THE ROLE OF ATTRIBUTION IN LEARNING FROM PERFORMANCE FEEDBACK: BEHAVIORAL PERSPECTIVE ON THE CHOICE BETWEEN ALLIANCES AND ACQUISITIONS
归因在从绩效反馈中学习的作用：关于联盟与收购之间选择的行为视角

Performance feedback theory postulates that past performance is an important source of organizational learning (Cyert & March, 1963; Greve, 1998). Central to performance feedback theory is the idea that a firm’s past performance relative to its aspiration level guides its future behavior and decisions (Baum, Rowley, Shipilov, & Chuang, 2005; Greve, 2003a; Kacperczyk, Beckman, & Moliterno, 2015; Miller & Chen, 2004). Much scholarly attention has been given to this proposition, significantly advancing our understanding of how performance shapes firm behavior and decisions (for a review, see Gavetti, Greve, Levinthal, & Ocasio, 2012; Greve & Gaba, 2017; Posen, Keil, Kim, & Meissner, 2018).
绩效反馈理论认为，过去的绩效是组织学习的重要来源（Cyert & March, 1963; Greve, 1998）。该理论的核心思想是，企业过去的绩效相对于其期望水平会指导其未来的行为和决策（Baum, Rowley, Shipilov, & Chuang, 2005; Greve, 2003a; Kacperczyk, Beckman, & Moliterno, 2015; Miller & Chen, 2004）。这一命题得到了大量学术关注，极大地推进了我们对绩效如何影响企业行为和决策的理解（综述见 Gavetti, Greve, Levinthal, & Ocasio, 2012; Greve & Gaba, 2017; Posen, Keil, Kim, & Meissner, 2018）。

To learn from performance feedback, firms should first build a causal model by identifying strategies and routines that have determined prior performance. Such a model will guide them to change the strategies and routines that have been attributed to failure and retain, or reinforce, those that have been attributed to success. Thus, causal attribution for outcomes is an essential step for learning from performance feedback. However, attribution has not been explicitly discussed in the performance feedback literature; consequently, the literature has developed an incomplete view on the process of learning from performance feedback. This research gap is surprising considering the important role of attribution in learning from performance feedback, and the increasing scholarly attention to psychological factors influencing how firms evaluate and respond to performance (Audia & Greve, 2021; Desai, 2015; Eggers & Suh, 2019). Indeed, Posen et al. (2018: 212), in their literature review, called for the integration of the attribution process with performance feedback research, describing existing studies as projecting “a rather mechanistic, nearly automatic, view of search triggered when performance falls below the threshold defined by the aspiration.”
从绩效反馈中学习，企业首先应通过识别决定先前绩效的策略和惯例来构建因果模型。此类模型将指导他们改变那些归因于失败的策略和惯例，并保留或强化那些归因于成功的策略和惯例。因此，对结果进行因果归因是从绩效反馈中学习的关键步骤。然而，绩效反馈文献中并未明确讨论归因问题；因此，该文献对从绩效反馈中学习的过程形成了不完整的看法。考虑到归因在从绩效反馈中学习的重要作用，以及学术界对影响企业评估和应对绩效的心理因素日益关注（Audia & Greve, 2021; Desai, 2015; Eggers & Suh, 2019），这一研究空白令人惊讶。确实，Posen 等人。 (2018: 212)在其文献综述中呼吁将归因过程与绩效反馈研究相结合，并将现有研究描述为“投射了一种相当机械的、近乎自动的搜索视角，这种搜索在绩效低于期望定义的阈值时被触发。”

We argue that an explicit consideration of attribution can significantly revise and improve the core predictions of performance feedback theory. For example, if a firm attributes a failure to exogenous factors over which it has little control (e.g., unexpected random events), it may persist with its current strategy even after the failure, concluding that its current strategy is not responsible for the failure, and thus does not need to be replaced. It is, therefore, critical to understand the role of attribution in learning from performance feedback to draw a precise picture of how performance feedback influences firms’ search behavior and decisions to change.
我们认为，明确考虑归因可以显著修正并提升绩效反馈理论的核心预测。例如，如果一家公司将失败归因于其几乎无法控制的外生因素（如意外随机事件），它可能会在失败后继续坚持当前策略，认为当前策略并非失败的原因，因此无需更换。因此，理解归因在从绩效反馈中学习的作用，对于精确描绘绩效反馈如何影响企业的搜索行为和变革决策至关重要。

To this end, we develop a theory of attribution in learning from performance feedback by integrating attribution and performance feedback theory. We specifically explore how different patterns of attribution of failure and success influence a firm’s response to past performance. To detect attribution mechanisms in learning from performance feedback, we identify three theoretical factors that can significantly influence attribution in performance evaluation and decision-making (the diffusion of responsibility, the perceived capability of partners, and the ambiguity of performance information), and examine how they moderate the effect of performance feedback on a firm’s strategic decision-making.
为此，我们通过整合归因理论与绩效反馈理论，发展了一种从绩效反馈中学习的归因理论。我们特别探讨了失败与成功归因的不同模式如何影响企业对过往绩效的响应。为了检测从绩效反馈中学习的归因机制，我们识别了三个能显著影响绩效评估与决策中归因的理论因素（责任分散、合作伙伴的感知能力以及绩效信息的模糊性），并考察了它们如何调节绩效反馈对企业战略决策的影响。

We test our theory by exploring how firms’ prior alliance performance influences the choice between acquisitions and alliances-the two most often-used alternative governance modes of sourcing external knowledge and resources (Folta, 1998; Villalonga & McGahan, 2005; Wang & Zajac, 2007). Using data on alliances and acquisitions completed by Fortune 500 companies between 1994 and 2017, we find strong support for our theory and demonstrate that our findings are robust across several alternative econometric specifications and samples.
我们通过探究企业既往联盟绩效如何影响收购与联盟之间的选择，来检验我们的理论——这两种是最常用的获取外部知识和资源的替代治理模式（Folta, 1998; Villalonga & McGahan, 2005; Wang & Zajac, 2007）。利用 1994 年至 2017 年间《财富》500 强公司完成的联盟与收购数据，我们找到了强有力的理论支持，并证明我们的发现在多种替代性计量经济学设定和样本中均具稳健性。

Our contributions are threefold. First, we advance research on performance feedback by developing a theory of attribution in learning from performance feedback. To the best of our knowledge, this is the first study on performance feedback that explicitly considers how attribution can influence a firm’s response to performance feedback (Greve & Gaba, 2017; Posen et al., 2018). Our theory and results demonstrate that a firm’s response to prior performance is significantly influenced by attribution, highlighting the importance of integrating attribution with performance feedback research. Second, by identifying three moderating factors that influence performance attribution, we uncover theoretical mechanisms by which attribution operates in learning from performance feedback. Taken
我们的贡献有三方面。首先，我们通过发展一种关于从绩效反馈中学习的归因理论，推进了绩效反馈的研究。据我们所知，这是首次明确考虑归因如何影响企业对绩效反馈反应的研究（Greve & Gaba, 2017; Posen et al., 2018）。我们的理论和结果表明，企业对先前绩效的反应显著受到归因的影响，强调了将归因与绩效反馈研究相结合的重要性。其次，通过识别影响绩效归因的三个调节因素，我们揭示了归因在从绩效反馈中学习过程中运作的理论机制。
together, our findings not only provide evidence that attribution plays a significant role in firms’ responses to performance feedback but also demonstrate theoretical mechanisms that strengthen or attenuate the effect of attribution in learning from performance feedback. Finally, extant literature has taken economic and agency perspectives in explaining the choice between alliances and acquisitions (e.g., Reuer, Tong, Tyler, & Ariño, 2013; Villalonga & McGahan, 2005; Wang & Zajac, 2007). Being the first to apply the behavioral perspective to examine this choice, our study shows that a firm’s governance choices can be influenced by heuristics and biases, such as self-serving attribution.
我们的研究结果不仅证明了归因在企业响应绩效反馈中扮演重要角色，还展示了加强或减弱归因在从绩效反馈中学习效果的理论机制。最后，现有文献多从经济和代理视角解释企业在联盟与收购之间的选择（例如，Reuer, Tong, Tyler, & Ariño, 2013; Villalonga & McGahan, 2005; Wang & Zajac, 2007）。作为首次应用行为视角来考察这一选择的研究，我们的研究表明，企业的治理决策可能受到启发式思维和偏见的影响，如自利性归因。

Performance feedback theory suggests that learning from performance feedback is a four-step process comprising (a) performance evaluation, (b) attribution, © search, and (d) risk-taking and change (Cyert & March, 1963; Greve, 2003b; Posen et al., 2018; cf., Jordan & Audia, 2012).
绩效反馈理论认为，从绩效反馈中学习是一个包含四个步骤的过程，即(a)绩效评估，(b)归因，(c)搜索，以及(d)风险承担与变革（Cyert & March, 1963; Greve, 2003b; Posen 等，2018; 参见 Jordan & Audia, 2012）。

In the first step, managers evaluate performance. Due to the bounded rationality and the complexity of performance assessment (Cyert & March, 1963; Obloj & Sengul, 2020; Zollo, 2009), they simplify performance evaluation by dichotomizing past performance into success and failure by comparing it with the aspiration level (Cyert & March, 1963). Performance that falls short of the aspiration level is considered a failure, while performance that exceeds it is considered a success. In the second step, managers seek to identify factors deemed to have contributed to the success or failure. Based on the outcome of this attribution, they build or update the causal model, specifying the causes of the success or failure, which they then use to make future decisions in the following steps. The strategies and actions associated with success will be reinforced, while those associated with failure will be scrutinized.
在第一步中，管理者评估绩效。由于有限理性和绩效评估的复杂性（Cyert & March, 1963; Obloj & Sengul, 2020; Zollo, 2009），他们通过将过去的表现与期望水平进行比较，将其简化为成功与失败（Cyert & March, 1963）。未达到期望水平的绩效被视为失败，而超过期望水平的绩效则被视为成功。在第二步中，管理者试图识别被认为导致成功或失败的因素。基于这种归因的结果，他们建立或更新因果模型，明确成功或失败的原因，并在后续步骤中用于未来决策。与成功相关的策略和行动将得到加强，而与失败相关的策略和行动将受到审查。
In the third step, different types of organizational search are triggered depending on whether a firm experiences success or failure. When a firm experiences failure, managers engage in a problemistic search to find alternatives to strategies and actions thought to have caused the failure, and they become more willing to take greater risks-opening a path to the fourth step of organizational change. Success may also lead to organizational change through an accumulation of a firm’s slack resources, which may promote slack search (cf., Xu, Zhou, & Du, 2019).
在第三步中，根据企业经历成功或失败的不同，会触发不同类型的组织搜索。当企业遭遇失败时，管理者会进行问题性搜索，以寻找被认为导致失败的策略和行动的替代方案，并且他们更愿意承担更大的风险——这为组织变革的第四步开辟了道路。成功也可能通过企业冗余资源的积累导致组织变革，这可能会促进冗余搜索（参见 Xu, Zhou, & Du, 2019）。

