Organizational strategies in the context of legitimacy loss: Radical versus gradual responses to disclosed corruption

Organizational corruption and changes in the institutional environment

Corruption can be generally defined as the “misuse of an organizational position or authority for personal gain or organizational (or sub-unit) gain” (Anand et al., 2004: 40). In line with existing studies on corruption from the institutional theory perspective (Gebhardt and Müller-Seitz, 2011; Misangyi et al., 2008), we refer to “misuse” not only as deviations from regulatory norms but also as deviations from accepted social norms and/or cognitive elements. We focus on systemic corruption at the level of the organization (see Ashforth et al., 2008: 673) and restrict our analysis to transnational cases of corruption illustrating the heterogeneous demands modern organizations face.

Regulatory sanctions for transnational corruption are a recent phenomenon. For a long time, existing legislation on, for example, cases of bribery extending beyond national borders was rarely enforced (Gibson Dunn, 2013; Weismann et al., 2014). The Foreign Corrupt Practices Act (FCPA; passed in 1977 and amended in 1988), a US law forbidding bribery of foreign public officials and requiring anti-corruption accounting rules for firms listed at the New York Stock Exchange (NYSE), has been in existence for more than three decades. However, the US Securities and Exchange Commission (SEC) and the Department of Justice (DOJ) have only recently started to enforce it extensively (Gibson Dunn, 2013). There are currently at least 78 corporations under investigation for possible FCPA violations and a total of US$3.74 billion has been paid by 58 companies to settle corruption charges (Wayne, 2012). In the meantime, many other countries have enacted similar laws (Nichols, 2012: 362).

Besides regulatory changes, the fight against corruption also involves normative and cultural-cognitive changes. Various international organizations such as the Organization for Economic Co-operation and Development (OECD) and the United Nations (UN) have established international conventions (e.g. the OECD Convention on Combating Bribery of Foreign Public Officials or the UN Convention against Corruption) that promote transnational anti-corruption norms without, however, being legally enforceable as such. Similarly, multi-stakeholder initiatives (e.g. the UNGC or the World Economic Forum’s Partnering against Corruption Initiative (PACI)) and nongovernmental organizations such as TI provide guidance on the fight against corruption (for overviews of changes in the institutional environment to fight corruption, see Gebhardt and Müller-Seitz, 2011; Hess, 2012). Despite these developments, however, there is no “shortage” of corruption-related transgressions involving MNCs.

Disclosed transgressions and organizational responses

A transgression generally refers to an unethical act such as corruption that places its stakeholders at risk (Coombs, 1995). We look at—disclosed—corrupt behavior as one particular type of an unethical act, or transgression (see also Pfarrer et al., 2008). Pfarrer et al. (2008) conceptualize a four-stage reintegration process and argue that passing through these stages successfully may increase the speed and likelihood of reintegration. A stage is passed when stakeholders concur about (1) what happened (discovery), (2) why it happened (explanation), (3) what punishment is appropriate (penance), and (4) what organizational changes have been implemented to prevent future misconduct (rehabilitation). Bertels et al. (2014) build on these four stages using a single-case study approach. As these two models apply a legitimacy approach in looking at organizational reintegration, they appear particularly relevant for our study. In contrast, other models of reintegration focus, for example, on trust (Gillespie et al., 2014; Gillespie and Dietz, 2009), restorative justice (Goodstein and Butterfield, 2010), or the micro level of analysis (Gromet and Okimoto, 2014). Further knowledge about organizational responses in the context of legitimacy loss seems necessary for the following reasons.

First, the model of Pfarrer et al. (2008) focuses on an ideal process of reintegration and is explicitly normative. It assumes that transgression responses can be separated into distinct stages that can be passed through in a certain linear sequence (Pfarrer et al., 2008: 740). In contrast, the model falls short when it comes to explaining deviations from the ideal, such as varying temporalities, sequences, and extents of transgressions and organizational responses that can be observed in in-depth case studies. In today’s globalized world with organizations facing varying jurisdictions and stakeholder demands and interests, one may wonder whether one reintegration stage is typically fully completed before progressing to the next. For example, for an MNC with offices around the world, disclosed bribery may induce several countries and jurisdictions to investigate the transgressor at different points in time. Hence, transgression response stages such as “discovery” and “penance” may occur in alternating waves or cycles rather than in a linear order. Furthermore, the transgressor may implement, for example, monitoring and due diligence processes that fall under the last “rehabilitation” stage, which may then trigger further “discovery.” Hence, it even seems unlikely that certain reintegration stages can be completed at all, for example, that all facts of a transgression can be “discovered” or the transgressor can be fully “rehabilitated” vis-a-vis all involved stakeholder groups. While Pfarrer et al. (2008) briefly refer to timing as a moderator in the transgression response process, we need to know more about how and why temporalities vary across cases.

