can. Within heterogeneous networks of cross-type alliances, both external information and knowledge have to flow through different functional units of multiple firms, subsequently weakening imitative pressure for additional alliance formation.

The potential carrying capacity of each industry for alliances is limited (Aldrich, 1979). Assuming new entry is also limited, cumulative alliances may possibly exhaust available industry partners. As the total number of alliances in an industry increases, momentum-reducing constraints will be stronger for same-type rather than for cross-type alliances. This limited carrying capacity will be an especially critical constraint when most industry participants seek similar partners and/or resources. As explained above, the potential risk of information leakage to rivals via common partners may occur more indirectly or slowly with cross-type alliances. Some partners that could be unacceptable for same-type purposes may be marginally acceptable for cross-type purposes (Axelrod, 1984). For instance, accumulating marketing alliance partners in an industry may not reduce the set of acceptable partners for production or technology alliances. In addition, assuming that firms are likely to avoid any possible sub-additivity problems in adding new alliance partners (Vassolo et al., 2004), the last available partners who may be less attractive for one type of alliance may become a reasonable partner for another type of alliance. Altogether, we expect that the effects of cross-type alliances will be less significant than those of same-type alliances at the industry level. Thus, we predict:

Hypothesis 4. There will be an inverted U-shaped relationship between the other industry firms’ previous number of alliances and the focal firm’s propensity to form new alliances for same-type alliances but not for cross-type alliances.

Methods

The international passenger airline industry context

The international passenger airline industry provides a very useful context for this study because international passenger airlines need to form cross-border alliances to increase their operational scopes and because regulation hinders acquisitions and stand-alone expansion abroad (Oum et al., 1993; Pustay, 1980). Accordingly, cross-border alliances are one of the most popular strategies in this industry. Most alliance announcements in this industry contain a well-defined set of reliable data about functional focus. Due to intense global competition, the decision to form alliances is very critical to organizational survival (Brander and Zhang, 1993). For example, once-strong airlines, such as Swissair, went bankrupt partly because it did not effectively manage alliances (Knorr and Arndt, 2004).

International Air Transport Association (IATA), the authoritative worldwide industry group, defines an international passenger airline as a firm that provides scheduled international passenger transportation services year-round. The international passenger airline industry grew after significant technological advances in cross-continent jet airplanes, especially after the launch of B707 with 200 passenger seats in 1955 (Kotha, 2010). Major industry characteristics include high product homogeneity, service route network expansions, frequent flier mileage programs, extremely low R&D intensity, and high government regulations (Holloway, 1997).

The above characteristics lead individual firms to concentrate on two critical resources: service route networks and passenger-carrying capacity. Airlines with well-established networks and sufficient passenger-carrying capacity can gain considerable benefits if their service networks access major traffic flows and capture the growth potential of these flows. Thus, adding an additional service route to the existing networks can create significant benefits from both increased traffic and economies of scope. Access to slots (gates) at congested airports helps airlines both maintain efficient network structures and improve their service schedules. Since market entry is severely limited at congested airports, airlines dominating those airports are better able to increase ticket prices (Kunz, 1999). It is also possible to increase cash flow by leasing slots to other airlines.

International airlines maintain competitive positions in their own home markets where government regulations prevent foreign airlines from entering on their own or engaging in outright mergers and acquisitions (M&As; Oum et al., 2001). Previous studies have suggested that strategic dominance in home markets is an important resource in geographically based competition (Bailey and Williams, 1988). The bilateral agreement system shaped by the Chicago Convention in 1944 stipulates the airports to be served in each country, the available capacity, and the number of airline signatories designated to serve specific routes. This convention significantly restricts foreign airlines from entering specific international routes and accessing landing slots. For example, under the bilateral regime, no airline is allowed to provide transportation services from a foreign country to another foreign country unless it receives the fly-beyond right permission. Over time, strong bilateral regulations evolved into a multi-lateral agreement among countries in the same region. For example, the single European aviation market came into existence on 1 April 1997 with an agreement allowing any European Union (EU)-registered carrier to provide transportation services as domestic routes within any of the EU’s 15 member countries. Even though restrictions have gradually eased since then, strong governmental regulations in the late 1990s made service routes and airport access all the more critical. The above features made cross-border alliances a critical strategic activity.

Cross-border alliances are usually classified as operation and marketing alliances. Airline companies utilize operation alliances to expand route networks, increase passenger-carrying capacity, and/or improve operational efficiencies. Typical examples of operation alliances include joint service, joint baggage handling service, and shared aircraft maintenance. Airlines also leverage marketing alliances to increase passenger-carrying capacity with joint promotions, shared sales offices, and cross-selling of partners’ tickets. There were only a few alliances before the 1980s. Since 1987, airline companies have significantly increased the frequency of forming multiple alliances with their competitors (see Figure 1). For example, American Airlines has established more than 40 cross-border alliances with 22 foreign airline companies during the last two decades. Among the top 65 airline companies in terms of annual revenue, the total number of strategic alliances increased to over 1200 alliances by 2000. Especially since the late 1990s, the typical bilateral characteristics of cross-border airline alliances significantly shifted to complex multi-lateral arrangements among multiple partners, such as Oneworld, Sky Team, and Star Alliance. One precursor to multi-airline alliances was the granting of antitrust immunity by the US Department of Transport to the bilateral alliance between Northwest Airlines and KLM airlines. This eventually led to the anticipation of a generalization of such immunity and open skies agreements, which paved the ground for multi-airline alliances. Although the bilateral alliance between Northwest and KLM was signed in late 1992, it did not receive immunity immediately and was fully rolled out only in 1994 (Doganis, 2010).

Figure 1.

Historical cumulative cross-border alliances by functional type (1982–1994).

Data

To investigate same-type and cross-type alliance momentum, this study requires a longitudinal sample of firm- and industry-level observations with a detailed timeline. We collected monthly observations of alliances formed by 32 international passenger airlines from January 1982 to December 1994 (the study period), controlling for earlier alliances and those of other major international airlines. Monthly observations allowed us to obtain precise estimates of the accumulated momentum as of the time a new alliance became effective. To avoid any left censoring in computing the independent variables, we collected data on each sample firm’s alliances all the way back to 1945. We intentionally terminated our observation period at the end of the year 1994 because the nature of cross-border alliances significantly changed from bilateral to multi-lateral agreements in the late 1990s, and multi-lateral alliances may be a unique phenomenon. However, it is theoretically and empirically challenging to identify unique motivations of alliance formation and functional foci. Therefore, these multi-lateral alliance agreements are beyond the scope of our research questions.

