Introduction

“The mounting evidence clearly demonstrates that this tragedy was preventable and the direct result of BP’s reckless decisions and actions. Frankly, we are shocked by the publicly available information that […] indicates BP operated unsafely and failed to monitor and react to several critical warning signs during the drilling of the Macondo well. BP’s behavior and actions likely represent gross negligence or willful misconduct and thus affect the obligations of the parties under the operating agreement” (Korosec, 2010, June 21)

This statement by the Chairman and CEO of Anadarko Petroleum, Jim Hackett, in the wake of the Deepwater Horizon disaster, illustrates one possible response to the threat of negative reputation spillover from an alliance partner accused of environmental misconduct.

It is a widely observed phenomenon that one firm’s preventable crisis or scandal (Coombs 2015) contaminates and causes reputation damage to related firms or those with similar characteristics (Comyns and Franklin-Johnson 2018; Jonsson et al. 2009; King et al. 2002). In alliances (Yu and Lester 2008), the response strategy alternatives in such reputation spillover crises are well defined (Bruyaka et al. 2018). Confronted with a risk of contagion (Laufer and Wang 2018) as well as pressure from various stakeholder groups (King et al. 2002), the firm (the focal firm hereafter) must decide whether to adopt supportive or non-supportive response (Jensen 2006); a non-response will soon be interpreted as silent support (Farrell 1983).

On the one hand, the focal firm may decide to criticize, distance itself from, and sever ties with its partner or, in other words, rapidly de-escalate its commitment (Drummond 1995) to the strategic alliance (i.e., interfirm collaborative agreement aimed at generating competitive advantage through pooling imperfectly tradable resources; Das and Teng 2000; Drees and Heugens 2013; Hoetker and Mellewigt 2009). We label this an ‘adversary response’. Here, the goals are to protect the firm’s ethical reputation (Baselga-Pascual et al. 2018) and to avoid involvement in the crisis or “tarring by the same brush” (King et al. 2002, p. 397) as well as the associated collective responsibility that exceeds the alliance’s scope and contractual obligations.

On the other hand, the focal firm may feel compelled to assist or defend its crisis-stricken partner, even if continued endorsement can lead “the public to perceive the endorsers as conspirators in the events” (Rhee and Valdez 2009, p. 164), thus increasing the risk of crisis contagion (Laufer and Wang 2018). The main underlying motive is protecting expected gains and the resources shared within the alliance. Such persistence with the alliance despite negative feedback (Brockner 1992; Delios et al. 2004), and delay of an eventual exit decision, is a type of commitment escalation (Hantula and Bragger 1999). We label this an ‘advocacy response’.

These opposing responses represent a cost–benefit trade-off (Bruyaka et al. 2018) and an escalation-decision dilemma for the focal firm (Staw 1976). Under which conditions will the focal firm prefer an adversary/advocacy response? According to Bruyaka et al. (2018, p. 446), this understudied “question is of paramount importance because it is essential for managers to anticipate their alliance partner’s behavior (Tjemkes and Furrer 2010) when their organization is stricken by an adverse event and is thus in a vulnerable position.”

Prior escalation research found alliances to stimulate escalation behavior (Delios et al. 2004; Sleesman et al. 2018), as they allow firms to access partners’ valuable resources and thereby reduce their own dependence on external resources (i.e., those of other firms and the market; Casciaro and Piskorski 2005; Hillman et al. 2009). Simultaneously and somewhat paradoxically, alliances tend to create “new patterns of interdependence” (Drees and Heugens 2013, p. 1670) including reputational interdependence (Barnett and Hoffman 2008)—an alliance formation incentive when reputation is positive, which makes high-reputation firms or those with high status or legitimacy more attractive (Norheim-Hansen 2015; Stuart 1998; Sullivan et al. 2007)Footnote 1.

However, when a partner is accused of misconduct, the reputational interdependence becomes problematic for the focal firm—a reputation commons problem (King et al. 2002)—and could contribute to de-escalation of commitment (Drummond 1995). Consequently, in this context, prior findings on escalation behavior in alliances (Delios et al. 2004; Sleesman et al. 2018) are not fully applicable. Stated differently, factors found to have a commitment escalation effect need to be reconsidered in the context of a reputation spillover crisis.

We address this gap by exploring how previously identified impactful escalation factors affect the focal firm’s adversary/advocacy response. Specifically, we argue for a reverse effect in our crisis contagion context, and four hypotheses are derived from this main proposition. Our theoretical reasoning mainly integrates insights from two complementary theoretical perspectives. First, we draw on the extended escalation theory, also known as decision dilemma theory, which states that high equivocality about the (reputation) loss outcome (i.e., it is very difficult to tell whether any significant negative reputation spillover will, in fact, occur) is necessary for escalation to occur (Bowen 1987; Drummond 1995; Hantula and Bragger 1999). Second, considering that we estimate such equivocality or uncertainty “through the lens of interdependence” (Sleesman et al. 2018, p. 199), we also build on resource dependence theory (Pfeffer and Salancik 1978).

To test the hypotheses, we use a policy-capturing method and develop scenarios concerning an environmental preventable crisis or scandal. The respondents are CEOs of Norwegian manufacturing firms. Globally, our study answers a call for a more nuanced understanding of escalation and the theory’s boundary conditions by introducing reputation spillover crisis as a contextual influencer of escalation behavior. It also extends reputation literature and provides new evidence that reputation concerns can instigate ethical decision-making (O’Fallon and Butterfield 2005). Specifically, it offers additional evidence of the value attributed to ethical reputation (Baselga-Pascual et al. 2018) and the effect it can have on alliances. We discuss further theoretical contributions, as well as managerial implications, in the discussion and conclusion section of this article.

Background Literature

Corporate Reputation: Ethical Reputation to the Front Stage

Corporate reputation can be defined as “a relatively stable, issue specific aggregate perceptual representation of a company’s past actions and future prospects compared against some standard” (Walker 2010, p. 370). A positive reputation is a strategic resource, viewed by CEOs as among those making “the most important contribution to business success” (Hall 1992, p. 135). Its strong influence on firm performance comes from the enhanced willingness of others to contract with reputable firms—whether that be customers, employees, investors or alliance partners (Dollinger et al. 1997; Sullivan et al. 2007). This is because good reputations serve as quality signals for both products/services and firm behavior. Accordingly, they provide a “reservoir of goodwill” (Zavyalova et al. 2016, p. 254) with stakeholders, valuable when there is a negative event.

