X Jiang et al. /Journal of International Management 14(2008)173-189 Furthermore, past studies have often cited high instability rates (or more precisely, a high termination ratio) and therefore have viewed alliances as inherently fragile and unstable. But this view has been challenged by other scholars Yan(1998), for instance, argued that such a view fails to explain why many alliances have operated stably over a long eriod of time. Hennart et al. (1998)found, after controlling for age and size, that the instability of JVs is similar to that of wholly-owned subsidiaries. In a recent large sample analysis (27, 974 Japanese foreign subsidiaries), Delios and Beamish(2004) similarly found evidence that JV exit rates(or survival rates) were almost identical to those of wholly owned subsidiaries. Thus, strategic alliances should not be casually treated as intrinsically fragile or unstable(Doz, 1996: Koza and Lewin, 1998). Because of these concems, this paper adopts a different perspective and views stability as a process-based construct This perspective requires an analysis of the evolutionary process of alliance development, so that we can develop a comprehensive framework of alliance stability. Further, we view stability as a determinant of alliance performance and success, rather than as an ultimate outcome emphasized by prior studies. Below, we will discuss these in greater detail 3.l.2.St We first discuss why stability is an important construct in different theories used in the alliance literature. Many theories have been used to address alliance stability issues. Among these, the competitive leaning perspective is a powerful tool(Inkpen and Beamish, 1997; Nakamura, 2005). It argues that learning and acquisition of knowledge over time may alter the relative bargaining power of the partners and thus seriously undermine the initial balance of collaborative relationships. Hamel(1991)describes alliances as a"race to learn, and the partner who first successfully accomplishes its learning objectives will tend to leave the alliance. For example, the competitive learning between Chrysler and Mitsubishi leads to the termination of their Diamond Star alliance in 1991, with Mitsubishi taking over the jointly owned automobile assembly plant. The more speedy and efficient learning enabled the Japanese partner to successfully develop its own distribution network in North America, while Chrysler failed to acquire the manufacturing technologies from Mitsubishi. This exerted a negative influence on the stability of the Diamond Star alliance. In this nstance, competitive learning becomes a critical source of instability. We dont deny that collaboration, while initially harmonious, will gradually strain as a result of such considerations as the race to learn. But we contend that before the partners have achieved their learning goals, the alliances must go through a relatively stable( developmental) period so that the learning can be carried out. Inkpen and Beamish(1997) posit that a partner who is learning from the knowledge provider will prefer to maintain rather than terminate the relationship. Toppan Moore, a JV between Canadas Moore Corporation and Japan's Toppan Printing Company, was widely considered as one of the most successful [Vs in Japan. Despite many years of accumulated experience in the local market, Moore Corporation continued to cooperate closely with its Japanese partner in order to gain more specialized knowledge about the market and the customers. Likewise, Toppan Printing relied on Moore Corporation in order to obtain advanced manufacturing and product technology. Thus the partners'realization that they have much to learn from each other has made a major contribution to the stability of the venture(Beamish and Inkpen, 1995) We also contend that, in most alliances, there are stabilizing and sustaining forces which may prevent instability and shelter alliance relationships from unplanned changes. Specifically, forming an alliance takes much time and effort. Substantial non-recoverable, relationship-specific investments must be put into the alliance, and these investments usually cannot be recovered for other uses(Anderson and Weitz, 1992; Parkhe, 1993). The efforts, time and resources involved in an alliance incur high exit costs and create inertia towards leaving the alliance( Ford et al, 1998). Even leaving is often done in a gradual way(Gadde and Mattsson, 1987) Furthermore, joint contributions and reciprocal commitments create attachments among partners. Attachment refers to"an inertial or binding force between exchange partners that can lead to the maintenance of an existing relationship to the exclusion of altematives"(Seabright et al, 1992: 126). By definition, partners with strong attachments may feel obligated to enhance the joint strength and maintain a high degree of"couplings", which may act as a cohesive force that fosters confidence, reciprocation and intimate cooperation (Luo, 2005). As a result, attachment serves as a stabilizing force that blocks the pressures for change in each cooperative relationship( Seabright et al, 1992) 3.1.3. Relationship to alliance outcomes A Alliance outcomes have often been evaluated in terms of alliance success or failure. Despite a significant amount of earch, however, alliance success remains a limitedly studied area, and thus far there are no all-accepted criteria and measures for alliance success. In the extant literature. a conventional measure is to examine whether an alliance hasFurthermore, past studies have