The IATA lists 64 airlines with substantial international operations at the end of the study period (IATA, 1995). All are scheduled international passenger carriers, ensuring comparability among firms. The data collection processes are extremely demanding and time-consuming because various activities of cross-border alliance record tracking and extensive language translation efforts were inherently necessary in the global context. Our data included 32 international passenger carriers, or half the total, in a total of 28 countries and a comprehensive historical alliance record for up to 49 years.

However, to ensure the sample was representative of the industry’s firm size, regional distributions, and industry-wide historical evolution, we relied on a random sampling method. We randomly selected 50% from each set of 10 consecutive airlines, as ranked by size in 1994. The sample includes 11 airlines from the Americas; 11 from Europe, the Middle East, and Africa; and 10 from Asia and Oceania (see Appendix 1). Our final sample approximately reflects each region’s share of worldwide traffic although a few well-known airline companies may be excluded due to the random sampling process. Of the 32 airlines, 24 had international operations before 1945, while the other 8 started international operations during the study period.

We identified and coded all cross-border alliances that these 32 airlines established with other international airlines. These alliances are typically horizontal, that is, these link international industry rivals. To verify that these alliances actually were established, we used multiple sources including annual reports, company publications, international newspapers, and leading industry sources, such as Airline Business, Flight International, and the World Airline Directory. We excluded a few reports of potential alliances that never materialized in practice. We also excluded alliances with purely domestic airline operators or non-airline partners, such as car rental or credit card companies, for three reasons. First, these vertical alliances among non-rivals are theoretically different from horizontal alliances, since these are often initiated by different motives. Second, most of these alliances are purely domestic, and the national regulations on these vertical alliances are too heterogeneous to allow meaningful comparisons across countries. Third, most of these domestic vertical alliances, compared to the cross-border horizontal alliances in our sample, were temporal or short-term arrangements that usually survived for a few months. It was beyond our research scope to investigate these domestic or vertical alliances. Finally, reliable and comprehensive data were significantly limited for most non-US countries.

We took multiple steps to ensure data validity and reliability. This approach resembles earlier studies of competitive actions in the airline industry and elsewhere (Chen et al., 1992; Ferrier et al., 1999). First, we drew a comprehensive list of key words describing airline alliances from Groenewege (1996) and Airline Business. We clarified and refined this list by examining 60 alliance announcements and thorough interviews with airline industry executives. To ensure a valid coding scheme, we conducted interviews with executives from four airlines on three continents, two industry analysts, and two industry consultants. We first asked these experts to describe the main types of alliances in the industry. They consistently indicated that the most fundamental distinction in practice is between marketing and operation alliances. Indeed, several interviewees mentioned that alliances were both designed and managed along functional lines. The experts suggested additional key words and helped us assign all key words to alliance types. We arranged multiple codings to ensure reliability. A rater with substantial experience as an airline consultant coded all announcements. Working separately, two graduate students with prior industry expertise also coded cross-border alliances. Perreault and Leigh’s (1989) index was used to assess initial coding reliability at the level of key words. For operation and marketing alliances, the index is 0.97, which indicates very high reliability. The few discrepancies were subsequently resolved by verifying original sources or by contacting experts and/or the airlines.

Dependent variables

We created three different count variables as dependent variables for our study. The first, New Alliances, counts all cross-border alliances, regardless of the alliance type, formed by a focal airline with all other international passenger airlines that became effective in a given month. The other two distinguish between the functional types of alliances. Based on experts’ opinions, we predicted and found that almost all cross-border alliances in the airline industry were of the operation or marketing type. Following existing activity-based taxonomies of alliances, we also sought to identify alliances for R&D/technology development. We found very few technology-related alliances—only 18, far less than one per sample airline and too few to include as a stand-alone type in the analysis. Our results are not sensitive to the inclusion of technology alliance counts in a broader third category. However, we did not include these alliances in our analysis and focused on the primary operation and marketing activities. Accordingly, the other two dependent variables we created based on alliance type are (1) New Operation Alliances, which counts cross-border alliances formed by a focal airline with all other international passenger airlines that became effective in a given month and dealt with operation activities, and (2) New Marketing Alliances, which counts cross-border alliances formed by a focal airline with all other international passenger airlines that became effective in a given month and dealt with marketing activities. The data coding yielded 315 cross-border alliances that could be classified as operation or marketing alliances: 218 operation alliances and 97 marketing alliances. Raw data included 56 other alliances that appeared to encompass some operation and some marketing activities. Since the functional purposes of these alliances were unclear, we conducted analyses excluding these 56 alliances, assigning them to both counts of functional alliance type, and assigning them to each count of functional alliance type in turn. The results reported below were robust and unaffected by all treatments of these 56 alliances.

Independent variables

We created two sets of independent variables at the firm and industry level, respectively. The first set counts the total number of alliances formed by a focal firm or other industry firms without distinction of alliance type. The variable Focal Firm’s Total Alliances was created by summing the number of alliances formed by a focal airline preceding the focal month. This represents a cumulative count of all past alliances involving a focal airline. Industry-level variables were obtained by summing past alliances of all sample firms except the focal firm. Thus, the variable Other Firms’ Total Alliances represents the cumulative count of alliances by all but the focal firm. The second set counts the number of alliances formed by a focal firm or other industry firms similarly but with alliance type distinction. The variable Focal Firm’s Operation Alliances was created by summing the number of operation alliances formed by a focal airline preceding the focal month. The variable Focal Firm’s Marketing Alliances was similarly computed to count a focal firm’s past marketing alliances. At the industry level, the variable Other Firms’ Operation Alliances represents the cumulative count of operation-related alliances by all but the focal firm. Similarly, the variable Other Firms’ Marketing Alliances counts all marketing alliances by firms other than the focal firm. To examine non-monotonic effects, we computed squared terms of each of these independent variables. We lagged each independent variable by 1 month to avoid modeling the dependent variable as a function of its own current value.