In today’s business context, a reputational dimension that has arrived at the front stage is a reputation for ethical behavior (Baselga-Pascual et al. 2018)—especially related to environmental and social conduct. Widespread concern about, for instance, climate change has made stakeholder sanctions such as protests, boycotts and lawsuits vis-à-vis firms accused of misconduct ever more commonplace (King et al. 2002). Therefore, when firms face negative reputation spillover crises in which their ethical reputation can be damaged due to a partner’s unethical behavior, it represents a threat to organizational performance (Hall 1992, 1993).

Reputation Spillover and Response Strategies

Reputation spillover occurs when one firm’s negative or positive reputation on some issue spreads to another firm with similar characteristics or to a closely related firm (Goins and Gruca 2008; Jonsson et al. 2009). In alliances, such spillover is due to interorganizational proximity signaling a certain overlap in strategies, orientations, or values as well as interorganizational endorsement (Stuart et al. 1999; Yu and Lester 2008). Consequently, the partnering firms are, to some extent, “painted with the same brush” (Barnett and Hoffman 2008, p. 1). This general social-categorization mechanism (Turner 1985), which results from stakeholders lacking information to being able to “fully distinguish the relative quality or performance of each firm” (King et al. 2002, p. 395), is comparable for positive and negative reputation spillovers. However, distinct research streams address these opposing outcomesFootnote 2.

The research stream on positive reputation and status spillovers centers largely on alliance formation. Ever since the publication of Stuart’s influential work in 1998, a number of studies showed that reputable firms are attractive partners (Dollinger et al. 1997) due, in part, to anticipated reputation spillovers (Norheim-Hansen 2015).

Negative reputation spillovers have been extensively explored at the industry level (Fauchart and Cowan 2014; Paruchuri and Misangyi 2015)—that is, between firms operating in the same industry. However, they were rarely examined at the interfirm alliance level (Yu and Lester 2008). Therefore, we know little about the reactions of partnering firms when faced with a threat of negative spillovers (Bruyaka et al. 2018; Jensen 2006). Such a threat is present whenever an alliance partner faces public accusations of misconduct or any preventable crisis (Coombs 2012, 2015). Metaphorically speaking, in this context of crisis contagion (Laufer and Wang 2018), an alliance is the conduit through which a dangerous disease (i.e., negative ethical reputation) can spread from a crisis-plagued firm to the focal firm.

Furthermore, given that various stakeholders are awaiting a response (Jensen 2006), the focal firm is required to adopt non-supportive or supportive behavior vis-à-vis its criticized partner. A wait-and-see strategy or initial neutral non-response will soon be interpreted as a silent supportive response (Bruyaka et al. 2018; Farrell 1983)Footnote 3. The threat of negative reputation spillover and pressures from external and internal stakeholders (Hillman et al. 2009) induce the focal firm to reassess its commitment to the alliance and decide on a response to the threat.

Extant theoretical insights suggest that an adversary response mitigates reputational dependence, and thus limits costly reputation loss and increases societal acceptance—which are important organizational motives (Drees and Heugens 2013; Jensen 2006). At the same time, however, an adversary response inflicts costs and financial losses related to severing ties. These include overcoming reintroduced external resource dependence (Hillman et al. 2009) or costs related to finding a new partner (Sullivan et al. 2007). An advocacy response should have the opposite effect (Rhee and Valdez 2009) and allow the focal firm to continue to enjoy financial gains from the alliance. In sum, as claimed earlier, these alternative responses represent a cost–benefit trade-off (Bruyaka et al. 2018) in the form of an escalation-decision dilemma (Staw 1976).

Escalation of Commitment and Its Determinants

Commitment escalation occurs when decision-makers “maintain or increase commitment to their goal” (Sleesman et al. 2018, p. 178) despite negative feedback (Brockner 1992; Staw 1976, 1981). It is a cognitive bias affecting loss and gains perspectives (Tversky and Kahneman 1979). Generally, the greater the investments made in terms of time, money, or effort, the stronger the escalation tendency (Sleesman et al. 2018.). This tendency has been observed in a wide variety of organizational contexts, including R&D (Conlon and Parks 1987), labor-management negotiations (Ross 1998), acquisitions (Meschi and Métais 2015), and alliances (Delios et al. 2004). Accordingly, an array of explanatory factors was identified and empirically supported (Sleesman et al. 2012).

In the alliance context, typically, known escalation determinants hint at the ‘strength’ of the alliance or ‘thickness’ of the interfirm tie—and three prominent factors can be extracted from prior literature (Delios et al. 2004): (1) strategic importance of the alliance, (2) alliance termination costs (both project-related factors; Staw and Ross 1987) and (3) personal responsibility of the focal firm’s CEO in initiating the alliance (a psychological/social factor; Ibid.).

However, we argue that factors previously found to be positively associated with escalation instead lead to de-escalation in our context of a reputation spillover crisis, as they strengthen reputational interdependence and thus reputational cost. In the following, we elaborate this proposition by describing each of our determinants and predicting each’s effect on the adversary/advocacy decision. A key assumption in these developments is focal firm’s acknowledgment of the fact that costs can be both economic and non-economic (Hall 1992, 1993).

Yet, first, we add a fourth highly relevant (de-)escalation determinant that directly affects the magnitude of the reputational cost, namely the focal firm’s reputation for management quality. To explain, a good reputation “may be a burden under particular circumstances” (Zavyalova et al. 2016, p. 254) and aggravate negative spillover, since negative events garner more attention when reputable firms are involved and such firms are held to higher standards by stakeholders—thus they are more severely sanctioned due to expectancy violation (Rhee and Valdez 2009). We extend the reverse effect proposition to this fourth factor, as detailed in the next section where we develop our hypotheses. An overarching theoretical model is presented in Fig. 1.

Fig. 1
figure 1

Overarching theoretical model

Full size image

Hypotheses

Escalation Determinants: Effects on Advocacy/Adversary Response

Strategic Importance of the Alliance

Strategic importance of the alliance pertains to whether the alliance is “strongly linked to [the focal firm’s] organizational values or missions” (Delios et al. 2004, p. 465). High strategic importance indicates that the alliance greatly allows to reduce external resource dependence (Hillman et al. 2009)—as the crisis-stricken partner contributes resources to the alliance that the focal firm needs to maintain or build a solid competitive position (Das and Teng 2000). If the focal firm decides to sever ties, the estimated economic costs are elevated. Accordingly, prior research found high strategic importance to be a forceful driver of escalation, suggesting an advocacy response.

