176 X Jiang et al. /Journal of International Management 14(2008)173-189 based observation approach and cited the high incidence of uV instability (for a review, see Sim and Ali, 2000). They therefore concluded that DVs are fragile organizational forms. Following this view, subsequent research has proposed several important conceptual models in discussing the causes and antecedents of IV instability, e.g., Inkpen and Beamish's(1997) bargaining power and dependence perspective and Yans(1998)structural instability and structural inertia perspective. More recently, some scholars have empirically tested the factors and antecedents proposed by prior studies and offered important evidence for their arguments(e.g, Sim and All, 2000; Nakamura, 2005) 2.1.2. Limitations in instability research Despite the rich literature on instability, previous studies have limitations(Yan and Zeng, 1999; Sim and Ali, 2000) First, several streams of research have examined alliance instability through a variety of theoretical perspectives and approaches(for a review, see Das and Teng, 2000), but no single theoretical approach can offer comprehensive insights nto its nature. The incompatible perspectives have also produced different and even contradictory findings, leading to the result that relevant research findings are not cumulative and comparable. Second, these studies have the problem of non-generalizability. For example, most research focuses on the instability of [JVs, but these research findings are not necessarily the same as those associated with other types of alliances. Third, while researchers have identified various kinds of destabilizing factors, it is unclear how these factors emerge and how they affect instability There are two distinct approaches for conceptualizing and operationalizing stability in the extant literature(Yan and Zeng, 1999; Gill and Butler, 2003). One is a static and outcome-oriented approach whose focus is on the final fates of an alliance, including the alliance being bought out, turning into an acquisition, or being liquidated. The other is a dynamic and process-oriented approach, in which instability is defined as changes in ownership structure in terms of ajor reorganizations and reconfigurations or contractual renegotiations. Since instability itself is a dynamic process variable, the more process-oriented perspective provides a better explanation for alliance instability. Unfortunately, prior research, with few exceptions (e. g, Killing, 1983; Yan and Zeng, 1999), has been dominated by the outcome-oriented approach. Das and Teng(2000: 96), who have conducted some of the most comprehensive and representative research so far, also acknowledge that their research only links alliance outcomes with imbalances of three pairs of competing forces and"does not clearly depict the evolutionary process by which these forces may develop". As a result, the process aspect and dynamic nature of instability are largely ignored, incurring serious methodological barriers and understanding biases 2.2. Previous research on alliance stabilin Research on the topic of stability is relatively scarce as opposed to the rich instability literature. Table I also lists prior representative studies on alliance stability, research contributions which also have limitations. For example, many studies do not conceptually differentiate instability from stability. Others(e.g, Yan, 1998; Yan and Zeng, 1999) parate the two but do not offer a reasonable theoretical rationale for this separation. In particular, while some of the udies nominally address the stability issue, they instead center most attention on the inverse side, i.e., alliance instability, or mix the two concepts in one study. Moreover, many studies deal with the stability of some particular type of alliances(typically JV or IJV stability), leading to the lack of consensus on the relative stability of different types of alliances. As a result, the conceptual blurring and inconsistency in research results increase the difficulty in understanding the dynamics within alliances. 3. Toward a comprehensive understanding of alliance stability 3. 1. Explaining the importance of stability research 3.1.1. Necessity of perspective conversion In contrast to prior relevant research, in this study we differentiate instability from stability Instability is a dynamic concept in itself, but research has traditionally taken an outcome-oriented approach and treated it as synonymous with the alliance's termination, death or failure(Franko, 1971; Killing, 1983; Kogut, 1989). In fact, however, an alliances termination does not necessarily imply instability. Some alliances are terminated according to plan(Inkpen and Beamish 1997), and others go out of existence because they have successfully accomplished the initial objectives( Gomes-Casseres 1987). Both of these possibilities indicate that not all terminated or short-life alliances should be considered unstablebased observation approach and cited the high incidence of IJV instability (for a review, see Sim and Ali, 2000). They therefore concluded that IJVs are fragile organizational forms. Following this view, subsequent research has proposed several important conceptual models in discussing the causes and antecedents of IJV instability, e.g., Inkpen and Beamish's (1997) bargaining power and dependence perspective and Yan's (1998) structural instability and structural inertia perspective. More recently, some scholars have empirically tested the factors and antecedents proposed by prior studies and offered important evidence for their arguments (e.g., Sim and Ali, 2000; Nakamura, 2005). 2.1.2. Limitations in instability research Despite the rich literature on instability, previous studies have limitations (Yan and Zeng, 1999; Sim and Ali, 2000). First, several streams of research have examined alliance instability through a variety of theoretical perspectives and approaches (for a review, see Das and Teng, 2000), but no single theoretical approach can offer comprehensive insights into its nature. The incompatible perspectives have also produced different and even contradictory findings, leading to the result that relevant research findings are not cumulative and comparable. Second, these studies have the problem of non-generalizability. For example, most research focuses on the instability of IJVs, but these research findings are not necessarily the same as those associated with other types of alliances. Third, while researchers have identified various kinds of destabilizing factors, it is unclear how these factors emerge and how they affect instability. There are two distinct approaches for conceptualizing and operationalizing stability in the extant literature (Yan and Zeng, 1999; Gill and Butler, 2003). One is a static and outcome-oriented approach whose focus is on the final fates of an alliance, including the alliance being bought out, turning into an acquisition, or being liquidated. The other is a dynamic and process-oriented approach, in which instability is defined as changes in ownership structure in terms of major reorganizations and reconfigurations or contractual renegotiations. Since instability itself is a dynamic process variable, the more process-oriented perspective provides a better explanation for alliance instability. Unfortunately, prior research, with few exceptions (e.g., Killing, 1983; Yan and Zeng, 1999), has been dominated by the outcome-oriented approach. Das and Teng (2000: 96), who have conducted some of the most comprehensive and representative research so far, also acknowledge that their research only links alliance outcomes with imbalances of three pairs of competing forces and “does not clearly depict the evolutionary process by which these forces may develop”. As a result, the process aspect and dynamic nature of instability are largely ignored, incurring serious methodological barriers and understanding biases. 2.2. Previous research on alliance stability Research on the topic of stability is relatively scarce as opposed to the rich instability literature. Table 1 also lists prior representative studies on alliance stability, research contributions which also have limitations. For example, many studies do not conceptually differentiate instability from stability. Others (e.g., Yan, 1998; Yan and Zeng, 1999) separate the two but do not offer a reasonable theoretical rationale for this separation. In particular, while some of the studies nominally address the stability issue, they instead center most attention on the inverse side, i.e., alliance instability, or mix the two concepts in one study. Moreover, many studies deal with the stability of some particular type of alliances (typically JV or IJV stability), leading to the lack of consensus on the relative stability of different types of alliances. As a result, the conceptual blurring and inconsistency in research results increase the difficulty in understanding the dynamics within alliances. 3. Toward a comprehensive understanding of alliance stability 3.1. Explaining the importance of stability research 3.1.1. Necessity of perspective conversion In contrast to prior relevant research, in this study we differentiate instability from stability. Instability is a dynamic concept in itself, but research has traditionally taken an outcome-oriented approach and treated it as synonymous with the alliance's termination, death or failure (Franko, 1971; Killing, 1983; Kogut, 1989). In fact, however, an alliance's termination does not necessarily imply instability. Some alliances are terminated according to plan (Inkpen and Beamish, 1997), and others go out of existence because they have successfully accomplished the initial objectives (Gomes-Casseres, 1987). Both of these possibilities indicate that not all terminated or short-life alliances should be considered unstable. 176 X. Jiang et al. / Journal of International Management 14 (2008) 173–189