This process of learning from performance feedback has received broad empirical support and generated valuable insights on how performance feedback shapes firm behavior (Gavetti et al., 2012; Greve & Gaba, 2017; Posen et al., 2018). However, existing research has largely focused on the third and fourth steps (search behavior and organizational change), giving scant attention to the second step (attribution) (Posen et al., 2018). Without explicitly considering different patterns of attribution, prior studies have often assumed that managers attribute observed performance outcomes, primarily, to their own strategies and capability. Haleblian, Kim, and Rajagopalan (2006), for example, argued that managers will perceive poor acquisition performance as a signal that their firm does not have what it takes to make successful acquisitions, thereby decreasing its likelihood of making future acquisitions. However, this prediction may need to be revised if managers attribute acquisition failure to external causes (e.g., a one-time black swan event). Such attribution will not discourage them from making acquisitions as much as when failure is attributed to internal causes (e.g., inferior resources and capabilities).
从绩效反馈中学习的过程已获得广泛的实证支持，并产生了关于绩效反馈如何塑造企业行为的宝贵见解（Gavetti 等，2012；Greve & Gaba，2017；Posen 等，2018）。然而，现有研究主要集中在第三和第四步（搜索行为和组织变革），对第二步（归因）关注较少（Posen 等，2018）。在没有明确考虑不同归因模式的情况下，先前的研究往往假设管理者将观察到的绩效结果主要归因于他们自己的战略和能力。例如，Haleblian、Kim 和 Rajagopalan（2006）认为，管理者会将收购表现不佳视为公司缺乏成功收购所需条件的信号，从而降低未来进行收购的可能性。然而，如果管理者将收购失败归因于外部原因（例如，一次性的黑天鹅事件），这一预测可能需要修正。 这种归因不会像将失败归因于内部原因（例如，资源不足和能力欠缺）那样，阻碍他们进行收购。

Identifying the true causes of performance outcomes is an arduous task due to the inherent difficulty of making causal attributions (Zollo, 2009), and this difficulty is exacerbated when firms attempt to establish a causal relationship from rare, complicated events that involve multiple organizations, such as major accidents and alliances (March, Sproull, & Tamuz, 1991). Thus, learning from such events can be susceptible to cognitive biases, such as selfserving attribution. Indeed, a large body of attribution research in social psychology (e.g., Burger, 1981; Mezulis, Abramson, Hyde, & Hankin, 2004) and management (e.g., Clapham & Schwenk, 1991; Ford, 1985; Salancik & Meindl, 1984; Wagner & Gooding, 1997) has suggested that managers tend to attribute poor performance to the causes external to themselves and their firm, such as a deteriorating industry or economic conditions; by contrast, they tend to attribute success to internal causes and take the credit.
识别绩效结果的真实原因是一项艰巨的任务，这归因于因果归因本身固有的困难（Zollo, 2009），而当企业试图从涉及多个组织的罕见、复杂事件（如重大事故和联盟）中建立因果关系时，这种困难会进一步加剧（March, Sproull, & Tamuz, 1991）。因此，从这类事件中学习可能容易受到认知偏差的影响，例如自我服务归因。事实上，社会心理学（如 Burger, 1981; Mezulis, Abramson, Hyde, & Hankin, 2004）和管理学（如 Clapham & Schwenk, 1991; Ford, 1985; Salancik & Meindl, 1984; Wagner & Gooding, 1997）中的大量归因研究表明，管理者倾向于将糟糕的绩效归因于自身及其公司之外的因素，如行业或经济状况的恶化；相比之下，他们倾向于将成功归因于内部因素并居功。

Research in attribution theory has posited that self-serving attribution can arise from both cognitive and motivational factors (e.g., Sedikides & Alicke, 2012; Shepperd, Malone, & Sweeny, 2008). Like other cognitive biases (e.g., confirmation bias or the anchoring effect), self-serving attribution is often viewed as information-processing rules or heuristics. When attributing performance outcomes, managers tend to rely on the heuristics and schema that arise from their backgrounds, experiences, beliefs,
归因理论的研究表明，自我服务归因可能源于认知和动机因素（例如，Sedikides & Alicke, 2012; Shepperd, Malone, & Sweeny, 2008）。与其他认知偏差（例如，确认偏误或锚定效应）类似，自我服务归因通常被视为信息处理规则或启发法。在归因绩效结果时，管理者倾向于依赖其背景、经验、信仰所产生的启发法和图式。
and routines, which can inadvertently lead them to make self-serving attributions in evaluating available information about the observed performance (e.g., Gilovich, 1991; Tetlock & Levi, 1982). For example, managers cannot fully observe others’ efforts toward a common task, while they are fully aware of their own and thus may develop a view that others’ contributions toward the task were not as important or as genuine as their own (Miller & Ross, 1975). They may also hold a positive view of their abilities and expect success more frequently than failure, considering success as a “normal” outcome that can be naturally attributed to their abilities while considering failure as an “abnormal” outcome, which can only be explained by exogenous factors (Mezulis et al., 2004). This view holds that self-serving attribution is a heuristic, nonmotivational process.
以及常规做法，这可能会无意中导致他们在评估关于观察到的表现的可用信息时做出自我服务的归因（例如，Gilovich, 1991; Tetlock & Levi, 1982）。例如，管理者无法完全观察到他人对共同任务的努力，而他们完全清楚自己的努力，因此可能会形成一种观点，认为他人对任务的贡献不如自己的重要或真诚（Miller & Ross, 1975）。他们也可能对自己的能力持积极看法，并期望成功比失败更频繁，将成功视为“正常”结果，可以自然地归因于他们的能力，而将失败视为“异常”结果，只能由外部因素解释（Mezulis et al., 2004）。这种观点认为，自我服务的归因是一种启发式的、非动机性的过程。
However, self-serving attribution has also been portrayed as motivational bias, which refers to a conscious or unconscious distortion driven by motivations or incentives (e.g., Campbell & Sedikides, 1999; Crocker, Voelkl, Testa, & Major, 1991). For example, motivated by the need to salvage their reputation, managers may attribute failure to others; or, driven by the desire to enhance their self-image and status, they may attribute success to themselves by taking more than their fair share of credit (e.g., Campbell & Sedikides, 1999). Consistent with this view, recent studies in management have suggested that performance evaluation and interpretation may be influenced by psychological processes akin to attribution, such as managers’ desire to enhance their self-esteem and their impression on key stakeholders (e.g., Audia & Greve, 2021; Fang, Kim, & Milliken, 2014; Harris & Bromiley, 2007).
然而，自我服务归因也被描述为动机性偏见，指的是由动机或激励驱动的有意识或无意识的扭曲（例如，Campbell & Sedikides, 1999; Crocker, Voelkl, Testa, & Major, 1991）。例如，出于挽回声誉的需要，管理者可能会将失败归咎于他人；或者，受提升自我形象和地位的欲望驱使，他们可能会将成功归功于自己，从而获得超出其应得的赞誉（例如，Campbell & Sedikides, 1999）。与此观点一致，最近的管理学研究表明，绩效评估和解释可能受到类似于归因的心理过程的影响，例如管理者提升自尊心及其对关键利益相关者印象的愿望（例如，Audia & Greve, 2021; Fang, Kim, & Milliken, 2014; Harris & Bromiley, 2007）。
These two vantage points have been amalgamated to form a perspective that self-serving attribution is a process in which cognitive and motivational processes work in tandem (e.g., Kunda, 1990; Sedikides & Alicke, 2012; Shepperd et al., 2008). In this study, we subscribe to this perspective and consider self-serving attribution as primarily driven by judgmental heuristics (just like most other cognitive biases), but which can be further distorted by motivational factors.
这两个观点被结合起来，形成了一种视角，即自我服务归因是一个认知和动机过程共同作用的过程（例如，Kunda, 1990; Sedikides & Alicke, 2012; Shepperd et al., 2008）。在本研究中，我们采纳这一视角，认为自我服务归因主要由判断启发法驱动（就像大多数其他认知偏差一样），但可能会被动机因素进一步扭曲。

We tackle our research questions by examining how the attribution of alliance performance influences the subsequent choice between alliance and acquisition. Alliances and acquisitions are the two most popular governance modes for sourcing external assets, knowledge, and capabilities that cannot be easily acquired
我们通过研究联盟绩效归因如何影响后续在联盟与收购之间的选择来探讨研究问题。联盟和收购是获取外部资产、知识和能力的两种最流行的治理模式，这些资源和能力难以轻易获得。
or purchased (Folta, 1998; Villalonga & McGahan, 2005; Wang & Zajac, 2007). They are not direct or perfect substitutes for each other, nor are they equally effective in accessing the same resources (Capron & Mitchell, 2012). However, because there is much overlap in resources required by acquisitions and alliances, such as top management attention, it is likely that the choices between acquisitions and alliances are intertwined. Indeed, both academic and practical research has extensively documented that they are strategic alternatives for achieving similar objectives, and that firms typically consider them when the need for sourcing external resources and knowledge arises (Capron & Mitchell, 2012; Folta, 1998; Reuer & Ragozzino, 2012; Villalonga & McGahan, 2005).
或购买（Folta, 1998; Villalonga & McGahan, 2005; Wang & Zajac, 2007）。它们并非彼此的直接或完美替代品，在获取相同资源方面也不具有同等效力（Capron & Mitchell, 2012）。然而，由于收购和联盟所需的资源存在大量重叠，例如高层管理者的关注，因此收购与联盟之间的选择很可能是相互交织的。事实上，学术和实务研究均广泛记载了它们是实现相似目标的战略替代方案，且企业通常在需要获取外部资源和知识时会考虑它们（Capron & Mitchell, 2012; Folta, 1998; Reuer & Ragozzino, 2012; Villalonga & McGahan, 2005）。

Performance feedback research has been previously conducted in the context of acquisitions and alliances, but most of these studies have focused on how a firm’s overall performance (e.g., the return on assets) leads to its decision to make acquisitions or alliances (Baum et al., 2005; Iyer & Miller, 2008; Kavusan & Frankort, 2019). While these studies have generated valuable insights, firm-level performance may not be an appropriate construct for studying attribution; due to the confounding influences of many strategies, actions, and events on firm-level performance, it is difficult to attribute firm-level performance to a specific action.
绩效反馈研究先前已在收购和联盟的背景下进行，但大多数研究都集中在公司整体绩效（如资产回报率）如何导致其做出收购或联盟决策上（Baum 等，2005；Iyer & Miller，2008；Kavusan & Frankort，2019）。尽管这些研究产生了有价值的见解，但公司层面的绩效可能不是研究归因的适当构建；由于许多策略、行动和事件对公司层面绩效的混杂影响，很难将公司层面的绩效归因于特定行动。