Control variables

We include six control variables—firm size, experience with international operations, size of the focal airline’s home market, number of available alliance partners, government ownership dummy, and number of newly added international airlines—which are all potential indicators of a firm’s attractiveness as a cross-border alliance partner and its exposure to alliance opportunities and congestion effects. Together with the fixed firm effects described below, these variables help control for firms’ propensities to form cross-border alliances and their (and their environment’s) alliance carrying capacity (Baum and Oliver, 1992). It is important to control for firm size because past research found conflicting results on the effect of firm size on alliance formation and strategic momentum (Eisenhardt and Schoonhoven, 1996). A standard measure of airline size is the total seat kilometers available for sale by the firm in a given year, which also represents the scale of route network (Oum et al., 1993). The resulting variable was expressed in millions, log-transformed, and lagged by 1 year. This measure of firm size also helps control for a firm’s financial and managerial resource endowment given that financial data are not available. Controlling for a firm’s international experience is also relevant because a firm’s own experience may act as a substitute for (or complement) experience obtained through alliances. We used the months elapsed since it started international route services as a proxy for an airline’s international experience. Using the firm’s overall age as an alternative measure for experience, which includes domestic operations, did not change the results. Controlling for the size of an airline’s domestic market is also relevant because growth in cross-border alliances may simply reflect increasing demand rather than alliance momentum. The primary international market for an airline, under the bilateral arrangements regime, consisted of travelers into or out of its home country. We used another standard indicator to measure this (Keeler and Formby, 1994): the total passenger kilometers performed by all airlines into and out of that country in a given year, expressed in millions, log-transformed, and lagged by 1 year. Replacing that variable with a measure including domestic passenger kilometers did not change the results. We also include the number of available alliance partners to control for the diminishing demand for alliances. We counted each year’s total number of airlines performing international flights from the International Civil Aviation Organization (ICAO) database. To control for ownership and governance characteristics, we include ownership dummy variable. We coded 1 if the focal firm is a private-owned airline and coded 0 if the focal firm is a state- or government-owned airline. Finally, we include the number of newly added international airlines in a given year. We counted each year’s newly added number of airlines performing international flights from the ICAO database.

Statistical methods

Using the number of new alliances as the dependent variable raises econometric issues that are common in studies of count variables. Following Hausman et al. (1984), we specify a Poisson regression to model the probability that a firm will form n alliances in a given month (with n = 0, 1, 2, …) as follows: (A) Prob(Y = y_j) = e⁻^λj λ_j^Yj/Y_j!, where Y_j is the count of alliances for the entries of the jth firm. To incorporate exogenous variables, lambda can be expressed as a function of the covariates: (B) λ_j = exp(ΣB_i X_ij), where Bs are the coefficients, Xs are the covariates (with X1 set to 1), i is the ith variable, and j is the jth firm. The exponential function ensures non-negativity.

The Poisson distribution contains a strong assumption that the mean and the variance of the explained variable are equal to lambda. Below, we report the diagnostic tests used to examine this assumption. To address the potential problem of overdispersion, whereby the mean differs from the variance, a firm-specific error term can be specified. Equation (B) then becomes (C) λ_j = exp(ΣB_i X_it)exp(u_j), where λ_j is no longer determined but is itself a random variable. As u_j is unobserved, it is integrated out of the expression by specifying a gamma distribution for the error term, whereupon the now-compound Poisson reduces to the negative binomial model. Only the scale of the distribution is permitted to vary as a function of the covariates. The variance of Y_j is parameterized to equal (1 + α) E(Y_j), yielding a constant variance–mean ratio. This specification is a standard way of accounting for overdispersion. Since each firm figures into the data multiple times (once per each period during which it operates internationally), a fixed- or random-effects specification can be used to control for unobserved firm-specific effects that may otherwise bias negative binomial estimates (Greene, 2002).

We successfully replicated the results to address two potential estimation problems. First, some firms may not be at risk of forming alliances, at least initially. This is a lesser concern in our case since all firms were clearly active in alliance formation soon after entering the sample, if not before. Nevertheless, to address this possibility, we ran analyses while including only firms that had already entered into at least one alliance. Second, some alliance observations were represented twice in the data if both partners were among the 32 focal airlines. To address the possible oversampling, we also ran maximum likelihood estimations where weight of such observations was reduced.

Results

Descriptive statistics

Our sample contains 4456 airline-month records. The descriptive statistics are shown in Table 1. The number of alliances formed by an airline in a given month varies from 0 to 4. The correlations between independent variables measuring past firm- and industry-level alliances are consistent with the presence of momentum. High correlation between the variables Other Firms’ Operation Alliances and Other Firms’ Marketing Alliances is an inherent feature of cumulative data.

Table 1.

Descriptive statistics and Pearson’s correlation coefficients.

Variables		M	SD	Min	Max	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1	New Alliances	0.98	0.37	0	4	1
2	New Operation Alliances	0.07	0.31	0	4	0.84	1
3	New Marketing Alliances	0.03	0.21	0	4	0.56	0.01	1
4	Focal Firm’s Operation Alliances	3.52	5.71	0	34	0.20	0.18	0.09	1
5	Other Firms’ Operation Alliances	72.22	60.30	2	216	0.16	0.13	0.10	0.44	1
6	Focal Firm’s Marketing Alliances	1.35	2.27	0	14	0.14	0.09	0.12	0.47	0.54	1
7	Other Firms’ Marketing Alliances	29.89	27.06	1	97	0.15	0.12	0.09	0.48	0.77	0.51	1
8	Other Firms’ Total Alliances	104.5	88.4	5	316	0.16	0.13	0.09	0.45	0.90	0.53	0.79	1
9	Firm size (in millions)	268.4	304.1	0	2139	0.11	0.08	0.07	0.29	0.22	0.52	0.20	0.21	1
10	International experience (in months)	400.3	209.9	0	895	0.04	0.05	−0.00	0.18	0.03	0.03	0.05	0.04	−0.14	1
11	Home-market demand (in millions)	169.3	307.0	0	2197	0.02	0.01	0.02	0.05	0.11	0.16	0.10	0.11	0.06	0.09	1
12	Available alliance partners	178.5	15.4	158	208	0.16	0.13	0.10	0.45	0.90	0.50	0.88	0.78	0.22	0.04	0.10	1
13	Ownership dummy	0.45	0.49	0	1	0.06	0.03	0.06	0.11	0.18	0.23	0.16	0.18	0.31	−0.35	0.16	0.20	1
14	Number of newly added airlines	4.27	9.31	−12	17	0.05	0.03	0.05	0.11	0.21	0.12	0.21	0.21	0.03	0.01	0.02	0.48	0.02	1

SD: standard deviation.

The number of observations for all variables is 4456.

Regression analysis

We estimated five regressions with the different sets of independent variables. The dependent variable of the first three regressions is New Alliances and the dependent variables for the fourth and fifth models are New Operation Alliances and New Marketing Alliances, respectively. We conducted Cameron and Trivedi’s (1990) T_opt test to examine whether the mean and the variance of λ are equal and found evidence of overdispersion in our regression models (p < 0.01 in each model). Furthermore, based on the Vuong (1989) test, a zero-inflated model does not improve model fit (p > 0.10 in each model). Accordingly, we report results of negative binomial regressions that conservatively account for overdispersion. We replicated results with the Poisson and zero-inflated models and with a binary dependent variable to indicate the formation of at least one alliance per period and did not find a substantial difference. Hausman tests indicate that a fixed-effects specification is more suitable than random effects. The fixed-effects specification is a powerful way to control for unobserved factors such as any residual heterogeneity in alliance opportunities and capabilities not explained by our independent and control variables (Greene, 2002). Recent research indicates that fixed-effects models are, in general, preferable to random-effects models when studying repeated organizational change patterns, such as strategic momentum (Beck et al., 2008).