often cited high instability rates (or more precisely, a high termination ratio) and therefore have viewed alliances as inherently fragile and unstable. But this view has been challenged by other scholars. Yan (1998), for instance, argued that such a view fails to explain why many alliances have operated stably over a long period of time. Hennart et al. (1998) found, after controlling for age and size, that the instability of JVs is similar to that of wholly-owned subsidiaries. In a recent large sample analysis (27,974 Japanese foreign subsidiaries), Delios and Beamish (2004) similarly found evidence that JV exit rates (or survival rates) were almost identical to those of wholly￾owned subsidiaries. Thus, strategic alliances should not be casually treated as intrinsically fragile or unstable (Doz, 1996; Koza and Lewin, 1998). Because of these concerns, this paper adopts a different perspective and views stability as a process-based construct. This perspective requires an analysis of the evolutionary process of alliance development, so that we can develop a comprehensive framework of alliance stability. Further, we view stability as a determinant of alliance performance and success, rather than as an ultimate outcome emphasized by prior studies. Below, we will discuss these in greater detail. 3.1.2. Stabilizing forces We first discuss why stability is an important construct in different theories used in the alliance literature. Many theories have been used to address alliance stability issues. Among these, the competitive learning perspective is a powerful tool (Inkpen and Beamish, 1997; Nakamura, 2005). It argues that learning and acquisition of knowledge over time may alter the relative bargaining power of the partners and thus seriously undermine the initial balance of collaborative relationships. Hamel (1991) describes alliances as a “race to learn”, and the partner who first successfully accomplishes its learning objectives will tend to leave the alliance. For example, the competitive learning between Chrysler and Mitsubishi leads to the termination of their Diamond Star alliance in 1991, with Mitsubishi taking over the jointly owned automobile assembly plant. The more speedy and efficient learning enabled the Japanese partner to successfully develop its own distribution network in North America, while Chrysler failed to acquire the manufacturing technologies from Mitsubishi. This exerted a negative influence on the stability of the Diamond Star alliance. In this instance, competitive learning becomes a critical source of instability. We don't deny that collaboration, while initially harmonious, will gradually strain as a result of such considerations as the race to learn. But we contend that before the partners have achieved their learning goals, the alliances must go through a relatively stable (developmental) period so that the learning can be carried out. Inkpen and Beamish (1997) posit that a partner who is learning from the knowledge provider will prefer to maintain rather than terminate the relationship. Toppan Moore, a JV between Canada's Moore Corporation and Japan's Toppan Printing Company, was widely considered as one of the most successful IJVs in Japan. Despite many years of accumulated experience in the local market, Moore Corporation continued to cooperate closely with its Japanese partner in order to gain more specialized knowledge about the market and the customers. Likewise, Toppan Printing relied on Moore Corporation in order to obtain advanced manufacturing and product technology. Thus the partners' realization that they have much to learn from each other has made a major contribution to the stability of the venture (Beamish and Inkpen, 1995). We also contend that, in most alliances, there are stabilizing and sustaining forces which may prevent instability and shelter alliance relationships from unplanned changes. Specifically, forming an alliance takes much time and effort. Substantial non-recoverable, relationship-specific investments must be put into the alliance, and these investments usually cannot be recovered for other uses (Anderson and Weitz, 1992; Parkhe, 1993). The efforts, time and resources involved in an alliance incur high exit costs and create inertia towards leaving the alliance (Ford et al., 1998). Even leaving is often done in a gradual way (Gadde and Mattsson, 1987). Furthermore, joint contributions and reciprocal commitments create attachments among partners. Attachment refers to “an inertial or binding force between exchange partners that can lead to the maintenance of an existing relationship to the exclusion of alternatives” (Seabright et al., 1992: 126). By definition, partners with strong attachments may feel obligated to enhance the joint strength and maintain a high degree of “couplings”, which may act as a cohesive force that fosters confidence, reciprocation and intimate cooperation (Luo, 2005). As a result, attachment serves as a stabilizing force that blocks the pressures for change in each cooperative relationship (Seabright et al., 1992). 3.1.3. Relationship to alliance outcomes Alliance outcomes have often been evaluated in terms of alliance success or failure. Despite a significant amount of research, however, alliance success remains a limitedly studied area, and thus far there are no all-accepted criteria and measures for alliance success. In the extant literature, a conventional measure is to examine whether an alliance has X. Jiang et al. / Journal of International Management 14 (2008) 173–189 177