To study attribution, we need to establish a tight causal relationship between an action and the performance of that action. A handful of recent studies on performance feedback have begun to examine the effect of a performance that is specific to an action (Eggers & Suh, 2019; Maslach, 2016), including two acquisition studies that examined the relationship between the performance specific to acquisition and future acquisition likelihood (Haleblian et al., 2006; Kim, Finkelstein, & Haleblian, 2015). However, no study has probed how alliance performance influences future alliance likelihood. Further, there has been little research on whether the performance of prior alliances influences the likelihood of acquisitions or vice versa. Addressing these empirical gaps, we study how alliance performance influences future decisions on acquisitions versus alliances.
为了研究归因，我们需要在行动与该行动的表现之间建立紧密的因果关系。近期一些关于绩效反馈的研究已开始探讨特定行动表现的影响（Eggers & Suh, 2019; Maslach, 2016），其中包括两项并购研究，这些研究考察了特定并购表现与未来并购可能性之间的关系（Haleblian 等, 2006; Kim, Finkelstein, & Haleblian, 2015）。然而，尚无研究深入探讨联盟表现如何影响未来联盟的可能性。此外，关于先前联盟的表现是否影响并购可能性，或者反之，这方面的研究也较为匮乏。针对这些实证空白，我们研究了联盟表现如何影响未来在并购与联盟之间的决策。

In this section, drawing on the existing research on performance feedback and acquisition, we first establish baseline hypotheses on the relationship between alliance performance and the likelihood of making acquisitions without considering the
在本节中，借鉴现有关于绩效反馈和收购的研究，我们首先在不考虑联盟绩效与进行收购可能性之间关系的情况下，建立了基线假设
influence of attribution, and use them as a building block for developing a theory of attribution in learning from performance feedback.
归因的影响，并将其作为开发从绩效反馈中学习的归因理论的基石。
Prior studies on acquisitions have found that firms with poor prior acquisition performance are less likely to make future acquisitions (Haleblian et al., 2006; Kim et al., 2015). When managers have experienced poor acquisition performance, they may come to believe that they do not have skills or capabilities necessary for making successful acquisitions or that there is a poor strategic or organizational fit between their firm and alliances; thus, poor acquisition performance may discourage them from choosing acquisitions as a future governance mode.
先前关于收购的研究发现，之前收购表现不佳的公司不太可能进行未来的收购（Haleblian 等，2006；Kim 等，2015）。当管理者经历过收购表现不佳时，他们可能会认为自己缺乏成功收购所需的技能或能力，或者认为公司与其联盟之间存在战略或组织上的不匹配；因此，糟糕的收购表现可能会阻止他们选择收购作为未来的治理模式。

Extending this logic to alliances, we posit that alliance performance that falls short of the aspiration level will raise the question of whether the alliance is an appropriate governance mode for the firm. Discouraged by poor alliance performance, managers will likely rely less on alliances and more actively explore alternative governance modes that may serve them better-especially acquisitions. Thus, we predict that as alliance performance decreases further below the aspiration level, the likelihood that managers will turn to acquisitions will increase.
将这一逻辑扩展到联盟，我们假设，若联盟表现未达到期望水平，将引发该联盟是否适合作为公司治理模式的质疑。由于表现不佳的联盟令管理者气馁，他们可能会减少对联盟的依赖，并更加积极地探索可能更适合他们的替代治理模式，尤其是收购。因此，我们预测，随着联盟表现进一步低于期望水平，管理者转向收购的可能性将增加。

Baseline Hypothesis 1a. When a firm’s alliance performance is below its aspiration level for alliances, the lower the alliance performance, the more likely the firm will subsequently make an acquisition.
基线假设 1a。当企业的联盟表现低于其联盟期望水平时，联盟表现越低，企业随后进行收购的可能性越大。
By contrast, when firms have experienced a successful alliance that exceeds their aspiration level, they may conclude that they have what it takes to make successful alliances and that alliances are an appropriate governance mode for them. Such inferences will encourage them to continue relying on alliances while discouraging them from exploring alternative governance modes, such as acquisitions. Hence, alliance performance above the aspiration level makes firms less likely to make future acquisitions.
相比之下，当企业经历了超出其期望水平的成功联盟后，他们可能会得出结论，认为自己具备成功联盟的能力，并且联盟是适合他们的治理模式。这种推断将鼓励他们继续依赖联盟，同时阻碍他们探索其他治理模式，如收购。因此，高于期望水平的联盟表现会降低企业未来进行收购的可能性。

Baseline Hypothesis 1b. When a firm’s alliance performance is above its aspiration level for alliances, the higher the alliance performance, the less likely the firm will subsequently make an acquisition.
基线假设 1b。当企业的联盟绩效高于其联盟的期望水平时，联盟绩效越高，企业随后进行收购的可能性就越低。

In this section, we discuss how integrating attribution into performance feedback theory will modify the relationships predicted in the baseline hypotheses. In Baseline Hypothesis 1a, we predict a negative relationship between alliance performance and the
在本节中，我们将探讨将归因整合到绩效反馈理论中，将如何改变基线假设中所预测的关系。在基线假设 1a 中，我们预测联盟绩效与...之间存在负相关关系。
likelihood of subsequently making acquisitions when alliance performance is below the aspiration level. This prediction is based on a view that managers will attribute poor alliance performance to internal causes, such as their lack of alliance capabilities. Research in attribution, however, has provided ample evidence that managers tend to shift the blame for failed strategic decisions to others (e.g., Clapham & Schwenk, 1991; Ford, 1985; Park, Park, & Ramanujam, 2018; Wagner & Gooding, 1997). Selfserving attribution is likely to manifest, particularly among managers who are involved in alliances, because alliances involve external partners whom managers can blame for failure. Indeed, disputes between former partners of failed alliances over the question of who is responsible for the failure are common (Ariño & Doz, 2000; Peng & Shenkar, 2002). Thus, it is possible that managers will attribute poor alliance performance to external causes, such as their partners or unexpected environmental events.
当联盟表现低于期望水平时，后续进行收购的可能性。这一预测基于一种观点，即管理者会将联盟表现不佳归因于内部原因，例如他们缺乏联盟能力。然而，归因研究提供了充分的证据表明，管理者倾向于将战略决策失败的责任转嫁给他人（例如，Clapham & Schwenk, 1991; Ford, 1985; Park, Park, & Ramanujam, 2018; Wagner & Gooding, 1997）。自利归因很可能显现，特别是在参与联盟的管理者中，因为联盟涉及外部合作伙伴，管理者可以将失败归咎于他们。事实上，失败的联盟前合作伙伴之间关于谁应对失败负责的争议很常见（Ariño & Doz, 2000; Peng & Shenkar, 2002）。因此，管理者有可能将联盟表现不佳归因于外部原因，例如他们的合作伙伴或意外的环境事件。

This kind of self-serving attribution may lead them to believe that they would not repeat the failure in future alliances insofar as they could find more capable partners. Supporting this argument, attribution theory has documented that when a failure on a task is attributed to external causes, individuals are more likely to redo the same task because they do not expect that they will fail on the task again (e.g., Fontaine, 1974; Inagi, 1977; Weiner, Heckhausen, & Meyer, 1972). As such, when poor alliance performance is attributed to external causes, it may not deter managers from making alliances again, nor encourage them in searching for an alternative to alliances to the same degree as when attributed to internal causes.
这种自我服务的归因可能会使他们相信，在未来结盟中只要找到更有能力的合作伙伴，就不会重蹈覆辙。支持这一论点的是，归因理论已证明，当任务失败归因于外部原因时，个体更有可能重新执行同一任务，因为他们不预期会再次失败（例如，Fontaine, 1974; Inagi, 1977; Weiner, Heckhausen, & Meyer, 1972）。因此，当联盟表现不佳被归咎于外部因素时，它可能不会阻止管理者再次结盟，也不会像归因于内部原因那样，促使他们寻找替代联盟的方案。

As alliance performance deteriorates further below the aspiration level, the need for managers to find ways to justify the poor performance increases. They will become increasingly wary and start questioning whether they have chosen the right partners. Thus, the poorer the alliance performance is, the stronger their tendency to attribute poor alliance performance to external causes, especially to their partners. Consistent with these arguments, research in attribution has shown that individuals facing severely negative outcomes are more likely to ascribe the outcomes to others than are those experiencing moderately negative outcomes (e.g., Schönbach, 1990; Shaw & McMartin, 1977). Hence, as alliance performance decreases further below the aspiration level, self-serving attribution will further decrease the likelihood of subsequently making an acquisition, flattening the slope from the dashed line to the solid line when alliance performance is below the aspiration level, as shown in Figure 1.
随着联盟绩效进一步恶化并低于期望水平，管理者为糟糕绩效寻找理由的需求增加。他们会变得更加谨慎，并开始质疑是否选择了正确的合作伙伴。因此，联盟绩效越差，他们越倾向于将糟糕的联盟绩效归因于外部因素，尤其是他们的合作伙伴。与这些论点一致，归因研究表明，面临严重负面结果的个体比经历中等负面结果的个体更可能将结果归咎于他人（例如，Schönbach, 1990; Shaw & McMartin, 1977）。因此，当联盟绩效进一步低于期望水平时，自利归因将进一步降低随后进行收购的可能性，如图 1 所示，当联盟绩效低于期望水平时，斜率从虚线变为实线。

FIGURE 1  图 1
The Effect of Self-Serving Attribution on the Relationship between Alliance Performance and Acquisition Likelihood
自利归因对联盟绩效与收购可能性关系的影响

Attribution can also modify the relationship predicted by Baseline Hypothesis 1b. As managers tend to take credit for success by attributing it to their own efforts and abilities (Burger, 1981; Ford, 1985; Mezulis et al., 2004), they are likely to attribute successful alliances to themselves. The self-serving attribution leads them to believe that they are capable of making successful alliances sustainably and that they will also succeed in subsequent alliances (e.g., Fontaine, 1974; Inagi, 1977; Weiner et al., 1972). This will decrease their search for alternatives to alliances, making future acquisitions less likely. Supporting these arguments, attribution theory suggests that, as the success of an action is attributed more to internal and controllable causes (e.g., own strategies and capabilities), the expectation of success through similar actions increases, thus promoting similar actions (e.g., Fontaine, 1974; Weiner et al., 1972).
归因也可以修改基线假设 1b 预测的关系。由于管理者倾向于将成功归功于自己的努力和能力（Burger, 1981; Ford, 1985; Mezulis et al., 2004），他们可能会将成功的联盟归因于自己。这种自我服务的归因使他们相信自己能够持续地建立成功的联盟，并且他们也会在后续的联盟中取得成功（例如，Fontaine, 1974; Inagi, 1977; Weiner et al., 1972）。这将减少他们对联盟替代方案的搜索，从而降低未来收购的可能性。支持这些论点，归因理论表明，当行动的成功更多地归因于内部和可控的原因（例如，自己的策略和能力）时，通过类似行动取得成功的期望会增加，从而促进类似的行动（例如，Fontaine, 1974; Weiner et al., 1972）。