However, in our reputation spillover context, we argue that significant non-economic costs in the form of reputation damage (which subsequently generates economic costs) accompanies an advocacy response under the high strategic importance condition. Specifically, this condition signifies a strong alliance acting as a powerful conduit for negative ethical reputation from the accused firm to its partner (Yu and Lester 2008).

First, as stakeholders expect more care and resources to be devoted to establishing (e.g., selecting partners and undertaking due diligence; Mitsuhashi 2002) and managing alliances of high strategic importance, the perceived interorganizational endorsement is elevated (Yu and Lester 2008). Continued endorsement through an advocacy strategy can lead “the public to perceive the endorsers as conspirators in the events” (Rhee and Valdez 2009, p. 164); thus clearly leading to crisis contagion or significant ethical reputation loss.

Second, while other arrangements for reducing resource dependence (e.g., equity joint ventures and acquisitions) are generally more visible than contractual alliances (Drees and Heugens 2013), high strategic importance contributes to making the interfirm tie more externally perceivable. Firms will communicate more about the alliance in strategy-related communications (Higgins 2002)—and key stakeholders such as employees, customers and shareholders will show stronger interest.

Importantly, stakeholders are also more affected by the advocacy/adversary decision and will sanction any supportive behavior by the focal firm (Goins and Gruca 2008; Jonsson et al. 2009), thus activating negative spillover (King et al. 2002). Especially high-status stakeholders (Perrault 2017) will react negatively towards an advocacy response; this, partly since their own reputation may in turn be affected (Sullivan et al. 2007). There is a real risk that these preferred stakeholders (due to their status) leave the focal firm (Ibid.); even addressing them individually (McVea and Freeman 2005) will hardly lead to acceptance and alignment with an advocacy response. Accordingly, accountability to important audiences (Hillman et al. 2009; Perrault 2017) was found to make firms more likely to sever ties with an affiliate having committed an unethical act (Jensen 2006).

Overall, under the high strategic importance condition, we argue that the core anticipated costs of an advocacy response come from reputation loss, and that these will be considered very significant. Therefore, we contend that the focal firm’s loss and gains perspectives will be asymmetric (Tversky and Kahneman 1979) in the sense that ethical reputation loss will dominate the decision (Shimizu 2007; Sullivan et al. 2007). Stated differently, we expect that the anticipation of a greater loss with advocacy response curbs escalation (Drummond 2014).

For an alliance of low strategic importance, on the other hand, we expect the focal firm to anticipate smaller loss with advocacy than adversary response—and thus react differently (Shimizu 2007). Specifically, jeopardizing gains from the alliance (even if they are less important than under the condition above) will be considered more costly than an eventual reputation loss. There is, in fact, high equivocality as to whether the focal firm will experience any significant reputational damage at all, since the fortifying forces for negative reputation spillover described above are much weaker. Thus, high equivocality about the loss outcome, and the aforementioned fact that alliances stimulate escalation behavior (they always signify at least a moderate level of commitment; Bowen 1987; Drummond 1995), suggest that the focal firm selects an advocacy response. On this basis, we formulate our first hypothesis:

Hypothesis 1

In a (negative) reputation spillover crisis, the focal firm is more likely to adopt an adversary response when the alliance’s strategic importance is high and more likely to adopt an advocacy response when the alliance’s strategic importance is low.

Alliance Termination Costs

Alliance termination costs are the costs “of closing an alliance” (Delios et al. 2004, p. 465). High termination costs result from formally agreed and extensive obligations in the partnership, with high penalties for contract breach (Hoetker and Mellewigt 2009). Exit clauses prevent unilateral and unexpected decisions to leave the alliance without adverse financial consequences. A departing firm may also face lawsuits. Consequently, high alliance termination costs often prompt escalation (Delios et al. 2004), suggesting an advocacy response.

However, in our reputation spillover context, we argue that substantial non-economic costs in the form of reputation loss accompanies an advocacy response under the high alliance termination costs condition. Specifically, this condition (like high strategic importance) signifies a strong alliance, acting as a forceful contamination vehicle. The main underlying causes are similar to those discussed above.

First, stakeholders should presume that the due diligence processes associated with an alliance design where there are high termination costs revealed various aspects of the partner’s strategies, orientations, and values (Mitsuhashi 2002). Such costs are thus testament to important interorganizational commitments and proximity (Hoetker and Mellewigt 2009). An advocacy strategy further strengthens these endorsement signals, propelling the focal firm into a collective responsibility situation or “tarring by the same brush” (King et al. 2002, p. 397); suggesting considerable ethical reputation loss.

Second, an extensively formalized alliance creates more visible interfirm linkages than a looser, arm’s-length alliance (Drees and Heugens 2013)—increasing the risk of stakeholder attention and sanctions (King et al. 2002) if a supportive response towards a partner accused of misconduct is implemented; as argued above.

Overall, under the high termination costs condition—consistent with our previous reasoning—we argue that the core costs associated with an advocacy response come from ethical reputation loss. At the same time, we also highlighted earlier that an advocacy response allows for protecting expected gains and the resources shared within the alliance. Yet, again, we expect that even an overvaluation of these gains will be triumphed by the motivation to avoid ethical reputation loss (Tversky and Kahneman 1979); i.e., the loss and gains perspectives will be asymmetric, with loss aversion as a prevailing motivator (Shimizu 2007). This argument is here further supported by the focal firm’s ability to eradicate some of “the obligations of the parties under the operating agreement” (Anadarko Chairman and CEO) due to the partner’s misconduct, and, thereby somewhat reduce the costs of the adversary response. Consequently, the anticipation that an advocacy response engenders a greater loss curbs escalation (Drummond 2014).

On the other hand, the reputational risk is lower when termination costs are low and (as for the low strategic importance condition), we expect the focal firm to anticipate smaller loss with advocacy than adversary response. We thus predict a different reaction (Shimizu 2007), consistent with the developments for hypothesis 1; the focal firm prefers an advocacy response. Hence, our next hypothesis reads as follows:

Hypothesis 2

In a (negative) reputation spillover crisis, the focal firm is more likely to adopt an adversary response when the alliance termination costs are high and more likely to adopt an advocacy response when the alliance termination costs are low.