Models 1–3 in Table 2 provide analysis for New Alliances without distinction of alliance type. These models have substantial explanatory power, as indicated by their overall χ² statistics. The χ² for model 1 is 274.8, the χ² for model 2 is 2556.5, and the χ² for model 3 is 1130.5. Each model is statistically significant as a whole (p < 0.01). Model 1 is a base model which investigates the influence of control variables only. It shows that all control variables have a significant impact on new alliance formation in Model 1. Model 2 examines the effect of firm-level momentum and confirms the inverted U-shaped momentum. Model 3 shows that the effect of industry-level momentum is significant and inverted U-shaped. Therefore, these results support our Hypotheses 1 and 2, which are the starting points of our research, confirming the argument of previous studies on alliance momentum (Chung et al., 2000; Gulati, 1995a).

Table 2.

Fixed-effects negative binomial analysis of alliance formation.

Dependent variable	New Alliances	New Alliances	New Alliances	New Operation Alliances	New Marketing Alliances
Independent variable	Model 1	Model 2	Model 3	Model 4	Model 5
Control variables
Firm size	−0.194*	−0.093⁺	0.090	0.206*	0.046
Firm size	(0.076)	(0.048)	(0.076)	(0.102)	(0.135)
International experience	0.272**	0.022	0.058⁺	0.091*	−0.014
International experience	(0.055)	(0.024)	(0.030)	(0.037)	(0.057)
Home-market demand	−0.203**	−0.036	−0.002	−0.026	−0.007
Home-market demand	(0.046)	(0.024)	(0.031)	(0.040)	(0.057)
Available alliance partners	−0.011**	−0.018**	−0.031**	−0.036**	−0.043**
Available alliance partners	(0.003)	(0.002)	(0.003)	(0.004)	(0.006)
Ownership (dummy)	0.774**	0.162	0.165	0.122	0.375
Ownership (dummy)	(0.181)	(0.105)	(0.142)	(0.174)	(0.282)
Newly added airlines	0.024**	0.020**	0.028**	0.024**	0.054**
Newly added airlines	(0.006)	(0.006)	(0.006)	(0.007)	(0.011)
Firm-level momentum
Focal Firm’s Total Alliances		0.270**	0.156**
Focal Firm’s Total Alliances		(0.018)	(0.024)
(Focal Firm’s Total Alliances)²		−0.626**	−0.402**
(Focal Firm’s Total Alliances)²		(0.061)	(0.070)
Industry-level momentum
Other Firms’ Total Alliances			0.026**
Other Firms’ Total Alliances			(0.003)
(Other Firms’ Total Alliances)²			−0.061**
(Other Firms’ Total Alliances)²			(0.009)
Firm-level momentum by function
Focal Firm’s Operation Alliances				0.229**	0.058
Focal Firm’s Operation Alliances				(0.033)	(0.047)
(Focal Firm’s Operation Alliances)²				−0.709**	−0.214
(Focal Firm’s Operation Alliances)²				(0.120)	(0.162)
Focal Firm’s Marketing Alliances				−0.103	0.250*
Focal Firm’s Marketing Alliances				(0.076)	(0.118)
(Focal Firm’s Marketing Alliances)²				0.400	−1.850⁺
(Focal Firm’s Marketing Alliances)²				(0.668)	(1.071)
Industry-level momentum by function
Other Firms’ Operation Alliances				0.084**	−0.031
Other Firms’ Operation Alliances				(0.017)	(0.023)
(Other Firms’ Operation Alliances)²				−0.027**	0.002
(Other Firms’ Operation Alliances)²				(0.007)	(0.009)
Other Firms’ Marketing Alliances				−0.129	0.238**
Other Firms’ Marketing Alliances				(0.039)	(0.057)
(Other Firms’ Marketing Alliances)²				0.098	−0.160**
(Other Firms’ Marketing Alliances)²				(0.032)	(0.045)
N	4456	4456	4456	4456	4456
Model χ²	274.8	2556.5	1130.5	933.4	610.7
Degrees of freedom	6	8	10	14	14

Standard errors in parentheses.

p < 0.10; *p < 0.05; **p < 0.01.

Models 4 and 5 in Table 2 provide analysis for New Operation Alliances and New Marketing Alliances, respectively. These models also have substantial explanatory power, as indicated by their overall χ² statistics. The χ² for model 4 is 933.4, and the χ² for model 5 is 610.7. Each model is statistically significant as a whole (p < 0.01). We classified firm- and industry-level alliances by functional type and used these alliance counts to measure momentum effects. In model 4, Focal Firm’s Operation Alliances has a positive main effect and a negative quadratic effect on the firm’s propensity to form further same-type alliances (p < 0.01 for both terms). The effect of Other Firms’ Operation Alliances also shows same-type bounded momentum at the industry level (p < 0.01 for both terms). However, neither Focal Firm’s Marketing Alliances nor Other Firms’ Marketing Alliances has a statistically significant main or quadratic effect. In Model 5, Focal Firm’s Marketing Alliances has a positive and significant main effect, and a negative and significant quadratic effect. The main and quadratic coefficients for Other Firms’ Marketing Alliances suggest a statistically significant inverted U-shaped effect at the industry level (p < 0.01 for both terms). Focal Firm’s Operation Alliances and Other Firms’ Operation Alliances have statistically non-significant effects. Our results show that the inverted U-shaped pattern of alliance momentum holds for same-type but not for cross-type alliances at the firm and industry levels. These results support Hypotheses 3 and 4. We display in Figure 2 the prediction of the estimated effects by same-type and cross-type alliances at both the firm and industry levels.

Figure 2.

Patterns of alliance momentum by same type and cross type.

We assess the effective scale of momentum effects by computing the combined effects of the main and quadratic terms for each significant set of effects. For firm-level operation alliances, the combined effect peaks between the 16th and 17th alliance that a firm forms. The net effect becomes negative past the 33rd alliance. At the beginning of the study period, no firm had accumulated 17 or more operation alliances. By the end of the study period, six firms had accumulated 17 or more operation alliances. Thus, according to our estimates, these firms experienced declining momentum for further operational alliances. One firm accumulated enough alliances, 34 by the end of the study period, that its estimated propensity to form further alliances was slightly less than if it had no prior alliances. For firm-level marketing alliances, the combined effect peaks between the sixth and seventh alliance and remains positive through the observed range. At the beginning, no firm had accumulated more than six alliances. By the end, according to our estimates, four firms experienced declining momentum for further marketing alliances, although none faced less momentum than it would have if it had no prior alliances. The results suggest that firm-level effects encouraging marketing alliances run out after fewer alliances than those encouraging operation alliances. Still, most sample firms maintained substantial alliance momentum throughout our entire study period.