The stronger the alliance performance is, the more likely managers will be to attribute the strong performance internally because extremely strong performance that far exceeds their aspiration level will boost their confidence and increase their tendency to overestimate their contribution in comparison to their partners. Such strong performance will also motivate them to seize the moment to enhance their reputation and status by taking more credit than
联盟表现越强劲，管理者越有可能将这一优异表现归因于内部因素，因为远超其期望水平的卓越表现会增强他们的信心，并增加他们相较于合作伙伴高估自身贡献的倾向。如此强劲的表现还会激励他们抓住时机，通过更多地邀功来提升自己的声誉和地位。
they deserve (Shi, Zhang, & Hoskisson, 2017). Thus, as alliance performance increases further above the aspiration level, self-serving attribution will further decrease the likelihood of subsequently making an acquisition, increasing the slope from the dashed line to the solid line when alliance performance is above the aspiration level, as shown in Figure 1.
他们应得的（Shi, Zhang, & Hoskisson, 2017）。因此，随着联盟绩效进一步超过期望水平，自我服务归因将进一步降低随后进行收购的可能性，当联盟绩效超过期望水平时，斜率从虚线增加到实线，如图 1 所示。

Above, we propose that the presence of attribution can significantly influence how firms respond to past performance beyond the existing literature on performance feedback theory. However, we cannot tease out the effects of attribution from other organizational processes due to the difficulties of directly measuring and testing a cognitive process, such as attribution at firm level. In this situation, researchers have typically explored the moderating factors that could influence the relationship between the cognitive processes and firm behavior to uncover the mechanisms by which such processes operate (e.g., Azoulay, Stuart, & Wang, 2014; Reitzig & Sorenson, 2013). Following this approach, we identify and theorize on the characteristics of alliances that could significantly affect the degree to which the selfserving attribution manifests.
我们提出，归因的存在可以显著影响企业如何回应过去的绩效，这超出了现有绩效反馈理论的文献。然而，由于难以直接测量和测试企业层面的认知过程（如归因），我们无法将归因的影响与其他组织过程区分开来。在这种情况下，研究人员通常探索可能影响认知过程与企业行为之间关系的调节因素，以揭示这些过程运作的机制（例如，Azoulay, Stuart, & Wang, 2014; Reitzig & Sorenson, 2013）。遵循这一方法，我们识别并理论化联盟的特征，这些特征可能显著影响自利归因的表现程度。

We identified three moderating factors firmly grounded in attribution theory using a two-step process. First, we systematically reviewed the attribution literature to identify theoretical factors that have been found to influence self-serving attribution (e.g., Gilovich, Griffin, & Kahneman, 2002; Gregg, Sedikides, & Gebauer, 2011; Harvey, Madison, Martinko, Crook, & Crook, 2014; Kwang & Swann, 2010; Sedikides & Alicke, 2012, 2019; Shepperd et al., 2008). Weiner (1992: 230) noted that “there is no unified body of knowledge that neatly fits into one specific attribution theory.” Hence, we focused on the factors that have been proven to influence attribution in performance assessment in line with our theoretical interests. Second, among the identified factors, we excluded those deemed irrelevant to our empirical context and setting (e.g., individuallevel factors, such as demographic profiles or psychological characteristics). This process helped us identify three theoretical factors: (a) the diffusion of responsibility (e.g., Beyer, Sidarus, Bonicalzi, & Haggard, 2017; Gregg et al., 2011), (b) the perceived capability of partners (e.g., Elliott, 1989; Sedikides & Alicke, 2012), and © the ambiguity of performance
我们通过两步流程确定了三个基于归因理论的调节因素。首先，我们系统地回顾了归因文献，以识别已被发现影响自我服务归因的理论因素（例如，Gilovich, Griffin, & Kahneman, 2002; Gregg, Sedikides, & Gebauer, 2011; Harvey, Madison, Martinko, Crook, & Crook, 2014; Kwang & Swann, 2010; Sedikides & Alicke, 2012, 2019; Shepperd et al., 2008）。Weiner (1992: 230) 指出，“没有一个统一的知识体系能够完全符合某一特定的归因理论。”因此，我们根据理论兴趣，重点关注那些已被证明在绩效评估中影响归因的因素。其次，在识别出的因素中，我们排除了被认为与我们的实证背景和设置无关的因素（例如，个体层面的因素，如人口统计特征或心理特征）。 这一过程帮助我们识别了三个理论因素：(a) 责任分散（例如，Beyer, Sidarus, Bonicalzi, & Haggard, 2017; Gregg et al., 2011），(b) 合作伙伴的感知能力（例如，Elliott, 1989; Sedikides & Alicke, 2012），以及 (c) 绩效的模糊性。
information (e.g., Dunning, Meyerowitz, & Holzberg, 1989; Gregg et al., 2011).
信息（例如，Dunning, Meyerowitz, & Holzberg, 1989; Gregg 等, 2011）。

When individuals believe they have little control over outcomes, they feel less responsible for those outcomes and are more likely to attribute negative outcomes to external causes (Weiner, 1979, 1986). When a task is performed by several individuals, each individual’s control over the outcome of the task decreases and responsibility can be diffused across several individuals; hence, each individual will feel less responsible for the outcome (Beyer et al., 2017; Forsyth, Zyzniewski, & Giammanco, 2002; Mynatt & Sherman, 1975). This heuristic is activated when a task fails, as individuals may attempt to diffuse their responsibility for the failure to others. The larger the number of individuals involved in a task, the more broadly the responsibility can be diffused, and the less likely that each individual will feel accountable for the failure of the task, setting the stage for external attribution (Forsyth et al., 2002). Thus, the diffusion of responsibility amplifies individuals’ tendency to attribute failure to others.
当个体认为自己对结果的控制力较弱时，他们会感到对这些结果的责任感降低，并更倾向于将负面结果归因于外部原因（Weiner, 1979, 1986）。当一项任务由多个个体共同完成时，每个个体对任务结果的控制力会减弱，责任可能会分散到多个个体身上；因此，每个个体会感到对结果的责任感降低（Beyer 等, 2017; Forsyth, Zyzniewski, & Giammanco, 2002; Mynatt & Sherman, 1975）。这种启发式机制在任务失败时会被激活，因为个体可能会试图将失败的责任推卸给他人。参与任务的个体数量越多，责任分散的范围就越广，每个个体对任务失败的责任感就越弱，从而为外部归因创造了条件（Forsyth 等, 2002）。因此，责任分散放大了个体将失败归因于他人的倾向。
The performance of an alliance is determined by the collective contribution of all partners, but each has only limited control over the alliance outcomes (Diestre & Rajagopalan, 2012; Lavie, 2007). As the number of alliance partners increases, accountability for the alliance outcome is diffused among a larger number of partners, decreasing each partner’s accountability for the alliance’s performance. Further, a larger number of partners makes it more difficult to clearly identify the relative contribution of each partner, due to the increased complexity and causal ambiguity, further promoting self-serving attribution.
联盟的绩效由所有合作伙伴的集体贡献决定，但每个合作伙伴对联盟结果的控制力有限（Diestre & Rajagopalan, 2012; Lavie, 2007）。随着联盟合作伙伴数量的增加，联盟结果的责任分散在更多的合作伙伴中，从而降低了每个合作伙伴对联盟绩效的责任感。此外，由于复杂性和因果模糊性的增加，更多的合作伙伴使得明确识别每个合作伙伴的相对贡献变得更加困难，进一步助长了自利归因。
Thus, as the number of partners increases, managers are more likely to attribute alliance failure to their partners and conclude that future alliances will not fail as long as they work with more capable partners, thereby decreasing the likelihood that they will turn to an acquisition as their next move. Hence, the poor performance of an alliance with a large number of partners will deter future alliances and increase future acquisitions to a lesser extent than the poor performance of an alliance with a small number of partners.
因此，随着合作伙伴数量的增加，管理者更有可能将联盟失败归因于其合作伙伴，并得出结论：只要与更有能力的合作伙伴合作，未来的联盟就不会失败，从而降低了他们转向收购作为下一步行动的可能性。因此，与拥有大量合作伙伴的联盟表现不佳相比，与少量合作伙伴的联盟表现不佳将在较小程度上阻碍未来的联盟并增加未来的收购。

On the one hand, when an alliance fails, the presence of multiple partners may help managers diffuse responsibility for the failure to other partners. On the other hand, when an alliance succeeds, it makes it difficult for them to take full credit for the success. Research on attribution theory has suggested that the low perceived control over outcomes decreases selfserving attribution for success (Forsyth et al., 2002; Li, Han, Lei, Holroyd, & Li, 2011). The fact that a firm has low control over an alliance outcome implies that its contribution to the success of the alliance is limited and that it cannot easily dismiss the contributions of external factors, especially its alliance partners (Ariño & Doz, 2000; Yin, Wu, & Tsai, 2012).
一方面，当联盟失败时，多个合作伙伴的存在可能有助于管理者将失败的责任分散到其他合作伙伴身上。另一方面，当联盟成功时，这使他们难以完全归功于自己的成功。归因理论的研究表明，对结果的低感知控制减少了对成功的自利归因（Forsyth 等人，2002；Li，Han，Lei，Holroyd，& Li，2011）。企业对联盟结果控制力低的事实意味着其对联盟成功的贡献有限，并且不能轻易忽视外部因素，尤其是其联盟合作伙伴的贡献（Ariño & Doz，2000；Yin，Wu，& Tsai，2012）。

Managers may be predisposed to attribute success to themselves and their firm regardless of the number of partners in an alliance. As the number of partners in an alliance increases, however, their perceived control over outcomes decreases, and it becomes more difficult for them to claim a larger share of the credit for a successful alliance because the success is a culmination of joint efforts of all the partners and each partner may vie for more credit and recognition for the success (Ariño & Doz, 2000). Supporting these arguments, prior studies in attribution theory have found that as the number of partners for a joint project increases, individuals are more likely to make a self-serving attribution for failure, but not for success (Beyer et al., 2017; Mynatt & Sherman, 1975), and they are more likely to fairly acknowledge others’ contributions to the success of the project (e.g., Forsyth et al., 2002).
管理者可能倾向于将成功归因于自己和公司，而不管联盟中有多少合作伙伴。然而，随着联盟中合作伙伴数量的增加，他们对结果的感知控制力下降，且更难在成功的联盟中主张更大的功劳，因为成功是所有合作伙伴共同努力的结果，每个合作伙伴都可能为获得更多的功劳和认可而竞争（Ariño & Doz, 2000）。支持这些论点，先前关于归因理论的研究发现，随着联合项目中合作伙伴数量的增加，个人更可能对失败做出自利的归因，而非对成功（Beyer 等，2017；Mynatt & Sherman, 1975），并且他们更可能公平地承认他人对项目成功的贡献（例如，Forsyth 等，2002）。