Personal Responsibility of the Focal Firm’s CEO in Initiating the Alliance

Personal responsibility of the focal firm’s CEO in initiating the alliance refers to whether the current CEO was the main proponent for alliance formation (Graebner 2009). A confirmative answer signifies that s/he (and not a predecessor) was involved in the partner selection decision, due diligence process, and in negotiations concerning the alliance’s governance. In this case, a decision to sever ties means admitting to a poor partner choice (Shah and Swaminathan 2008) and “strong ego threat” (Sleesman et al. 2012, p. 554). Personal responsibility creates a “need to restore the ‘appearance’ of rationality [of poor decisions]” (Staw 1976, p. 40). This responsibility effect is a powerful driver of commitment escalation (Zhang and Baumeister 2006), suggesting an advocacy response.

However, we argue that considerable reputational costs accompanies an advocacy response under the personal responsibility condition. Specifically, this condition denotes a strong alliance which acts as a potent conduit for transmitting negative ethical reputation from the accused partner to the focal firm. As regards underlying causes, they are again mainly twofold.

First, personal responsibility of the focal firm’s CEO in initiating the alliance indicates a closer relationship between the two firms and their CEOs, as well as high interorganizational endorsement (Yu and Lester 2008). Extensive interactions over a certain period of time make the partners “appear as peers” (Veil et al. 2016, p. 325). Therefore, the risk of stakeholders attributing moral co-responsibility in the crisis is elevated, resulting in pressure and sanctions (King et al. 2002; Perrault 2017).

Second, in strategy-related communications (Higgins 2002), a CEO is arguably more likely to talk about an alliance that s/he personally initiated, thereby increasing its visibility among internal and external audiences (Jensen 2006).

Overall, under the personal responsibility condition, we argue that the principal costs associated with an advocacy response come from ethical reputation loss here as well. At the same time, an advocacy response allows to continue enjoying financial gains from the alliance, and the CEO to rationalize his/her initial decision (Staw 1976, 1981). Yet, as before, we contend that the incentive to avoid ethical reputation loss will dominate the decision (Shimizu 2007); possible overvaluation of gains will be countered by loss aversion (Tversky and Kahneman 1979). This argument is further supported by the fact that the CEO’s own reputation and reputation of the firm are intertwined (Graebner 2009)—and proof of ethical decision-making (i.e., an adversary response) should compensate for the burden of acknowledging a poor partner choice. Consequently, the focal firm estimates that an advocacy response will engender a bigger loss and this curbs escalation (Drummond 2014).

In contrast, if the CEO’s predecessor initiated the alliance, the reputational risk is again much lower compared to the opposite condition. Reducing reputational dependence is then not a main priority, and there is an anticipated smaller loss with advocacy than adversary response. Accordingly, the focal firm sees staying committed to the alliance (Bowen 1987) and continuing to mitigate external resource dependence as the better option (Drummond 2014). We thus predict the following:

Hypothesis 3

In a (negative) reputation spillover crisis, the focal firm is more likely to adopt an adversary response if its CEO was personally responsible for the decision to initiate the alliance and more likely to adopt an advocacy response when its CEO was not personally responsible for the decision to initiate the alliance.

Focal Firm’s Reputation for Management Quality

Focal firm’s reputation for management quality. Dollinger et al. (1997, p. 140) define reputation for high-quality management as follows: “known for the high quality of its top managers and their integrity. The executives of the firm display concern for their community and are known as responsive to environmental concerns. They have a reputation for being able to attract, develop, and keep talented people.” Reputation is another previously identified escalation factor in the social category (Staw and Ross 1987), but it has received little attention in the commitment escalation literature (Sleesman et al. 2012). Social factors concern preoccupation with stakeholders’ perceptions, especially those of high status (Perrault 2017), and avoiding diffusion of negative impressions. Accordingly, a firm may maintain or increase commitment to a goal or a partner, despite negative feedback, if it allows to preserve a good management reputation (Sleesman et al. 2018)—which takes extensive time, money and effort to build. Any failing corporate strategy, including an alliance strategy, represents a threat to the focal firm’s management reputation, suggesting an advocacy response.

Moreover, the focal firm may decide to rely on the aforementioned “reservoir of goodwill” (Zavyalova et al. 2016, p. 254), valuable when there is a negative event. Specifically, research showed that reputation can act as a ‘buffer’, because audiences presume high-reputation firms possess a capability to deal with adverse situations more constructively (Rhee and Haunschild 2006). It thus functions as a shield against reputation contamination.

However, a good reputation can be both a benefit and a burden (Zavyalova et al. 2016). While reputation for high-quality management can “give the organization the benefit of the doubt” (Ibid., p. 254), it simultaneously gives rise to more stakeholder attention (King et al. 2002) and enhances stakeholder expectations (Rhee and Haunschild 2006; Rhee and Valdez 2009). Hence, an unethical decision—in this case an advocacy response according to most stakeholders (Sullivan et al. 2007)—will be perceived as a serious expectancy violation, resulting in heavier sanctions for a firm holding a high reputation (Ibid.). Consequently, we argue that substantial reputational cost accompanies an advocacy response under the reputation for high-quality management condition. Furthermore, while the focal firm may also suffer some reputation loss with an adversary response (due to a failed alliance strategy), it is arguably negligible compared with reputation loss associated with an advocacy response.

Overall, as high-reputation firms seek to maintain their standing in the eyes of stakeholders, and the focal firm anticipates greater reputation loss with an advocacy response (Drummond 2014), it adopts an adversary response. This is consistent with the results of Sullivan et al. (2007, p. 68), suggesting that faced with unethical acts, “firms with good reputations are most likely to leave the networks of others who have committed unethical acts.” Besides, reputation for high-quality management implies having many suitors (Jensen and Roy 2008), which should drive down costs of finding a new partner. Despite an alliance “gone awry” (Shafi et al. 2020), the focal firm will arguably preserve its attractiveness.

On the other hand, the focal firm may have a reputation for average management quality—that simply meets the industry standardFootnote 4. Here, with less scrutiny and less “vulnerable to stakeholder pressure” (King et al. 2002, p. 396), reputational risk is far lower. This prompts the focal firm to prioritize mitigating external resource dependence over abating reputational dependence. The focal firm views reducing its commitment to the alliance (Bowen 1987) as the costlier option (Drummond 2014), favoring escalation behavior. Consequently, our last hypothesis equally aligns with the aforementioned reverse effect proposition:

Hypothesis 4: In a (negative) reputation spillover crisis, the focal firm is more likely to adopt an adversary response when it has a reputation for management quality that is above the industry standard and more likely to adopt an advocacy response when it has a reputation for management quality that meets the industry standard.