Some momentum also remained at the industry level. According to the results of the first model where we found a significant industry-level same-type effect, the combined effect of other firms’ operation alliances peaks after 151 alliances. According to this computation, the firms in our sample started experiencing downward effects of other firm operation alliances, although only in the last 23 months of the study period. For each firm, the net cumulative effect of other firms’ operation alliances remained substantially higher in the early 1990s, implying that industry-level momentum for operation alliances had yet to run its course.

Robustness of results

We conducted additional analyses to verify the robustness of these findings. First, we examined whether the non-significant results reported above are due to correlated variables obscuring actual effects by rerunning models after excluding independent variables, in turn counting same-type and cross-type alliances. We compared the results with the full models, given that in cases of extreme collinearity and over-fitting, the addition of a problematic variable will cause other variables to exhibit unrealistic increases in estimated coefficients and standard errors, typically by several orders of magnitude. At the firm level, excluding either set of variables did not substantially change the remainder of the coefficients and standard errors. Excluding Focal Firm’s Operation Alliances and its quadratic term from model 4 decreases explanatory power significantly (χ²(2) = 48.54, p < 0.01), while excluding Focal Firm’s Marketing Alliances and its quadratic term does not (χ²(2) = 4.22). Excluding Focal Firm’s Marketing Alliances and its quadratic term from model 5 decreases explanatory power significantly (χ²(2) = 8.14, p < 0.01), while excluding Focal Firm’s Operation Alliances and its quadratic term does not (χ²(2) = 1.76). Thus, at the firm level, the pairs of cross-type effects do not contribute significant explanatory power (beyond same-type effects), while the pairs of same-type effects contribute substantial explanatory power (beyond the cross-type effects). The sensitivity analysis of industry-level effects also complements our main findings. Excluding Other Firms’ Operation Alliances and its quadratic term from model 4 reduces its explanatory power significantly (χ²(2) = 30.46, p < 0.05), while excluding Other Firms’ Marketing Alliances and its quadratic term does not (χ²(2) = 3.91). Excluding Other Firms’ Marketing Alliances and its quadratic term from model 5 decreases explanatory power significantly (χ²(2) = 19.29, p < 0.01), while excluding Other Firms’ Operation Alliances and its quadratic term does not (χ²(2) = 2.76). Thus, also at the industry level, the pairs of cross-type effects do not contribute significant explanatory power (beyond same-type effects), while the pairs of same-type effects contribute substantial explanatory power (beyond the cross-type effects).

We also checked the robustness of our findings to possible omitted variables. Statistically, fixed effects stand to absorb many plausible forms of unobserved heterogeneity among firms (Greene, 2002). For completeness of analysis, we examined several possible sources of heterogeneity. One possible source consists of a time effect. When we included a clock variable measuring calendar time in the analyses, its effect was very low and non-significant, and the remaining effects were essentially unchanged. This confirms that momentum differs conceptually and empirically from a simple time trend (Kelly and Amburgey, 1991). Another source may be the possibility that recently applied momentum may be more salient in driving future momentum than those utilized in the more distant past (Gulati, 1995a). To test the effect of organizational short memories, we used narrow windows of both 3 and 5 years to count alliance-related independent variables. We also conducted exactly the same analysis using annual data rather than monthly data. We found no significant difference compared to the reported results. To investigate the possible omission of some momentum-related factors, we replicated the results while including variables counting technology and other (including mixed and ambiguous) types of alliances. Their effects were negligible, and the effects of operation and marketing alliances did not change materially. We also checked the robustness of our findings by controlling for the potential impact of firm performance by using the variables focal airlines’ market share and annual operating profit. Moreover, we replicated our analysis with a traffic demand–related control variable. These additional control variables did not change our original results.

Finally, we examined whether equity-based alliances might exhibit a different momentum from non-equity-based alliances. We found no evidence of such an effect. This may be because airline regulators impose severe limits on international equity transactions.

Discussion

The exploration and exploitation literature highlighted the importance of balancing exploration and exploitation in various function domains (Lavie et al., 2011; Lavie and Rosenkopf, 2006; March, 1991; Rothaermel and Deeds, 2004). However, existing studies have not paid sufficient attention to how and why firms execute alliances for exploration and exploitation over time, leaving the question of whether firms pursue exploration and exploitation simultaneously or sequentially in function domains unsolved. To address this question, we analyzed cross-border alliances of 32 international airlines from 1945 to 1994. Unlike previous studies which simply treated previous alliances at an aggregated level, we examined how and why the function domains of inter-organizational alliances influence future alliance formation decision at the firm and industry level. Our results show that there are limitations to momentum and that the bounded momentum is influential only for same-type but not for cross-type alliances at the firm as well as the industry level. The drivers that hinder momentum—including resource constraints, the challenges of maintaining strategic flexibility, and the sub-additivity of alliance portfolio—ultimately counteract drivers that initially foster momentum. Our study also extends the bounded momentum model to cross-border alliances at the industry level. Specifically, at the industry level, the alliance momentum accelerated by several drivers—including isomorphism, vicarious learning, and lower causal ambiguity—cannot ultimately overcome the momentum-reducing factors. We also find that bounded momentum drivers are more function-specific for both operation and marketing alliances.

Our study highlights the need for academic attention to the role of functional focus in momentum phenomena. We show that the theoretical drivers of same-type momentum do not transfer readily across functional specialties, neither within firms, nor across firms in the same industry. These findings imply that alliance momentum with a specific functional focus evolves sequentially rather than simultaneously. That is, if a firm establishes a few alliances with a functional focus, it tends to sequentially repeat same-type alliances rather than simultaneously also engaging in other cross-type alliances. This repetition is pursued until it reaches an inverted U-shaped momentum. Our follow-up interviews with industry experts confirmed that, by nature, cross-border alliances require sharing more sensitive information, such as passenger travel patterns, competitive dynamics in a specific route, operation know-how of flight allocation, customer-related data, tacit know-how about promotion channels, pricing, and advertising. Considering the typically localized competition in the airline industry (Chen et al., 1992), the momentum drivers of cross-border alliances may be less evident because rival firms may not directly observe specific details about other firms’ alliances in geographically distant countries. Another important factor might be the nature of strong government regulations. Since strong government regulations do not allow alternative means of international expansion, such as organic growth (adding routes) or acquisitions, alliance momentum could be much stronger for same-type alliances than cross-type alliances. Our findings suggest that functional boundaries may significantly affect momentum drivers not only within firms but also across firms in an international context.