In sum, we propose that a high number of alliance partners will reduce managers’ tendency to attribute alliance success to themselves. As success is attributed less to themselves, the expectation for future alliance success decreases, consequently reducing the reliance on alliances and promoting an exploration of alternatives (e.g., Fontaine, 1974; Weiner et al., 1972). Hence, a strong performance of multipartner alliances will promote future alliances and deter future acquisitions to a lesser extent than would single-partner alliances.
总之，我们提出，大量的联盟伙伴将减少管理者将联盟成功归因于自身的倾向。由于成功较少归因于自身，对未来联盟成功的预期会降低，从而减少对联盟的依赖，并促进对替代方案的探索（例如，Fontaine, 1974; Weiner et al., 1972）。因此，多伙伴联盟的强劲表现将比单伙伴联盟在较小程度上促进未来联盟并阻碍未来收购。

Figure 2 depicts the three hypothesized moderating effects on the relationship between alliance performance relative to the aspiration level and the likelihood of choosing an acquisition (proposed in Hypotheses 1a through 3b).
图 2 描绘了联盟绩效相对于期望水平与选择收购可能性之间关系的三个假设调节效应（假设 1a 至 3b 中提出）。

Please note that we predict that the number of alliance partners will engender asymmetric effects for success and failure. Specifically, it will increase selfserving bias when alliance performance is below the aspiration level but will decrease self-serving bias when alliance performance is above the aspiration level. We propose that the other two factors will produce symmetrical effects for success and failure.
请注意，我们预测联盟伙伴的数量将对成功和失败产生不对称影响。具体而言，当联盟绩效低于期望水平时，这会增加自我服务偏差；而当联盟绩效高于期望水平时，则会减少自我服务偏差。我们提出，其他两个因素将对成功和失败产生对称影响。

The perceived capability of self and others has been deemed an important factor for attribution in performance assessment in both attribution (Weiner, 1979, 1986) and management (Rhee, 2009) research. In particular, the perceived capability of others in a joint task significantly influences how individuals assess and attribute the outcomes of the task (Silvia & Duval, 2001; Strube & Berry, 1985). Building on and extending these arguments, we posit that highly capable alliance partners will reduce managers’ tendency to attribute a failed alliance to their partners.
自我与他人感知到的能力，在归因（Weiner, 1979, 1986）与管理学（Rhee, 2009）研究中，均被视为绩效评估中归因的重要因素。特别是在联合任务中，对他人能力的感知，显著影响个体如何评估和归因任务结果（Silvia & Duval, 2001; Strube & Berry, 1985）。基于并扩展这些论点，我们提出，高能力的联盟伙伴将减少管理者将联盟失败归因于其伙伴的倾向。

When evaluating the outcomes of a complex event such as an alliance, it is difficult for a firm’s managers to methodically measure each partner firm’s actual contributions (Zollo, 2009). Thus, their perceptions of the partner firms’ capabilities will be instrumental in estimating the contribution of each partner (Simcoe & Waguespack, 2011). Managers’ perception of other firms’ capabilities regarding a task is often determined by the amount of experience managers have with the same or similar tasks (Eggers & Kaplan, 2013). Organizational learning research has recognized experience as a primary source of acquiring and enhancing capabilities, suggesting that a firm’s level of experience is positively related to its performance (Argote, 2013). Studies on alliances have similarly reported that a firm’s alliance experience is positively related to its alliance capability (for a review, see Wang & Rajagopalan, 2015).
在评估诸如联盟等复杂事件的结果时，企业的管理者很难系统性地衡量每个合作伙伴的实际贡献（Zollo, 2009）。因此，他们对合作伙伴能力的感知将有助于估算每个合作伙伴的贡献（Simcoe & Waguespack, 2011）。管理者对其他企业在某项任务上的能力的感知，通常取决于他们在相同或类似任务上的经验量（Eggers & Kaplan, 2013）。组织学习研究已认识到经验是获取和提升能力的主要来源，表明企业的经验水平与其绩效呈正相关（Argote, 2013）。关于联盟的研究同样报告称，企业的联盟经验与其联盟能力呈正相关（详见 Wang & Rajagopalan, 2015 的综述）。

Managers tend to attribute failed alliances to partners, but such an attribution is less likely if they perceive their partners to have strong alliance capabilities, as the partners’ strong capabilities render such an attribution less credible (Salancik & Meindl, 1984). Supporting this argument, attribution research has shown that individuals are less likely to attribute the failure of a joint task to other collaborators who are perceived to be experienced and capable of performing the task (Elliott, 1989; Silvia & Duval, 2001; Strube & Berry, 1985). Even if such an attribution is made, it is less likely to be channeled to those with strong
管理者倾向于将联盟失败归咎于合作伙伴，但如果他们感知到合作伙伴具有强大的联盟能力，这种归咎就不太可能发生，因为合作伙伴的强大能力使这种归咎显得不那么可信（Salancik & Meindl, 1984）。支持这一论点的是，归因研究显示，个体不太可能将联合任务的失败归咎于那些被认为经验丰富且有能力执行任务的合作者（Elliott, 1989; Silvia & Duval, 2001; Strube & Berry, 1985）。即使进行了这样的归咎，也不太可能指向那些能力强的人。

perceived capability than those with weak perceived capability. Further, a firm’s stakeholders will believe that the partners with high perceived capability are less likely to be responsible for the failed alliance (Fombrun, 1996; Obloj & Capron, 2011; Simcoe & Waguespack, 2011). Thus, the presence of alliance
感知能力强的合作伙伴比感知能力弱的合作伙伴更受青睐。此外，企业的利益相关者会认为，具有高感知能力的合作伙伴不太可能对联盟失败负责（Fombrun, 1996; Obloj & Capron, 2011; Simcoe & Waguespack, 2011）。因此，联盟的存在
partners with strong perceived capability reduces managers’ tendency to attribute alliance failure to their partners. Consequently, they may have to take greater responsibility for an alliance failure, and their ability to make successful alliances in the future may be scrutinized.
合作伙伴的强大感知能力降低了管理者将联盟失败归因于合作伙伴的倾向。因此，他们可能不得不为联盟失败承担更大的责任，并且他们未来成功建立联盟的能力可能会受到审查。

Taken together, we predict that the higher the partners’ alliance experience, the less likely a firm will attribute a failed alliance to them. Hence, poor alliance performance will deter future alliances but promote future acquisitions to a greater extent when partners are perceived to have strong alliance capability than when they are perceived to be less capable (e.g., Fontaine, 1974; Weiner et al., 1972).
综上所述，我们预测合作伙伴的联盟经验越丰富，企业将联盟失败归咎于他们的可能性就越小。因此，当合作伙伴被认为具有较强的联盟能力时，联盟表现不佳会更大幅度地阻碍未来的联盟，但会促进未来的收购，相比之下，当合作伙伴被认为能力较弱时，这种影响则较小（例如，Fontaine, 1974; Weiner 等, 1972）。

A firm’s stakeholders may also believe that alliance partners with strong alliance capabilities have made a greater contribution to the success of an alliance compared to those with weak alliance capabilities, just as they do not consider highly capable participants to have led an alliance to failure (Fombrun, 1996; Simcoe & Waguespack, 2011). Thus, the presence of capable partners may reduce managers’ tendency to take more credit than they deserve for a successful alliance because they are aware that such a self-serving attribution will not be taken as credible unless they can present supporting evidence (Salancik & Meindl, 1984). Consistent with these arguments, attribution research has found that partners’ high perceived or realized individual performance limits the self-serving attribution for a successful collaborative task (e.g., Elliott, 1989; Strube & Berry, 1985).
公司的利益相关者可能也会认为，与联盟能力较弱的伙伴相比，拥有强大联盟能力的伙伴对联盟成功的贡献更大，正如他们不会认为能力强的参与者会导致联盟失败一样（Fombrun, 1996; Simcoe & Waguespack, 2011）。因此，有能力的伙伴的存在可能会减少管理者在成功联盟中过度邀功的倾向，因为他们意识到，除非能提供支持性证据，否则这种自我服务的归因不会被视作可信（Salancik & Meindl, 1984）。与这些论点一致，归因研究发现，伙伴的高感知或实际个人表现限制了对成功协作任务的自我服务归因（例如，Elliott, 1989; Strube & Berry, 1985）。
In sum, when a firm that collaborates with partners with high perceived alliance capabilities makes a successful alliance, the firm’s managers will be less likely to attribute the success internally, in turn reducing their confidence and expectation that future alliances would be also successful (compared to when their partners are perceived to have weak alliance capabilities) (e.g., Fontaine, 1974; Weiner et al., 1972). Thus, strong alliance performance will promote future alliances and deter future acquisitions to a lesser extent when the partners’ capabilities are perceived to be strong than when they are perceived to be weak.
总之，当一家与被认为具有高联盟能力的合作伙伴成功建立联盟时，该公司的管理者将不太可能将成功归因于内部，从而降低他们对未来联盟也会成功的信心和期望（与合作伙伴被认为联盟能力较弱时相比）（例如，Fontaine, 1974; Weiner et al., 1972）。因此，当合作伙伴的能力被认为较强时，强大的联盟表现将在较小程度上促进未来联盟并阻止未来收购，相比于合作伙伴被认为能力较弱时。