Methods and Data

Policy-Capturing Method and Sample

We use a policy-capturing method to test our hypotheses, and collected primary data with a manipulated scenario-based survey directed at CEOs of Norwegian manufacturing firms. The policy-capturing method has a number of advantages for our research. First, it is particularly appropriate for studying managerial reactions to sensitive issues (e.g., misconduct and preventable accidents), since it substantially mitigates the norm-abiding or social acceptability bias that often affects traditional surveys. It also handles perceptual variables while limiting the threat of retrospective bias (Shah and Swaminathan 2008). Second, a manipulated scenario-based survey allows to present respondents with scenarios including multiple variables, and “force respondents to make judgements based on trade-offs” (Auspurg and Hinz 2015, p. 11). This multidimensionality creates complex decision-making contexts that simulate real-life situations. Third, this method allows to combine a traditional questionnaire using multi-item scales with a field experimental approach, thereby contextualizing respondents in real-life and manipulated scenarios (Karren and Barringer 2002).

There are also some disadvantages, including difficulty in attaining external validity, and the risk that respondents may not complete a survey comprised of numerous, repetitive scenarios. However, we largely overcame these disadvantages through pre-tested, well-designed, and realistic scenarios. The scenarios were based on real events and examples, and the scenarios and the measurement items were subjected to substantive pre-tests with policy-capturing method specialists and one cohort of executive MBA students.

To construct our sample, we extracted Norwegian manufacturing firms with more than 50 employees from a commercial database available through Proff Forvalt. 403 CEOs were invited to participate (and guaranteed confidentiality). Those agreeing received a link to the online scenario-based survey in a second email.

The respondents first read a text describing the decision-making context. They then reviewed short scenarios providing additional and hypothetical information on the alliance with the crisis-stricken partner and on reputational factors pertaining to the focal firm. We asked the respondents to judge the likelihood of selecting an adversary or advocacy response after reading each scenario; all four embedded variables had two levels and varied independently from one another (with intervariable correlations equal to zero). This is consistent with the assumption of orthogonality in the policy-capturing method.

In total, each CEO reviewed 16 non-randomized vignettes, which corresponds to a 2 × 2 × 2 × 2 within-respondent design (fully factorial and orthogonal; Auspurg and Hinz 2015)—as well as two duplicate test–retest scenarios to check whether the respondents exhibited consistent and reliable decision-makingFootnote 5. We assessed this within-respondent decision-making consistency using a mean test–retest correlation between the duplicate scenarios. The correlation coefficient for the full sample was 0.88, which reflects a high degree of within-respondent consistency and is similar to previous studies (0.82 in Choi and Shepherd 2004; 0.84 in Patzelt and Shepherd 2008; 0.87 in Shepherd et al. 2013)Footnote 6.

We collected 51 usable questionnaires, giving a response rate of 13% (16% when including questionnaires with missing responses). Explanatory factors for the rather low response rate include the nature and length of the surveyFootnote 7, as well as the respondents’ profile. Similar or even lower rates have been reported elsewhere (Connelly et al. 2016; Shepherd 1999). Moreover, our sample size is relatively similar to that of policy-capturing studies specifically surveying CEOs and top managers (64 CEOs in Mitchell et al. 2011; 66 top managers in Shepherd 1999). In total, with 16 decisions per respondent (the two duplicates are not included in the subsequent statistical analysis), the final sample corresponds to 816 decisions nested within 51 respondents—providing adequate statistical power. The data collection period ran from November 2016 to January 2017.

Decision-Making Context and Variables

As mentioned above, our online survey opened with a cover page describing the general decision-making context. First, we asked participants to imagine that they were the CEO of a hypothetical firm that had the same characteristics as their own firm. Next, we explained that they had a Norwegian alliance partner from the same industry and of approximately the same size. Thereafter, they were informed that this alliance partner was involved in an environmental scandal and accused of disseminating misleading information about its environmental performance—a crisis threatening the partner’s management reputation. In this regard, we drew upon the example of the Volkswagen crisis to fabricate the crisis described in the decision-making context. Moreover, we asked the respondents to visualize that they believed the partner was guilty and that its crisis management was unsatisfactory. The description stated that the alliance partner received a significant amount of negative media coverage, public criticism, and moral reprobation. Finally, we explained that the respondent’s (hypothetical) firm was in the spotlight because of its alliance with the accused partner and that there was a risk of negative reputation spillover.

The questionnaire then provided the respondents with additional information in 18 short scenarios, each of which was followed by two questions—one on the likelihood of selecting an advocacy response and another on the likelihood of selecting an adversary response. At the same time, we stated “that the attributes and environmental variables not specified in the decision profiles but possibly influencing their judgment should be considered as constant across all profiles” (Patzelt and Shepherd 2008, p. 1229).

Dependent Variables

The focal firm’s response is measured using two seven-point Likert scale items (measured with ordered categories: from 1 = strongly disagree to 7 = strongly agree), one corresponding to the advocacy response variable and the other to the adversary response variable. The two scale items are the following: “I will adopt an advocacy response, meaning that I will publicly defend and support my partner” and “I will adopt an adversary response, meaning that I will publicly take distance and criticize my partner.”

By splitting the response into two scale items, we allowed respondents to adopt the aforementioned initial neutral response by assigning a low value to both items. Although we explained that the firms must choose between a supportive or non-supportive strategy, respondents have the possibility to express a wait-and-see or no-choice position. Consequently, by means of our dual measure, we included this flexibility for respondents.

Independent Variables

Our four independent variables correspond to the four decision attributes embedded in the 18 scenarios. Consistent with the extant literature, two levels exist for each variable (Patzelt and Shepherd 2008).

The strategic importance of the alliance is a decision attribute pertaining to Hypothesis 1. Building on Das and Teng (2000), we depict a high level of this attribute as follows: “It is an alliance which is highly important for meeting the strategic goals of your firm, and thus the alliance which is attributed the most resources—compared with the firm’s other alliances.” When the level is low, we state the following: “It is an alliance which is not crucial for meeting the strategic goals of your firm, and thus is not the alliance which is attributed the most resources—compared with the firm’s other alliances.” We coded strategic importance as a binary variable (1 for high level and 0 for low level).