The significance of same-type effects implies a need to revisit previous research which overlooks the impact of function-specific effects. For example, if alliance momentum is type-specific, research designs pooling distinct alliance types may generate confounding errors. Not fully considering the function-specific effects at either the firm or industry level may introduce omission biases, thus possibly overestimating the results. In a similar vein, Vasudeva and Anand (2011) refined absorptive capacity into “latitudinal” and “longitudinal” components, corresponding to the use of diverse and distant knowledge, respectively. This kind of strict approach can better explicate in-depth mechanisms of various momentums.

Our results may also refine previous research on alliance portfolio management, alliance termination, inter-firm networks, and performance research (Greve et al., 2010; Park et al., 2004; Wassmer, 2010). Previous network studies investigated the formation of complex networks within industries, albeit only with monotonic specifications (Hagedoorn et al., 2011; Powell et al., 1996). However, we have argued that the propensity and the potential gains for firms engaged in network-shaping alliances are type-specific and stand to be reversed. Thus, our study may extend network-theoretic concepts by identifying suitable timing and positioning stances (centrality, etc.) as alliances spread within (and between) industries.

Our study may also inform momentum literature of other strategic actions, such as business expansions, diversifications, and organizational transformation if these strategies have a strong focus on functional aspects. For instance, relative to alliances, while diversifications may alleviate problems with knowledge spillovers due to common ownership, they tend to generate greater complexity due to asset indigestibility for future activities (Hennart and Reddy, 1997). Researchers should be aware that functional foci of bounded momentum may entirely reshape the strategic momentum at either the firm or the industry levels—an important refinement of existing models such as that of Amburgey and Miner (1992).

Our research also has implications for research that seeks to compare firm and industry effects in corporate strategies, as strategy research often does (e.g. Hoehn-Weiss and Barden, 2014; Madhok et al., 2015). Our evidence confirms that both the firm and industry effects matter, but differ in type specificity. This means that at least for cross-border alliances and plausibly for other forms of strategic actions, it is important to take functional types into account when comparing firm and industry effects. More precise models of cooperative strategies and competitive dynamics should likewise examine the theoretical association between the functional foci and cross-level effects.

Our study implies that managers should select alliances with bounded and type-specific momentum effects in mind. Initial alliance activity, especially its functional focus, may determine not only what learning opportunities a firm obtains but also what strategic constraints it will face in subsequent alliances. Thus, early alliances should be chosen carefully with planned future partnerships in mind. Firms should add alliances cautiously, as their optimal number is bounded, especially within a limited scope of single function. Furthermore, managers who recognize alliances as learning devices or real options should understand that the potential benefits and constraints of successive alliances may not transfer across functions as much as expected. Managers in other industries where alternative entry modes are readily available should make sure to compare each alliance opportunity with these alternatives. Finally, managers should also be keenly aware of isomorphism, the bandwagon effect, vicarious learning, as well as associated strategic constraints.

Limitations and future research

Limitations from our study may provide future research opportunities. First, the international airline industry is subject to potential data limitations since existing databases do not fully account for the survival of past alliances. Cross-border alliances between airlines are usually of indefinite duration and/or automatically renewable. Thus, we cannot assume that these alliances end naturally after a given period of time. Based on very few published records and information obtained directly from airlines covering 30 alliances, we estimate that approximately 20% of alliances were terminated by 1995. Unfortunately, for most of the sample, alliance terminations cannot be documented or logically inferred. However, we do not believe this limitation invalidates our results. Terminated alliances can have a strong effect past their apparent duration on residual learning, resource constraints, and strategic outlook. If they do not, then their momentum-inducing and momentum-reducing effects should be weaker than those of ongoing alliances, if not opposite in sign. Unlike terminated alliances, the market demand-side causal mechanism can have a strong influence on decisions regarding cross-border alliance formation. Such decisions include expending route networks to serve a new country, new airline companies participating in international operations, and reactions to an increase in world market demand as well as an expansion of route networks. However, our study could not include the detailed mechanisms of demand-side influences and their effect on cross-type alliances, which will certainly be an interesting topic for future research.

Second, this research also suggests the importance of alliance heterogeneity. Our research design allows the effect of each alliance to vary based on its functional focus and its order relative to other alliances. However, data limitations prevented us from incorporating all other dimensions of alliances. For instance, the relative scale of various alliances may affect the magnitude of positive and negative momentum. Future research could examine distinctions between horizontal cooperative strategies among industry participants (as studied here) and vertical cooperative strategies between buyers and suppliers.

Third, while marketing and operation alliances represent the overwhelming majority of partnerships among international passenger airlines, in other contexts, different categories of alliances may prevail, such as R&D alliances. In such contexts, researchers should examine whether these alliances also exhibit bounded and type-specific momentum. Since our study focused upon the services industry, we also believe there is a need for the direct comparison of momentum in the manufacturing and other industries. Given the unique phenomenon of the international passenger carrier industry in which global competition is limited by geographical locations, route networks, and size, for most other industries, we believe further investigations into industry-level momentum with the functional foci of alliances may be clearly warranted.

Finally, our findings suggest opportunities for research on performance implications of alliances (and, by extension, of other forms of strategic actions subject to bounded momentum). Given that a firm already possesses a heterogeneous functional portfolio of cross-border alliances, the potential impact of an additional alliance may be affected by the order of an alliance relative to the previous alliance portfolio of the firm and by industry participants, together with alliance types. Relevant performance indicators may include financial returns (Anand and Khanna, 2000; Park and Mezias, 2005), innovation (Deeds and Hill, 1996; Rothaermel and Deeds, 2004), firm growth, and longevity (Singh and Mitchell, 1996).

Conclusion

Focusing on the time dimension of executing exploration and exploitation activities, we investigated whether firms establish multiple alliances in different function domains sequentially or simultaneously, that is, sequentially exploiting multiple alliances in a single-function domain or simultaneously exploring multiple alliances in various function domains. Unlike most previous alliance studies which analyzed existing alliances at an aggregated level, we examined how and why the functional focus of inter-organizational alliances affects future alliance formation. We found that the inverted U-shaped pattern of alliance momentum holds for same-type but not for cross-type alliances at the firm as well as the industry level. These findings imply that alliance momentum with a specific functional focus evolves sequentially rather than simultaneously. The implications are substantive in regard to predicting what alliances firms will form and how they execute alliances for exploitation and exploration over time. Further research, especially with a refined functional focus, is well warranted for various forms of strategic actions at multiple levels of analysis.

Footnotes

Appendix 1 Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author biographies

Namgyoo K. Park (PhD, New York University) is a Professor of Strategy and International Business at Seoul National University. His research investigates knowledge creation, creativity management, firm boundary expansion, and firm performance in both domestic and global settings. His research has appeared in the Strategic Management Journal, Journal of Management, Creativity Research Journal, Transportation Research, Journal of Air Transport Management, Asia Pacific Journal of Management, Industry and Innovation, and International Journal of E-business Strategy.