The consistency of performance over time contains crucial information for evaluating and attributing
随时间推移的性能一致性包含评估和归因的关键信息
current performance, and has been recognized as an important factor in attribution (Meyer, 1980; Weiner, 1979, 1986). Consistent prior performance allows firms to clearly identify a common pattern and make accurate causal attributions of their performance. In contrast, inconsistent prior performance complicates the process of identifying a common pattern and drawing unambiguous lessons, increasing the difficulty of making accurate causal attributions (Arrfelt, Wiseman, & Hult, 2013). We posit that the ambiguity arising from inconsistent prior alliance performance will not only obscure managers’ ability to learn from performance feedback but also amplify their tendency to make self-serving attributions.
当前的表现，已被认为是归因中的一个重要因素（Meyer, 1980; Weiner, 1979, 1986）。一致的前期表现使企业能够清晰地识别出一个共同模式，并对其表现做出准确的因果归因。相反，不一致的前期表现则使识别共同模式和吸取明确教训的过程复杂化，增加了做出准确因果归因的难度（Arrfelt, Wiseman, & Hult, 2013）。我们认为，前期联盟表现不一致所带来的模糊性不仅会掩盖管理者从绩效反馈中学习的能力，还会放大他们做出自利归因的倾向。
Low variability ${}^1$${}^1$^(1){ }^{1} in alliance performance allows managers to forecast their future alliance performance more accurately, which in turn helps them make more informed decisions on which governance mode to choose next time. It also limits their tendency to make self-serving attributions when they experience alliance performance that falls short of their aspiration. Low variability in alliance performance reduces ambiguity and noise in performance feedback and provides a consistent pattern of action-outcome relationships that allow managers to more clearly identify the cause of the failure (Dahl & Ransom, 1999; Felson, 1981; Willard & Gramzow, 2008). This causal clarity leaves little room for managers to attribute the failure to their alliance partners or other contingent factors because they cannot “use divergent criteria and draw on disparate types of behavioral evidence” (Dunning et al., 1989: 1082).
联盟绩效的低变异性 ${}^1$${}^1$^(1){ }^{1} 使管理者能够更准确地预测未来的联盟绩效，这反过来有助于他们在下次选择治理模式时做出更明智的决策。它还限制了他们在联盟绩效未达到预期时做出自利归因的倾向。联盟绩效的低变异性减少了绩效反馈中的模糊性和噪音，并提供了行动与结果之间的一致模式，使管理者能够更清楚地识别失败的原因（Dahl & Ransom, 1999; Felson, 1981; Willard & Gramzow, 2008）。这种因果关系的明确性使管理者几乎没有空间将失败归因于联盟伙伴或其他偶然因素，因为他们无法“使用不同的标准并依赖不同类型的行为证据”（Dunning et al., 1989: 1082）。
In contrast, highly variable alliance performance complicates managers’ tasks in establishing causal relationships between their actions and the outcomes of past alliances. Highly variable alliance performance indicates that a firm has experienced a wide range of alliance performance levels (both success and failure) and sends mixed signals about its ability to make successful alliances (Kim et al., 2015). Even if its recent alliance performance was unsatisfactory, the fact that it has also experienced successful alliances implies that it is possible to make a successful alliance in the future. In light of the possibility of achieving alliance performance that positively deviates from recent alliance failure, managers may frame the recent failure as an outcome that does not reflect their true capabilities and
相比之下，联盟绩效的高度可变性使得管理者在建立自身行为与过去联盟结果之间的因果关系时任务变得复杂。高度可变的联盟绩效表明，一家公司经历了从成功到失败的各种联盟绩效水平，这发出了关于其能否成功建立联盟的混杂信号（Kim 等，2015）。即使其最近的联盟绩效不尽如人意，但考虑到它也曾有过成功的联盟经历，这意味着未来仍有可能达成成功的联盟。鉴于存在实现联盟绩效积极偏离近期失败的可能性，管理者可能会将最近的失败视为一个并不反映其真实能力的结果。

attribute it to external factors (Gilovich et al., 2002; Tversky & Kahneman, 1974). This framing and attribution will not discourage them from making future alliances.
将其归因于外部因素（Gilovich et al., 2002; Tversky & Kahneman, 1974）。这种框架和归因不会阻止他们未来建立联盟。

Supporting these arguments, attribution research has found that individuals are more likely to make a self-serving attribution for failure when the information pertaining to the failure is ambiguous and contradictory (e.g., Dahl & Ransom, 1999; Dunning et al., 1989; Felson, 1981). Jordan and Audia (2012) pushed this argument further to assert that managers may exploit greater information ambiguity to portray poor firm performance in a positive light-for example, by manipulating comparison groups and emphasizing the organizational goals that were better met.
支持这些论点的是，归因研究发现，当与失败相关的信息模糊且矛盾时，个体更可能做出自我服务的归因（例如，Dahl & Ransom, 1999; Dunning 等, 1989; Felson, 1981）。Jordan 和 Audia（2012）进一步推进了这一论点，断言管理者可能会利用更大的信息模糊性，将糟糕的公司表现描绘成积极的一面——例如，通过操纵比较组和强调更好地实现的组织目标。

Taken together, we predict that firms with inconsistent, highly variable prior alliance performance will be more likely to attribute a failed alliance to external causes compared to those with consistent prior alliance performance; hence, they will be likely to switch to acquisitions after experiencing poor alliance performance that fell below the aspiration level to a lesser extent versus those with consistent prior alliance performance.
综上所述，我们预测，与之前联盟表现一致的企业相比，那些先前联盟表现不一致且高度波动的企业更有可能将联盟失败归因于外部原因；因此，在经历低于期望水平的糟糕联盟表现后，它们转向收购的可能性相较于先前联盟表现一致的企业要小。

The mixed signals from highly variable prior alliance performance may also make managers attribute successful alliances to themselves. Seen from a negative perspective, highly variable prior alliance performance signifies that they have been unable to reliably make successful alliances. Seen from a positive perspective, however, it can be considered evidence that they can make successful alliances under the right conditions.
来自高度不稳定的先前联盟表现的混合信号也可能使管理者将成功的联盟归功于自己。从负面角度看，高度不稳定的先前联盟表现意味着他们一直无法可靠地建立成功的联盟。然而，从积极的角度看，这可以被视为他们在适当条件下能够建立成功联盟的证据。
In light of the strong performance in the recent alliance, managers recollect previous successes and draw parallels between recent and past successes. In attributing recent success, they can give more weight to the actions they took in the focal alliance as well as in successful past alliances (e.g., close monitoring and specific contractual agreement) to conclude that such actions led to the success. When individuals have both positive and negative experiences, they tend to rely much more on positive memories when explaining a recent positive event than on negative recollections (e.g., Mayer, McCormick, & Strong, 1995).
鉴于近期联盟的强劲表现，管理者们回顾了以往的成功，并在近期的成功与过去的成功之间进行了类比。在归因于近期的成功时，他们可能会更加重视他们在焦点联盟以及过去成功的联盟中所采取的行动（例如，密切监控和具体的合同协议），从而得出这些行动促成了成功的结论。当个体既有积极也有消极的经历时，他们在解释近期积极事件时，往往更多地依赖积极的记忆，而非消极的回忆（例如，Mayer, McCormick, & Strong, 1995）。

Thus, when previous alliance performances have been more variable, managers are more likely to make
因此，当以往联盟表现更加多变时，管理者更有可能采取行动
the internal attribution of a successful alliance. In line with these arguments, evidence from attribution research has consistently shown that ambiguous performance-assessment situations facilitate individuals’ tendency to make a positive assessment of their achievements (Dahl & Ransom, 1999; Dunning et al., 1989; Felson, 1981; Willard & Gramzow, 2008).
成功联盟的内部归因。根据这些论点，归因研究的证据一致表明，模糊的绩效评估情境促进了个体对其成就做出积极评估的倾向（Dahl & Ransom, 1999; Dunning et al., 1989; Felson, 1981; Willard & Gramzow, 2008）。

In sum, we predict that firms with a high variability in past alliance performance will be more likely to attribute their success to themselves, and, consequently, will be more likely to expect future alliances to be successful and continue choosing alliances; hence, firms with a high variability in past alliance performance are likely to switch to acquisitions after experiencing a successful alliance to a lesser extent than those with a low variability.
总之，我们预测，过去联盟绩效波动较大的企业更有可能将成功归因于自身，因此更可能预期未来的联盟会成功并继续选择联盟；因此，与波动较小的企业相比，过去联盟绩效波动较大的企业在经历一次成功的联盟后，转向收购的可能性较小。

Our sample consists of acquisitions and alliances announced and completed by Fortune 500 firms between 1994 and 2017. Such activities by these large firms are widely published, and thus we can obtain accurate data on their alliances. The study period was chosen with two considerations: (a) the Securities Data Corporation (SDC) databases, our main data sources, are known to be unreliable prior to 1989 (Anand & Khanna, 2000); and (b) data from the five-year window between 1989 and 1993 were used to measure alliance and acquisition experiences (e.g., Argote, 2013; Wang & Zajac, 2007).
我们的样本包括 1994 年至 2017 年间《财富》500 强企业宣布并完成的收购和联盟。这些大型企业的此类活动被广泛报道，因此我们可以获得关于它们联盟的准确数据。选择研究时期基于两点考虑：(a) 我们主要的数据来源——证券数据公司（SDC）数据库，在 1989 年之前被认为不可靠（Anand & Khanna, 2000）；(b) 1989 年至 1993 年这五年间的数据被用于衡量联盟和收购经验（例如，Argote, 2013; Wang & Zajac, 2007）。

We first identified companies listed on the 1994 Fortune 500 list. Then, we collected data on alliances completed by these firms from the SDC Joint Ventures and Alliances database. We included alliances made not only by the sample firms but also by their subsidiaries, because decisions on alliances and acquisitions are typically made by the headquarters (Villalonga & McGahan, 2005). Data on acquisitions were collected from the SDC Mergers and Acquisitions database. We eliminated duplicate deal entries and verified the announcement dates by comparing them with news articles from major business news outlets such as Financial Times and Bloomberg News. The announcement dates from the SDC and the news articles matched for about $77\mathrm{\%}$$77\mathrm{\%}$77%77 \% of the
我们首先确定了 1994 年《财富》500 强榜单上的公司。随后，我们从 SDC 合资与联盟数据库中收集了这些公司完成的联盟数据。我们不仅纳入了样本公司自身的联盟，还包括其子公司的联盟，因为关于联盟和收购的决策通常由总部做出（Villalonga & McGahan, 2005）。收购数据则来源于 SDC 并购数据库。我们删除了重复的交易记录，并通过与《金融时报》和彭博新闻等主要商业新闻媒体的报道对比，核实了公告日期。SDC 与新闻报道中的公告日期大约在 $77\mathrm{\%}$$77\mathrm{\%}$77%77 \% 的比例上吻合。
alliances; when there was a discrepancy, we prioritized the information from the news articles (e.g., Anand & Khanna, 2000; Villalonga & McGahan, 2005).
联盟；当存在差异时，我们优先考虑新闻文章中的信息（例如，Anand & Khanna, 2000; Villalonga & McGahan, 2005）。

For our main analyses, we included only equity alliances and joint ventures, excluding nonequity alliances, such as licensing and manufacturing arrangements, marketing agreements, and research collaborations, for two reasons. First, previous research has suggested that joint ventures and equity alliances are more comparable to acquisitions than are nonequity alliances (e.g., Reuer & Ragozzino, 2012; Teece, 1992). Second, equity alliances and joint ventures are generally more consequential for the strategy and performance of a firm than are nonequity alliances; hence, they capture greater managerial attention, which in turn has a stronger influence on the firm’s attribution and learning processes. Consistent with this reasoning, prior studies on governance mode choice have typically focused on equity alliances and joint ventures (e.g., Folta, 1998; Reuer & Ragozzino, 2012; Yang, Lin, & Peng, 2011). We tested the sensitivity of our findings using different classifications of alliances and obtained consistent results (see the “Robustness Check” section for more details).
在我们的主要分析中，我们仅纳入了股权联盟和合资企业，排除了非股权联盟，如许可和制造安排、营销协议以及研究合作，原因有二。首先，先前的研究表明，合资企业和股权联盟与收购相比，比非股权联盟更具可比性（例如，Reuer & Ragozzino, 2012; Teece, 1992）。其次，股权联盟和合资企业通常比非股权联盟对公司的战略和绩效影响更大；因此，它们吸引了更多的管理关注，进而对公司的归因和学习过程产生更强的影响。与此推理一致，先前关于治理模式选择的研究通常聚焦于股权联盟和合资企业（例如，Folta, 1998; Reuer & Ragozzino, 2012; Yang, Lin, & Peng, 2011）。我们通过不同联盟分类测试了研究结果的敏感性，并得出了一致的结论（详见“稳健性检验”部分）。