Alliance termination costs is the decision attribute addressed in Hypothesis 2. Drawing on Delios et al. (2004), we depict a high level as follows: “Contract-based restrictions and exit clauses are high with high penalties for unplanned one-sided termination.” When the level is low, we state the following: “Contract-based restrictions and exit clauses are moderate with low penalties for unplanned one-sided termination.” We coded alliance termination costs as a binary variable (1 for high level and 0 for low level).

The personal responsibility of the focal firm’s CEO is the decision attribute for Hypothesis 3. In the version in which the CEO is responsible for alliance formation and partner selection, we depict this variable using the following phrase: “Four years ago, with you as the main proponent.” In the version in which the CEO is not responsible, we use “Six years ago, with the previous CEO as the main proponent.” We coded personal responsibility as a binary variable (1 for yes and 0 for no).

Reputation for management quality is the decision attribute for Hypothesis 4. We follow Dollinger et al. (1997) in depicting a high level as follows: “Your company is known for the high quality of its top managers and their integrity. The executives of the company display concern for their community and are known as responsive to environmental concerns. They have a reputation for being able to attract, develop, and keep talented people.” To ensure external validity and create plausible scenarios, we depict a low level of this value as an ‘average’ reputation (as firms rarely ally with partners with poor reputations): “Your company has top managers of average (for its industry) quality and integrity. The executives of the firm display no more concern for their community than most other firms and are not known as particularly responsive to environmental concerns. They have an average reputation for being able to attract, develop, and keep talented people.” We coded reputation for management quality as a binary variable (1 for high level and 0 for low level).

Control Variables

Consistent with prior research using the policy-capturing method in the context of interorganizational alliances (Patzelt and Shepherd 2008; Shah and Swaminathan 2008; Tong et al. 2015), we included control variables pertaining to the industry, firm, and respondent. The selected control variables are likely to affect managers’ decisions and preferences with regard to forming, persisting with, and terminating alliances (Ibid.). As such, they could influence the focal firm’s response when faced with an alliance partner’s crisis.

First, we controlled for industry using industry growth (ordinal variable; 1 for declining; 2 for mature/growth rate close to zero; and 3 for growing). Its influence on managers’ judgment and preferences has been demonstrated in different strategic decision-making contexts, including joint venture formation and divestiture (Tong et al. 2015).

Second, we controlled for firm-specific variables. When forming, persisting with, and terminating alliances, top managers are often influenced by their firm’s ownership type, performance, and resource richness (Patzelt and Shepherd 2008). To account for these, we included four variables. We measured the focal firm’s ownership type using listed firm (binary variable; 1 for listed firm and 0 for privately owned firm), the focal firm’s performance using firm competitive position (ordinal variable assessing the firm’s standing in relation to direct competitors; 1 for below most competitors/below average; 2 for even with most competitor(s)/average; 3 for above most competitors/above average; and 4 for industry leader/number 1), and the focal firm’s resource richness using firm size (measured as the number of employees, logarithmically transformed).

Last, we controlled for respondent-specific variables. Strategy research using the policy-capturing method in the context of interorganizational alliances has shown that managers’ professional experience as well as their individual characteristics and values affect their judgement criteria and decisions when they face the critical choice between continuing or terminating an alliance. To account for these dimensions when analyzing the (adversary/advocacy) response, we included gender (binary variable; 1 for female and 0 for male), education level (measured in years of education completed after high school), CEO tenure (measured as years in CEO position in the present firm, logarithmically transformed), and alliance experience (measured as the number of alliances negotiated and formed over the CEO’s entire professional career, logarithmically transformed).

We also controlled for respondents’ pro-environmental values by adopting the four “respect for nature” (natural resources, welfare of the natural environment, current consumption, and future generations) items from Shepherd et al. (2013). These items are measured using alternative statements anchored at the two extremes of a seven-point scale. Contrary to Shepherd et al. (2013) who obtained high inter-item reliability (Cronbach’s alpha = 0.79), the Cronbach’s alpha of our four items was low (Cronbach’s alpha = 0.51) and did not allow for coding of an averaged, single respect for nature variable. Consequently, we included each of the four respect for nature items as distinct control variables in the subsequent statistical analysis.

In keeping with practice and recommendations in extant studies applying a policy-capturing method and using nested data (Patzelt and Shepherd 2008; Tong et al. 2015), we relied on hierarchical linear modeling (HLM) to conduct our statistical analysis. Our data are nested and involves two distinct (within- and between-respondent) levels: a first (within-respondent) level corresponds to the CEO response (or decision) which is nested within each manipulated scenario. At this first level, the HLM procedure accounts for the four decision attributes (see independent variables). A second (between-respondent) level corresponds to the 816 responses (or decisions) which are nested within the 51 CEOs of the sample. At this second level, the HLM procedure accounts for the CEO’s individual, firm and industry characteristics (see control variables).

As we collected data using a policy-capturing method, the 16 decisions reported for each respondent might be autocorrelated. HLM is able to control for such autocorrelation (Raudenbush and Bryk 2002). HLM is also recommended for examining within- and between-respondent variance, and for controlling for any heteroskedasticity that may be observed in nested data (Hofmann 1997). Moreover, accounting for the ordered nature of the two dependent variables and consistent with recommendations to analyze Likert scale categories as ordinal variables (Grilli and Rampichini 2012), we estimated an ordinal logistic HLM. We also considered ordinal logistic regression as more appropriate than the alternative estimation procedure, the OLS regression. In fact, we checked whether residuals were normally distributed, which is a key assumption when using OLS regression (and not when using ordinal logistic regression). The different OLS post-estimation normality tests (Jarque–Bera and Shapiro–Wilk tests) reported that this assumption was violated in our data.

Results

Hypothesis Testing

Table 1 presents descriptive statistics and correlations for all variables. This table shows that the two advocacy response and adversary response variables used to reflect a multidimensional measure of the response of the focal firm’s CEO are significantly and negatively correlated (= − 0.27 at p < 0.001). However, despite this significant correlation, the relatively low Cronbach’s alpha (= 0.42) and inter-item reliability confirm that the two variables reflect distinct response strategies, including the possibility of a firm initially remaining passive in a reputation spillover crisis.

Table 1 Descriptive statistics and correlation matrix
Full size table

Table 2 presents the results of the HLM analysis with robust standard errors. Model 1 examines the direct effects of the control and independent variables on the likelihood of adopting an advocacy response. Model 2 examines the direct effects on the likelihood of adopting an adversary response.