Xavier Martin (PhD, University of Michigan) is a Professor of Strategy, International Business and Innovation at Tilburg University (The Netherlands). His research examines how corporate strategies, inter-firm relationships, and knowledge-based assets affect each other and jointly affect firm performance. This article is a study of alliances, but he also researches buyer–supplier relations and other forms of inter-organizational relations in corporate strategy, international, innovation, and entrepreneurial contexts.

Jinju Lee is a researcher at the Seoul National University, Management Research Center. She received her PhD in Strategic Management and International Business at Seoul National University. Her research interests include alliance portfolio management, team creativity management, and the collaborative process in the creative industries. Her current research examines how alliance portfolio characteristics affect alliance formations and terminations in the global airline context.

John M. Mezias is an Associate Professor at the University of Miami’s School of Business Administration. His research examines international human resource management, liabilities of foreignness, and legal consequences of strategic actions. He has published in the Strategic Management Journal, Journal of International Business Studies, Harvard Business Review, Organization Science, Journal of Management, Long Range Planning, British Journal of Management, Journal of International Management, Asia Pacific Journal of Management, and Journal of Organizational Behavior. He is Associate Editor of AIB Insights and has long served on the Editorial Boards of the Strategic Management Journal and the Journal of International Business Studies.

References

Abrahamson

(1991) Managerial fads and fashions: The diffusion and rejection of innovations. Academy of Management Review 16(3): 586–612.

Abrahamson

Rosenkopf

(1993) Institutional and competitive bandwagons: Using mathematical modeling as a tool to explore innovation diffusion. Academy of Management Review 18(3): 487–517.

Aldrich

(1979) Organizations and Environments. Englewood Cliffs, NJ: Prentice Hall.

Amaldoss

Staelin

(2010) Cross-function and same-function alliances: How does alliance structure affect the behavior of partnering firms? Management Science 56(2): 302–317.

Amburgey

Miner

(1992) Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity. Strategic Management Journal 13(5): 335–348.

Anand

Khanna

(2000) Do firms learn to create value? The case of alliances. Strategic Management Journal 21(3): 295–315.

Axelrod

(1984) The Evolution of Cooperation. New York: Basic books.

Bailey

Williams

(1988) Sources of economic rent in the deregulated airline industry. Journal of Law and Economics 31: 173–202.

Baum

Calabrese

Silverman

(2000) Don’t go it alone: Alliance network composition and startups’ performance in Canadian biotechnology. Strategic Management Journal 21(3): 267–294.

10.

Baum

Oliver

(1992) Institutional embeddedness and the dynamics of organizational populations. American Sociological Review 57(4): 540–559.

11.

Beck

Brüderl

Woywode

(2008) Momentum or deceleration? Theoretical and methodological reflections on the analysis of organizational change. Academy of Management Journal 51(3): 413–435.

12.

Brander

Zhang

(1993) Dynamic oligopoly behaviour in the airline industry. International Journal of Industrial Organization 11: 407–435.

13.

Cameron

Trivedi

(1990) Regression-based tests for overdispersion in the Poisson model. Journal of Econometrics 46(3): 347–364.

14.

Chen

M-J

Smith

Grimm

(1992) Action characteristics as predictors of competitive responses. Management Science 38(3): 439–455.

15.

Chung

Singh

Lee

(2000) Complementarity, status similarity and social capital as drivers of alliance formation. Strategic Management Journal 21(1): 1–22.

16.

Cohen

Bacdayan

(1994) Organizational routines are stored as procedural memory: Evidence from a laboratory study. Organization Science 5(4): 554–568.

17.

Cohen

Levinthal

(1990) Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly 35(1): 128–152.

18.

Cyert

March

(1963) A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice Hall.

19.

Das

Teng

B-S

(2000) A resource-based theory of strategic alliances. Journal of Management 26(1): 31–61.

20.

Deeds

Hill

(1996) Strategic alliances and the rate of new product development: An empirical study of entrepreneurial biotechnology firms. Journal of Business Venturing 11(1): 41–55.

21.

DiMaggio

Powell

(1983) The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review 48(2): 147–160.

22.

Doganis

(2010) Flying Off Course: Airline Economics and Marketing. New York: Routledge.

23.

Doz

Hamel

(1998) Alliance Advantage: The Art of Creating Value through Partnering. Boston, MA: Harvard Business School Press.

24.

Eisenhardt

Schoonhoven

(1996) Resource-based view of strategic alliance formation: Strategic and social effects in entrepreneurial firms. Organization Science 7(2): 136–150.

25.

Ferrier

Smith

Grimm

(1999) The role of competitive action in market share erosion and industry dethronement: A study of industry leaders and challengers. Academy of Management Journal 42(4): 372–388.

26.

Ghemawat

Porter

Rawlinson

(1986) Patterns of international coalition activity. In: Porter

(ed.) Competition in Global Industries. Boston, MA: Harvard Business School Press, pp. 345–365.

27.

Gomes-Casseres

(1996) The Alliance Revolution: The New Shape of Business Rivalry. Cambridge, MA: Harvard University Press.

28.

Greene

(2002) Econometric Analysis. Upper Saddle River, NJ: Prentice Hall.

29.

Greve

Baum

Mitsuhashi

et al . (2010) Built to last but falling apart: Cohesion, friction, and withdrawal from interfirm alliances. Academy of Management Journal 53(2): 302–322.

30.

Groenewege

(1996) IATA Compendium of International Civil Aviation. Geneva: International Air Transport Association.

31.

Gulati

(1995a) Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal 38(1): 85–112.

32.

Gulati

(1995b) Social structure and alliance formation patterns: A longitudinal analysis. Administrative Science Quarterly 40(4): 619–652.

33.

Gulati

Gargiulo

(1999) Where do interorganizational networks come from? American Journal of Sociology 104(5): 1439–1493.

34.

Hagedoorn

Letterie

Palm

(2011) The information value of R&D alliances: The preference for local or distant ties. Strategic Organization 9(4): 283–309.

35.

Hamel

Doz

Prahalad

(1989) Collaborate with your competitors and win. Harvard Business Review 67(1): 133–139.

36.

Hausman

Hall

Griliches

(1984) Econometric models for count data with an application to the patents-R&D relationship. Econometrica 52(4): 909–938.

37.

Heimeriks

Duysters

Vanhaverbeke

(2007) Learning mechanisms and differential performance in alliance portfolios. Strategic Organization 5(4): 373–408.

38.

Hennart

Reddy

(1997) The choice between mergers/acquisitions and joint ventures: The case of Japanese investors in the United States. Strategic Management Journal 18(1): 1–12.

39.

Hoehn-Weiss

Barden

(2014) The influences of capital market munificence on new-venture alliance formation in emerging industries. Strategic Organization 12(4): 225–243.

40.