Finally, we collected quarterly data on the demographic and financial information of the sample firms and alliance partners from the COMPUSTAT/ CRSP merged database. After accounting for missing observations, the sample consisted of 39,570 firmquarter observations. Because our analysis estimates the likelihood that a firm, whose most recent transaction in a given quarter was an alliance (thereby being at risk of changing its governance mode to an acquisition in the next quarter), would make an acquisition in the next quarter, we included only firm-quarter observations, where the most recent transaction was an alliance. This resulted in the final sample of 13,535 firm-quarter observations; 4,017 incidents were found where firms changed from an alliance to an acquisition. If some firms had made only acquisitions but no alliances (or vice versa), this might have caused sample-induced endogeneity bias (Certo, Busenbark, Woo, & Semadeni, 2016). However, all firms in our sample made both alliances and acquisitions during our study period.
最后，我们从 COMPUSTAT/CRSP 合并数据库中收集了样本公司及其联盟伙伴的人口统计和财务信息的季度数据。在考虑了缺失观测值后，样本包含 39,570 个公司季度观测值。由于我们的分析估计了在某一季度中最近一次交易为联盟的公司（因此在下一季度有可能将其治理模式更改为收购）在下一季度进行收购的可能性，我们仅包括了最近一次交易为联盟的公司季度观测值。这最终得到了 13,535 个公司季度观测值的样本；其中发现了 4,017 起公司从联盟转变为收购的事件。如果某些公司仅进行收购而没有联盟（或反之），这可能会导致样本诱导的内生性偏差（Certo, Busenbark, Woo, & Semadeni, 2016）。然而，在我们的研究期间，样本中的所有公司都进行了联盟和收购。

Our dependent variable is the unobserved hazard rate for a firm that used an alliance at $t$$t$tt to change to an acquisition at $t+1$$t+1$t+1t+1. The dependent variable
我们的因变量是企业在 $t$$t$tt 时使用联盟后，在 $t+1$$t+1$t+1t+1 时转变为收购的未观测到的风险率。The dependent variable
equals 1 if the first event that a firm made at $t+1$$t+1$t+1t+1 was an acquisition, and 0 otherwise. ${}^2$${}^2$^(2){ }^{2}
如果一家公司在 $t+1$$t+1$t+1t+1 进行的首次事件是收购，则等于 1，否则为 0。 ${}^2$${}^2$^(2){ }^{2}

The hazard rates of a firm changing from an alliance to an acquisition were estimated using the Cox model (Tuma & Hannan, 1984). Firm identifiers were included to control for potential heterogeneity, and standard errors were clustered on firms. This model estimates the likelihood that a firm whose most recent transaction was an alliance at the time of observation would subsequently make an acquisition, given its characteristics (Blossfeld, Golsch, & Rohwer, 2007). ${}^3$${}^3$^(3){ }^{3} We estimated models with several alternative econometric specifications and obtained consistent results (see the “Robustness Check” section for details).
使用 Cox 模型（Tuma & Hannan, 1984）估计了企业从联盟转变为收购的风险率。企业标识符被纳入以控制潜在的异质性，标准误差在企业层面进行了聚类。该模型估计了在观察时最近交易为联盟的企业，基于其特性，随后进行收购的可能性（Blossfeld, Golsch, & Rohwer, 2007）。 ${}^3$${}^3$^(3){ }^{3} 我们采用了几种替代的经济计量学规范进行模型估计，并得出一致的结果（详见“稳健性检验”部分）。

Alliance performance. We measured alliance performance using cumulative abnormal returns (CAR). CAR is the most frequently used method for measuring major corporate events, such as alliances (Zollo & Meier, 2008), and is appropriate for our study for two reasons.
联盟绩效。我们使用累积异常收益（CAR）来衡量联盟绩效。CAR 是衡量诸如联盟等重大企业事件最常用的方法（Zollo & Meier, 2008），并且适用于我们的研究，原因有二。

First, our theory stipulates that managers consider performance feedback when making decisions; thus, testing it requires a performance measure that significantly influences managers’ decision-making. Stock market reaction is widely regarded as an important performance feedback measure by corporate managers, stakeholders, and observers. Davis (2009: 22) noted that, “for companies, share price reactions upon the announcement of corporate actions [provides] immediate feedback on the wisdom of those actions, allowing firms to quickly change course when necessary to avoid the damages foreseen by the market.” In line with this view, a manager of one of the sample firms noted in an interview with us that they “undertake a comprehensive analysis of our stock price changes after each deal announcement… Once we even withdrew a bid because of the disappointing reaction.” Further, managers are frequently
首先，我们的理论规定，管理者在决策时会考虑绩效反馈；因此，测试这一理论需要一个能显著影响管理者决策的绩效衡量标准。股票市场反应被企业管理者、利益相关者和观察者广泛视为重要的绩效反馈指标。Davis（2009: 22）指出，“对于公司而言，公司行动宣布时的股价反应提供了这些行动明智与否的即时反馈，使公司能够在必要时迅速改变方向，以避免市场预见的损害。”与这一观点一致，样本公司中的一位管理者在接受我们采访时提到，他们“在每次交易宣布后都会对股价变化进行全面分析……有一次我们甚至因为市场反应不佳而撤回了投标。”此外，管理者经常

rewarded or sanctioned based on changes in their firm’s stock value after a major corporate event (e.g., Hall & Liebman, 1998), incentivizing them to heed this performance measure carefully.
根据公司重大事件后公司股票价值的变化来奖励或制裁（例如，Hall & Liebman, 1998），激励他们仔细关注这一绩效指标。
Second, while alliance performance could be measured with accounting data from financial reports, it is often impossible to isolate the effects of an alliance on accounting performance from other corporate actions and events because financial reports are only published periodically (quarterly or yearly). CAR is generally immune to such confounding effects because of the short time window it employs; therefore, we can establish a tighter link between an alliance and its outcome using the CAR. Several prior studies on performance feedback have adopted CAR for the same reasons (e.g., Haleblian et al., 2006; Kim et al., 2015; Mishina, Dykes, Block, & Pollock, 2010). We tested the robustness of our findings using accounting data (quarterly return on assets) instead of CAR, and obtained consistent results.
其次，虽然可以通过财务报告中的会计数据来衡量联盟绩效，但由于财务报告仅定期发布（季度或年度），通常无法将联盟对会计绩效的影响与其他公司行为和事件分离开来。CAR 由于采用的时间窗口较短，通常不受此类混杂效应的影响；因此，我们可以利用 CAR 在联盟与其结果之间建立更紧密的联系。出于同样的原因，先前一些关于绩效反馈的研究也采用了 CAR（例如，Haleblian 等，2006；Kim 等，2015；Mishina, Dykes, Block, & Pollock, 2010）。我们使用会计数据（季度资产回报率）而非 CAR 测试了研究结果的稳健性，并得到了一致的结果。

CAR measures the total return surplus or shortage relative to the expected returns based on an economic model over a predefined period. ${}^4$${}^4$^(4){ }^{4} We used a short event window of $[-1,+1]$$[-1,+1]$[-1,+1][-1,+1] (Diestre & Rajagopalan, 2014; Haleblian et al., 2006) to measure stock market reactions to an alliance, for three reasons: (a) it captures most of the stock price movement, as the largest price movement is typically observed on the first day after the announcement; (b) boundedly rational managers may focus on the most salient stock price changes; and © a longer window may confound the effects of other corporate actions and events. We obtained consistent results from five alternative windows ([-10, 1], $[-10,3],[-5,5],[-3,3]$$[-10,3],[-5,5],[-3,3]$[-10,3],[-5,5],[-3,3][-10,3],[-5,5],[-3,3], and $[0,2]).{}^5$$[0,2]).{}^5$[0,2]).^(5)[0,2]) .{ }^{5}
CAR 衡量的是在预定义时期内，相对于基于经济模型的预期回报的总回报盈余或短缺。 ${}^4$${}^4$^(4){ }^{4} 我们使用了 $[-1,+1]$$[-1,+1]$[-1,+1][-1,+1] 的短期事件窗口（Diestre & Rajagopalan, 2014; Haleblian et al., 2006）来衡量市场对联盟的反应，原因有三：(a) 它捕捉了大部分的股价变动，因为通常在公告后的第一天观察到最大的价格变动；(b) 有限理性的管理者可能会关注最显著的股价变化；以及© 更长的窗口可能会混淆其他公司行动和事件的影响。我们从五个替代窗口（[-10, 1], $[-10,3],[-5,5],[-3,3]$$[-10,3],[-5,5],[-3,3]$[-10,3],[-5,5],[-3,3][-10,3],[-5,5],[-3,3] , 和 $[0,2]).{}^5$$[0,2]).{}^5$[0,2]).^(5)[0,2]) .{ }^{5} ）中获得了一致的结果。
Alliance performance feedback. Previous studies on performance feedback have often used two types of aspiration levels: historical and social (Greve & Gaba, 2017; Posen et al., 2018). We measured alliance performance feedback based on the social aspiration level. Since this level is formed
联盟绩效反馈。先前关于绩效反馈的研究通常使用两种类型的抱负水平：历史的和社会的（Greve & Gaba, 2017; Posen 等, 2018）。我们基于社会抱负水平来衡量联盟绩效反馈。由于这一水平形成