Table 2 Results of ordinal logistic HLM analysis
Full size table

Hypotheses 1, 2 and 3 relate three factors, indicative of alliance strength, to the likelihood of (de-)commitment escalation. Specifically, they predict that the focal firm will de-escalate its commitment to the accused partner if the alliance is of high strategic importance, its termination costs are high, or the focal firm’s CEO was personally responsible for the decision to initiate this alliance. Following the same rationale, these three hypotheses predict an advocacy response if the alliance’s strategic importance is low, its termination costs are low, or the focal firm’s CEO was not personally responsible for initiating this alliance. All of the HLM coefficients in Table 2 show a significant impact of strategic importance (at p < 0.001 in Models 1 and 2) and alliance termination costs (at p < 0.001 in Models 1 and 2) on the two response variables. As the significant impact for both variables is consistent with our predictions, we find support for the first two hypotheses (1 and 2). As regards the impact of personal responsibility on both dependent variables, Table 2 shows that this third alliance strength factor is significant only in Model 1 (at p < 0.01). Its impact is negative and consistent with the prediction of Hypothesis 3. Thus, Hypothesis 3 is partially supported.

The last hypothesis (Hypothesis 4) predicts a positive impact of one reputational variable—(high) reputation for management quality—on the likelihood of selecting an adversary response and an advocacy response for the low level. Table 2 shows that this independent variable is neither significant in Model 1 nor in Model 2. Thus, Hypothesis 4 is not supported.

As regards control variables, only Model 1 displays one significant and positive impact for alliance experience (at p < 0.05). In other words, the greater the number of alliances negotiated and formed over the CEO’s entire professional career, the higher the likelihood of selecting an advocacy response (and conversely).

Robustness Checks and Complementary Analyses

We conducted two additional statistical analyses in order to provide important clarifications for the data collected and to check the robustness of our results. First, we chose a two-item-scale approach to assessing advocacy and adversary preferences for our sample of Norwegian CEOs. In doing so, we intended to avoid forced responses, to offer our respondents the possibility to adopt a wait-and-see or no-choice position, and finally to get a comprehensive assessment of our respondents’ preferences. However, there is a non-negligible risk associated with this approach that our respondents may select the no-choice option more frequently than in real-life situations. This risk inherent to offering a no-choice option is also known as the “status quo bias” (Auspurg and Hinz 2015). Moreover, the issue with this no-choice option, which translates here into no clear preferences and similar or close responses (scores) on the two scale items, is that it reflects not only one but several (often complex and ambiguous) decision profiles within our sample.

To gain a better understanding of these profiles, as well as to examine to which extent our respondents were likely to take a strong and exclusive advocacy or adversary position, we followed a two-step complementary procedure: in a first step, we sought to identify all the possible no-choice decision profiles within our sample. To this aim, we computed the difference score between advocacy and adversary scale items (= advocacy response score – adversary response score) for each respondent and, on this basis, we created three categories of differential scores: the first [− 6 to− 3] category, with high and negative differential scores, corresponds to respondents with a stronger orientation towards adversary response, the second [− 2 to + 2], with low to zero differential scores, to respondents with no clear advocacy or adversary (or no-choice) response, and the last [+ 3 to + 6] category, with high and positive differential scores, to respondents with a stronger orientation towards advocacy response. The distribution of responses across the three categories shows that the advocacy- or adversary-oriented decision profiles account for 38.6% (= 315/816; 156 advocacy-oriented responses and 159 adversary-oriented responses) of total responses, while the no-choice decision profiles account for 61.4% (= 501/816).

In a second step, we checked whether the significant results reported individually for the two, advocacy response and adversary response, dependent variables in Table 2 remain the same when estimating an ordinal logistic HLM with a new aggregate and ordered dependent variable, decision profile (coded with the three categories of differential scores as follows: 1 for [− 6 to − 3] category; 2 for [− 2 to + 2] category, and 3 for [+ 3 to + 6] category) (see Table 3).

Table 3 Results of ordinal logistic HLM analysis with decision profile dependent variable
Full size table

Table 3 confirms the results obtained in Models 1 and 2 in Table 2. More specifically, we observe that strategic importance, alliance termination costs and personal responsibility are statistically significant (at p < 0.001, p < 0.001 and p < 0.05, respectively) and negative. As shown previously in Table 2, these results indicate that the likelihood of having stronger advocacy-oriented responses (corresponding to the [+ 3 to + 6] differential score category) is significantly higher when the alliance is of low strategic importance for the focal firm, its termination costs are low and the focal firm’s CEO was not personally responsible for the decision to initiate the alliance. Conversely, the likelihood of having stronger adversary-oriented responses (corresponding to the [− 6 to − 3] differential score category) is significantly higher when the alliance is of high strategic importance for the focal firm, its termination costs are high and the focal firm’s CEO was personally responsible in initiating the alliance with the accused partner.

Second, in our survey design, the presentation order of the 16 vignetttes was not randomized across respondents (see Appendix 1). A potentially adverse outcome of non-randomizing the vignettes is that later vignettes may be affected by survey fatigue, and thus biased. In order to control for this, we focused on the [− 2 to + 2] differential score category, indicative of respondents with no clear advocacy or adversary response but also of possible random or perfunctory answers, and checked how many differential scores match this category in the last four vignettes (excluding the duplicate vignette). On this basis, comparing these specific response patterns at the end of the survey with those observed (i.e., the number of [− 2 to + 2] differential scores) for the first four vignettes could allow us to investigate evidence of respondent fatigue in our survey. We compared these two sets of [− 2 to + 2] differential scores by computing a correlation test. For this test, we obtained a coefficient of 0.73 (significant at p < 0.001), indicating similar response patterns between the earlier and later vignettes and allowing us to minimize the influence of survey fatigue on our results.

Discussion and Conclusion

The main objective of this article was to examine influencing factors to firm reactions when an alliance partner is accused of misconduct and, consequently, there is a threat of negative reputation spillover. We considered the decision of an adversary or advocacy response as an escalation dilemma, and explored “escalation through the lens of interdependence” (Sleesman et al. 2018, p. 199). Specifically, we argued that factors traditionally found to increase escalation tendencies instead lead to de-escalation in our context, as they strengthen reputational interdependence.

For three of four factors, we found significant effects in accordance with the predictions in our hypotheses. According to our results, firms are more likely to select an adversary response when the alliance is of high strategic importance and has high termination costs. Conversely, firms are more likely to select an advocacy response when the alliance is of low strategic importance and has low termination costs, and when the CEO was not involved in the formation of the alliance.