Hoehn-Weiss

Karim

(2014) Unpacking functional alliance portfolios: How signals of viability affect young firms’ outcomes. Strategic Management Journal 35(9): 1364–1385.

41.

Holloway

(1997) Straight and Level: Practical Airline Economics. Sydney: Ashgate Publishing.

42.

International Air Transport Association (IATA) (1995) World Air Transport Statistics. Geneva: IATA Economics and Industry Finance Department.

43.

Jiang

Tao

Santoro

(2010) Alliance portfolio diversity and firm performance. Strategic Management Journal 31(10): 1136–1144.

44.

Keeler

Formby

(1994) Cost economies and consolidation in the U.S. airline industry. International Journal of Transport Economics 21: 21–45.

45.

Kefalas

Schoderbek

(1973) Scanning the business environment: Some empirical results. Decision Sciences 4(1): 63–74.

46.

Kelly

Amburgey

(1991) Organizational inertia and momentum: A dynamic model of strategic change. Academy of Management Journal 34(3): 591–612.

47.

Ketokivi

Castaner

(2004) Strategic planning as an integrative device. Administrative Science Quarterly 49(3): 337–365.

48.

Knorr

Arndt

(2004) Alliance strategy and the fall of Swissair: A comment. Journal of Air Transport Management 10(2): 119–123.

49.

Kotha

(2010) Spillovers, spill-ins, and strategic entrepreneurship: America’s first commercial jet airplane and Boeing’s ascendancy in commercial aviation. Strategic Entrepreneurship Journal 4(4): 284–306.

50.

Koza

Lewin

(1998) The co-evolution of strategic alliances. Organization Science 9(3): 255–264.

51.

Kunz

(1999) Airport regulation: The policy framework. In: Pfahler

Niemeier

Mayer

(eds) Airports and Air Traffic: Regulation, Privatization and Competition. New York: Peter Lang, pp. 11–55.

52.

Lavie

Kang

Rosenkopf

(2011) Balance within and across domains: The performance implications of exploration and exploitation in alliances. Organization Science 22(6): 1517–1538.

53.

Lavie

Rosenkopf

(2006) Balancing exploration and exploitation in alliance formation. Academy of Management Journal 49(4): 797–818.

54.

Levinthal

March

(1993) The myopia of learning. Strategic Management Journal 14(S2): 95–112.

55.

Levitt

March

(1988) Organizational learning. Annual Review of Sociology 14: 319–340.

56.

Rowley

(2002) Inertia and evaluation mechanisms in interorganizational partner selection: Syndicate formation among US investment banks. Academy of Management Journal 45(6): 1104–1119.

57.

Madhok

Keyhani

Bossink

(2015) Understanding alliance evolution and termination: Adjustment costs and the economics of resource value. Strategic Organization 13(2): 91–116.

58.

March

(1991) Exploration and exploitation in organizational learning. Organization Science 2(1): 71–87.

59.

March

Olsen

(1989) Rediscovering Institutions: The Organizational Basis of Politics. New York: Free Press.

60.

March

Simon

(1958) Organizations. New York: Wiley.

61.

Miner

Haunschild

(1995) Population-level learning. In: Cumming

Staw

(eds) Research in Organizational Behavior. Greenwich, CT: JAI Press, pp. 115–166.

62.

Nelson

Winter

(1982) An Evolutionary Theory of Economic Change. Cambridge, MA: Harvard University Press.

63.

Oum

Taylor

Zhang

(1993) Strategic airline policy in the globalizing airline networks. Transportation Journal 32(3): 14–32.

64.

Oum

Zhang

(2001) Global airline alliances: International regulatory issues. Journal of Air Transport Management 7(1): 57–62.

65.

Park

Mezias

(2005) Before and after the technology sector crash: The effect of environmental munificence on stock market response to alliances of e-commerce firms. Strategic Management Journal 26(11): 987–1007.

66.

Park

Mezias

Song

(2004) A resource-based view of strategic alliances and firm value in the electronic marketplace. Journal of Management 30(1): 7–27.

67.

Perreault

Leigh

(1989) Reliability of nominal data based on qualitative judgments. Journal of Marketing Research 26(2): 135–148.

68.

Powell

Koput

Smith-Doerr

(1996) Interorganizational collaboration and the locus of innovation: Networks of learning in biotechnology. Administrative Science Quarterly 41: 116–145.

69.

Pustay

(1980) Airline competition and network effects. Transportation Journal 19: 63–72.

70.

Radner

(1993) The organization of decentralized information processing. Econometrica 61: 1109–1146.

71.

Reuer

Koza

(2000) On lemons and indigestibility: Resource assembly through joint ventures. Strategic Management Journal 21(2): 195–197.

72.

Rothaermel

Deeds

(2004) Exploration and exploitation alliances in biotechnology: A system of new product development. Strategic Management Journal 25(3): 201–221.

73.

Schwenk

(1984) Cognitive simplification processes in strategic decision-making. Strategic Management Journal 5(2): 111–128.

74.

Shaver

Mezias

(2009) Diseconomies of managing in acquisitions: Evidence from civil lawsuits. Organization Science 20(1): 206–222.

75.

Shaver

Mitchell

Yeung

(1997) The effect of own-firm and other-firm experience on foreign direct investment survival in the United States, 1987–92. Strategic Management Journal 18(10): 811–824.

76.

Simonin

(1999) Ambiguity and the process of knowledge transfer in strategic alliances. Strategic Management Journal 20(7): 595–623.

77.

Singh

Mitchell

(1996) Precarious collaboration: Business survival after partners shut down or form new partnerships. Strategic Management Journal 17(S1): 99–115.

78.

Turner

Mitchell

Bettis

(2013) Strategic momentum: How experience shapes temporal consistency of ongoing innovation. Journal of Management 39(7): 1855–1890.

79.

Vassolo

Anand

Folta

(2004) Non-additivity in portfolios of exploration activities: A real options-based analysis of equity alliances in biotechnology. Strategic Management Journal 25(11): 1045–1061.

80.

Vasudeva

Anand

(2011) Unpacking absorptive capacity: A study of knowledge utilization from alliance portfolios. Academy of Management Journal 54(3): 611–623.

81.

Vuong

(1989) Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica 57(2): 307–333.

82.

Waller

Huber

Glick

(1995) Functional background as a determinant of executives’ selective perception. Academy of Management Journal 38(4): 943–974.

83.

Wassmer

(2010) Alliance portfolios: A review and research agenda. Journal of Management 36(1): 141–171.

84.

Zander

Kogut

(1995) Knowledge and the speed of the transfer and imitation of organizational capabilities: An empirical test. Organization Science 6(1): 76–92.

85.

Zollo

Reuer

(2010) Experience spillovers across corporate development activities. Organization Science 21(6): 1195–1212.