based on comparisons with the contemporary performance of similar firms, it reflects current business conditions more accurately than the historical aspiration level, and managers consider it a more up-to-date and relevant representation of their firm’s ability to make acquisitions (Arrfelt et al., 2013; Hu, He, Blettner, & Bettis, 2017; Short & Palmer, 2003). Further, the performance of an alliance is determined not only by the focal firm but also by its alliance partners. As the partners change from one alliance to another, the historical performance comparison of alliances with different partners may not be as meaningful as social comparisons. Additionally, the models based on social aspirations provide the best model fit compared to when the historical aspiration level and the weighted aspiration level are used. However, our hypotheses remained supported when they were tested with measures based on these alternative aspiration levels (see the “Robustness Check” section for more details).
基于与类似公司当前表现的比较，它比历史期望水平更准确地反映了当前的商业状况，管理者认为这是对其公司进行收购能力的更现代和相关的体现（Arrfelt 等，2013；Hu, He, Blettner, & Bettis, 2017；Short & Palmer, 2003）。此外，联盟的表现不仅由焦点公司决定，还由其联盟伙伴决定。随着合作伙伴从一个联盟变为另一个联盟，与不同伙伴的历史表现比较可能不如社会比较有意义。另外，与使用历史期望水平和加权期望水平相比，基于社会期望的模型提供了最佳的模型拟合度。然而，当我们用基于这些替代期望水平的措施测试时，我们的假设仍然得到支持（更多详情请参见“稳健性检验”部分）。
The social aspiration level of a firm is measured by the average of CAR of every alliance made by all other firms (not only by Fortune 500 firms) in the same industry (defined by 2-digit SIC codes of the firm’s primary industry) during the year prior to the focal alliance. ${}^6$${}^6$^(6){ }^{6} A spline function was used to construct the alliance performance above and below the aspiration level (e.g., Greve, 1998; Lungeanu, Stern, & Zajac, 2016). $ALL<ASP$$ALL<ASP$ALL < ASPA L L<A S P represents alliance performance below the aspiration level. It was measured by the CAR of the focal alliance minus the social aspiration level if the CAR of the focal alliance was below the aspiration level; it equals 0 if the CAR was higher than the aspiration level. $ALL>ASP$$ALL>ASP$ALL > ASPA L L>A S P represents alliance performance above the aspiration level. It was measured by the CAR of the focal alliance minus the aspiration level if the CAR of the focal alliance exceeded the aspiration, while it equals 0 if CAR was lower than the aspiration level. ${}^7$${}^7$^(7){ }^{7}
企业的社会抱负水平是通过同一行业内（由企业主要行业的两位 SIC 代码定义）所有其他企业（不仅仅是财富 500 强企业）在焦点联盟形成前一年内每项联盟的 CAR 平均值来衡量的。 ${}^6$${}^6$^(6){ }^{6} 使用样条函数构建了高于和低于抱负水平的联盟绩效（例如，Greve, 1998; Lungeanu, Stern, & Zajac, 2016）。 $ALL<ASP$$ALL<ASP$ALL < ASPA L L<A S P 表示低于抱负水平的联盟绩效。如果焦点联盟的 CAR 低于抱负水平，则通过焦点联盟的 CAR 减去社会抱负水平来衡量；如果 CAR 高于抱负水平，则等于 0。 $ALL>ASP$$ALL>ASP$ALL > ASPA L L>A S P 表示高于抱负水平的联盟绩效。如果焦点联盟的 CAR 超过抱负水平，则通过焦点联盟的 CAR 减去抱负水平来衡量；如果 CAR 低于抱负水平，则等于 0。 ${}^7$${}^7$^(7){ }^{7}
Number of alliance partners. Alliance partner number represents the total number of partners that
联盟合作伙伴数量。联盟合作伙伴数量代表合作伙伴的总数

participated in the focal alliance, and was measured accordingly.
参与了焦点联盟，并相应进行了测量。

Alliance experience of partners. Partner alliance experience represents the perceived alliance capabilities of the partners in an alliance and was measured by the average of the total number of alliances that were completed by each partner during a fiveyear period prior to the focal alliance. The results remained consistent when this variable was measured using alternative time periods (four-, seven-, and 10-year windows) (Diestre & Rajagopalan, 2012).
合作伙伴的联盟经验。合作伙伴的联盟经验代表了联盟中合作伙伴感知到的联盟能力，并通过每个合作伙伴在焦点联盟前五年期间完成的所有联盟总数量的平均值来衡量。当使用不同的时间段（四年、七年和十年窗口）来衡量这一变量时，结果保持一致（Diestre & Rajagopalan, 2012）。

Alliance performance variability. Alliance performance variability represents the degree to which a focal firm’s past alliance performances fluctuated, and was measured by the standard deviation of the performance of all the alliances that a firm made during the five-year period prior to the focal alliance. The results remained consistent when four, seven-, and 10-year windows were used (Diestre & Rajagopalan, 2012).
联盟绩效变异性。联盟绩效变异性代表了一个焦点公司过去联盟绩效的波动程度，并通过该公司在焦点联盟前五年内所有联盟绩效的标准差来衡量。当使用四年、七年和十年窗口时，结果保持一致（Diestre & Rajagopalan, 2012）。

We included seven firm-level control variables that could affect a firm’s choice between alliances or acquisitions. Since firms may repeat the governance mode with which they have more experience (e.g., Zollo & Reuer, 2010), we included acquisition experience and alliance experience, measured by the total number of prior acquisitions and alliances that were completed by a firm during the previous five years, respectively. Since strong financial performance may motivate firms to pursue active growth via acquisitions (Kim, Haleblian, & Finkelstein, 2011), we included firm performance, measured by the return on equity. A firm’s divestiture may release resources and management capacity, increasing the firm’s appetite for acquisitions (Kuusela, Keil, & Maula, 2017). We thus included a firm’s recent divestiture experience, calculated by the number of divestitures completed by the firm during the previous five years. Slack resources may also increase a firm’s preference for acquisitions. We controlled for three types of slack resources (Bromiley, 1991; Greve, 2003a; Kuusela et al., 2017): financial slack, which was measured by the equity-to-debt ratio; unabsorbed slack, which was measured by the ratio of working capital (current assets minus current liabilities) to total sales; and absorbed slack, which was calculated by the ratio of selling, general, and administrative expenses to sales.
我们纳入了七个可能影响企业在联盟或收购之间做出选择的公司层面控制变量。由于企业可能会重复使用他们更有经验的治理模式（例如，Zollo & Reuer, 2010），我们分别通过企业在过去五年内完成的收购和联盟总数来衡量其收购经验和联盟经验。由于强劲的财务表现可能会激励企业通过收购积极寻求增长（Kim, Haleblian, & Finkelstein, 2011），我们纳入了以股本回报率衡量的企业绩效。企业的剥离可能会释放资源和管理能力，从而增加其对收购的胃口（Kuusela, Keil, & Maula, 2017）。因此，我们纳入了企业最近的剥离经验，通过企业在过去五年内完成的剥离数量来计算。闲置资源也可能增加企业对收购的偏好。 我们控制了三种类型的闲置资源（Bromiley, 1991; Greve, 2003a; Kuusela et al., 2017）：财务闲置，通过权益负债比衡量；未吸收闲置，通过营运资本（流动资产减去流动负债）与总销售额的比率衡量；以及吸收闲置，通过销售、一般及行政费用与销售额的比率计算。

We controlled for diversification, measured by the number of different SIC codes reported by a firm,
我们控制了多元化程度，这是通过公司报告的不同 SIC 代码数量来衡量的，
because more diversified firms may be predisposed to make acquisitions over alliances (Lungeanu et al., 2016). Average past acquisition performance of a firm was included, as a strong performance in prior acquisitions might make firms prefer acquisitions to alliances. It was measured through the average of CARs of all acquisitions that a firm had made during the previous five years. Firms that are considered to be attractive alliance partners are highly sought after by other firms and are more likely to make alliances. The performance of a firm’s alliance partners is a good proxy of its attractiveness as an alliance partner (Fombrun, 1996; Obloj & Capron, 2011; Villalonga & McGahan, 2005). We included partner firm performance, calculated by the average of the return on the equity of the alliance partners in all the alliances that the focal firm had made during the previous five years.
因为更多元化的公司可能更倾向于通过收购而非联盟（Lungeanu 等，2016）。我们纳入了公司过去收购的平均表现，因为之前收购中的强劲表现可能使公司更偏好收购而非联盟。这是通过公司在过去五年内所有收购的累计异常收益率（CARs）的平均值来衡量的。被视为有吸引力的联盟伙伴的公司会备受其他公司追捧，并更有可能建立联盟。公司联盟伙伴的表现是衡量其作为联盟伙伴吸引力的良好指标（Fombrun，1996；Obloj & Capron，2011；Villalonga & McGahan，2005）。我们包含了伙伴公司的表现，这是通过焦点公司在过去五年内所有联盟中联盟伙伴的股本回报率的平均值来计算的。
Finally, a firm’s decision regarding acquisitions and alliances can be influenced by overall industry trends (Kim et al., 2015), so we included industry acquisition density and industry alliance density, measured by the total number of acquisitions and alliances made during the previous five years in the industry to which the focal firm belonged, respectively. ${}^{\text{.}}$${}^{\text{. }}$^(". "){ }^{\text {. }}
最后，企业关于收购和联盟的决策可能受到整个行业趋势的影响（Kim 等，2015），因此我们分别纳入了行业收购密度和行业联盟密度，这两个指标是通过焦点企业所属行业在过去五年内进行的收购和联盟总数来衡量的。 ${}^{\text{.}}$${}^{\text{. }}$^(". "){ }^{\text {. }}

We summarize our predictions in Table 1. Table 2 shows the descriptive statistics and zero-order correlations of the variables used for the main analysis. The correlations are low to moderate, and none of the models have a variance inflation factor over 3.7, which is well below 10 , a standard value that is often used to detect multicollinearity. We also did not observe any instability in the coefficients when the independent and moderating variables were entered separately. Overall, we detected no potential multicollinearity issue.
我们在表 1 中总结了我们的预测。表 2 展示了用于主要分析的变量的描述性统计和零阶相关性。相关性从低到中等不等，且所有模型的方差膨胀因子均未超过 3.7，远低于常用于检测多重共线性的标准值 10。此外，当独立变量和调节变量分别输入时，我们也没有观察到任何系数不稳定的情况。总体而言，我们没有发现潜在的多重共线性问题。
Table 3 presents the main results for testing the hypotheses. Model 1 only includes control variables, while the variables for performance feedback for alliances are added to Model 2. Three moderating variables are added to Models 3-5. ${}^9$${}^9$^(9){ }^{9}
表 3 展示了假设检验的主要结果。模型 1 仅包含控制变量，而模型 2 则加入了联盟绩效反馈的变量。模型 3 至模型 5 中分别加入了三个调节变量。 ${}^9$${}^9$^(9){ }^{9}

TABLE 1  表 1

TABLE 2  表 2
Descriptive Statistics and Correlations
描述性统计与相关性

Note: Correlations larger than $|.18|$$|.18|$|.18||.18| are significant at $p<0.01$$p<0.01$p < 0.01p<0.01; those larger than $|.14|$$|.14|$|.14||.14| are significant at $p<0.05$$p<0.05$p < 0.05p<0.05; those larger than $|.11|$$|.11|$|.11||.11| are significant at $p<0.1$$p<0.1$p < 0.1p<0.1.
注意：大于 $|.18|$$|.18|$|.18||.18| 的相关性在 $p<0.01$$p<0.01$p < 0.01p<0.01 水平上显著；大于 $|.14|$$|.14|$|.14||.14| 的在 $p<0.05$$p<0.05$p < 0.05p<0.05 水平上显著；大于 $|.11|$$|.11|$|.11||.11| 的在 $p<0.1$$p<0.1$p < 0.1p<0.1 水平上显著。

TABLE 3  表 3
Estimation for the Likelihood of an Alliance or Acquisition
联盟或收购可能性的估计