The non-significant results on reputation for management quality may reflect a canceling out effect (Drummond 1995, 2014), equally observable in our theoretical development. Although we predicted a result in keeping with the extended commitment escalation argument, we noted that, at first sight, we might expect firms with good managerial reputations to choose an advocacy response. This because high-quality management reputation can also be damaged due to a failed strategy, such as an alliance strategy (Shafi et al. 2020). Moreover, the focal firm could presume its reputation acts as a defense against negative spillover (Zavyalova et al. 2016).

In sum, “managers may confront multiple and conflicting forces” (Drummond 2014, p. 431) when making decisions. When facing misconduct of an alliance partner with clear risk for substantial negative reputation spillover, CEOs prioritize reducing reputational dependence rather than reducing external resource dependence.

Contributions to Theory and Practice

Theoretical Implications

Our findings make several contributions to escalation theory. First, escalation research found that, mostly, “interdependence accentuates escalation” (Sleesman et al. 2018, p. 200). We extend this stream by revealing de-escalation effects resulting from reputational interdependencies. These insights shed new light on mixed evidence regarding the interdependence-escalation link (Sleesman et al. 2018) and answer the call for “a more nuanced understanding of escalation” (Drummond 2014, p. 431). Second, our findings add empirical support to the extended commitment escalation argument by accounting for equivocality about the (reputation) loss outcome. We see this as an important contribution, since counterforces to this cognitive bias and escalation theory’s boundary conditions have not been adequately researched (Drummond 2014). Third, we extend the application scope of commitment escalation beyond persisting with a collaborative project (i.e., an alliance) despite negative feedback to continuing a collaborative project with a partner engaged in a failing course of action (i.e., the focus of the commitment escalation lens is not the same).

Our study also contributes to reputation literature and provides new evidence that reputation concerns can instigate ethical decision-making (O’Fallon and Butterfield 2005); i.e., prompt the decision to sever ties with a partner accused of misconduct. It offers additional evidence of the value attributed to ethical reputation (Baselga-Pascual et al. 2018) and the effect it can have on alliance partnerships. More broadly, we add to research on response strategies to negative reputation spillover, which have rarely been studied at the interfirm alliance level. Simultaneously, we extend the literature on reputational damage and repair by showing that firms’ closest alliance partners are likely to harm the reparative processes that follow a preventable crisis.

Finally, to the alliance termination literature, we contribute a reputational reason for why firms may decide to end alliances. Negative reputation spillover between partners can constitute a partner-specific reason for terminating an alliance, just like a partner’s decision to exit an alliance following the implementation of a refocusing or debt reduction strategy (Meschi 2005).

Managerial Implications

The information provided in this article enables managers to anticipate likely response strategies of alliance partners following a preventable crisis; something truly valuable (Bruyaka et al. 2018). When an endorsement is withdrawn, “the public will perceive the events as very serious” (Rhee and Valdez 2009, p. 161), making it difficult to repair the reputational damage. An alliance partner’s response can also influence media framing of the issue (Rhee and Valdez 2009).

Moreover, partners’ likely reactions should be considered when selecting a crisis response strategy, as alliances with (non-)supportive partners may allow for different communication strategies. Accordingly, our findings provide some guidelines for issues management and the development of crisis communication plans.

Not least, awareness of the possibility that alliance partners may become crisis stricken over the course of the partnership should prompt the inclusion of extra exit clauses in alliance contracts. Besides, it is important for managers to understand when other partners will cut ties, as it may have consequences for whether stakeholders perceive a supportive behavior as legitimate.

Limitations and Future Research

Among our study’s limitations is the focus on some escalation factors only. We encourage scholars to investigate other CEO-, firm-, alliance-, and industry-level constructs. Also, we call for further attention to CEO perceptions around our fourth factor. For instance, while prior studies found that high reputation can protect against accusations of unethical behavior, it is unclear to what extent CEOs do hold that view. Moreover, while the empirical context of Norwegian manufacturing firms is well suited for the purpose of our study, whether our results can be generalized to other geographical and institutional contexts requires additional research.

Concerning the method employed, a bias could have occurred in our data because some of the CEOs lack experience with interfirm alliances. Yet, additional statistical tests showed that the occurrence of this bias is very limited in our data. Another bias, linked to our manipulated scenario-based survey, could result from the presentation order of the two (adversary and advocacy) scale items. In fact, the response score to the first scale item is likely to condition the response score to the second scale item as advocacy and adversary items are clearly perceived by respondents as opposing positions. In our survey design, the adversary scale item was placed first and this presentation order might bias results in favor of the hypothesized adversary responses and against the hypothesized advocacy responses.

As regards the scenarios, we had to make choices on: (i) level of similarity between the focal firm and its crisis-stricken partner, and (ii) type of crisis. Partner similarity was set relatively high (same industry and firm size) and the reputational contamination effect could be stronger in this alliance context, as it adds a ‘layer’ to the perceived partner proximity. Role of partner similarity should thus be further examined in future research. The latter scenario-related choice concerns crisis type. While we believe our results apply to misconduct other than environmental, this needs verification. Moreover, evidently, results are likely different for pure accidents or victim crises. As noted by Rhee and Valdez (2009, p. 164), “firms’ endorsers may be more likely to withdraw their endorsement in the case of scandals than accidents.” However, we contend that the escalation lens is less relevant in accident cases.

Furthermore, “a variety of theoretical lenses can be used to explore the [escalation] phenomenon” (Sleesman et al. 2012, p. 555). Accordingly, other frameworks, especially social/organizational identity theory (Tajfel and Turner 1986), might provide additional insights. We excluded identity theory in our theoretical reasoning for the sake of parsimony, but we note that predictions from this viewpoint are consistent with our hypotheses. The same comment applies to prospect theory (Tversky and Kahneman 1979) as well as the names-and-faces approach to stakeholder management (McVea and Freeman 2005; Perrault 2017), but from which we borrowed a few selected highly pertinent arguments to further clarify and strengthen our developments.

Finally, we did not account for the fact that firms may change their response strategies from, for instance, supportive to non-supportive. Meanwhile, over time, mimetic effects could lead to de-escalation among various alliance partners. We propose testing the impact of mimetic isomorphism on (de-)escalation behavior using a sociological institutional framework.

In conclusion, while further scholarly attention is called for, we believe our research makes important advances toward enhancing our understanding of interfirm dynamics in the context of a reputation spillover crisis.