Second, Pfarrer et al. (2008) look at reintegration as a process that is primarily, if not exclusively, stakeholder-driven. The authors expect firms to respond in full accordance with the stakeholder demands and assume that concurrence among stakeholders, for example, about the transgression and its key elements, is likely (Pfarrer et al., 2008: 736). In contrast, they neglect the possibility of more firm-driven actions in, for example, cases when firms act proactively or respond deliberatively to heterogeneous stakeholder demands. While Bertels et al. (2014) avoid the normative stance in their reintegration model and consider in greater depth demonstrated firm actions, they restrict their analysis to the role of only one particular type of stakeholder—governments (i.e. regulators, prosecutors, and courts)—in supporting organizational reintegration efforts. Consequently, they only look at regulatory (civil and criminal) sanctions leaving mostly unaddressed the role of social sanctions as defined below.

Third, both reintegration models presented above focus on rational reintegration processes based on economic calculations (e.g. Hart, 1995). Business firms may calculate ex ante the control costs of compliance (including opportunity costs for loss of orders or delayed businesses due to anti-corruption policies), on the one hand, and costs of noncompliance (including anticipated sanctions for disclosed transgressions and legitimacy loss), on the other. After a transgression, Western enforcement agencies such as the SEC and DOJ (in the United States) play a key role in enforcing anti-corruption regulations, and impose severe monetary sanctions and compliance costs on transgressors. For example, enforcing the FCPA often entails out-of-court settlements with US authorities (Schwarz, 2011). The Deferred Prosecution Agreements (DPAs), a proceeding that is closed only after the successful completion of the agreed terms, is a popular arrangement in such a settlement (see, for example, Bertels et al., 2014; Gebhardt and Müller-Seitz, 2011; Schwarz, 2011). Based on the same economic calculation, however, rational transgressors may still be tempted to avoid implementation costs by introducing only gradual changes. Beyond such rational responses, Pfarrer et al. (2008) only occasionally include references to emotions such as shame. If they do so, they focus on “shaming” from the stakeholder perspective, rather than discussing the organizational response of “feeling shame” (p. 739). Similarly, Bertels et al. (2014) mention only once that transgressors “would be very embarrassed” (p. 355), without outlining the organizational response to such emotions.

In line with institutional theory (see Greenwood et al., 2011; Pache and Santos, 2010; Scherer et al., 2013), not only expected sanctions and compliance costs, but also the consistency of societal (anti-)corruption expectations is likely to influence organizational response. For example, Western anti-corruption and compliance demands may become more urgent for a transgressor than bribery demands by a public official in a developing country. Lange (2008) distinguishes two major types of sanctions in the context of a transgression, which we draw on in this article: regulatory sanctions and social sanctions. Regulatory sanctions include, for example, the disgorgement and penalty payments listed in the settlement agreements plus other legal expenses, for example, for lawsuits, lawyers, or consultants. In contrast, social sanctions primarily refer to the extent of negative press coverage that transgressors receive in the course of disclosure, settlement, and reintegration.

Legitimacy strategies in anti-corruption environments

From an institutional point of view, MNCs seek to portray themselves as legitimate actors by incorporating institutional elements of internal and external origin (Kostova et al., 2008). When an organization’s actual operations and institutional expectations are mismatched, the organization needs to adapt to the perceptions of its constituents. Incorporating global institutional demands relating to anti-corruption, however, poses serious challenges to MNCs: while the enforcement of legislation against transnational corruption originates mainly in developed Western countries, many developing countries in which MNCs operate still present a high corruption risk (TI, 2012). Consequently, MNCs can easily become involved in corrupt practices along their supply chains, thus diminishing a corporation’s legitimacy.

Scholars have distinguished three broadly defined types of legitimacy (see Suchman, 1995): (1) Pragmatic legitimacy “rests on the self-interested calculations of an organization’s most immediate audiences” (Suchman, 1995: 578). It can be maintained or regained when the most important constituents expect to benefit from the organization. (2) Moral legitimacy is based on normative evaluations of the organization and is granted if the organization’s behavior is considered the “right thing to do” (see also Aldrich and Fiol, 1994). (3) Cognitive legitimacy, in turn, is distinct from evaluation in that it refers to the “mere acceptance of the organization as necessary or inevitable based on some taken-for-granted cultural account” (Suchman, 1995: 582). How organizational responses to legitimacy loss shift over time and why such responses may differ across organizations, depending on which types of legitimacy are threatened, remain largely unaddressed (see Bertels et al., 2014; Pfarrer et al., 2008).

Rather than referring to reintegration stages (Pfarrer et al., 2008), we decided to focus on legitimacy strategies, considering that the latter come with the advantage of not prescribing a distinct and linear sequence of application. Scholars have identified several types of legitimacy strategies; our initial analysis was guided by the following typology, developed for sustainability issues in complex institutional environments (Scherer et al., 2013): isomorphic adaptation, moral reasoning, and strategic manipulation (see also Driscoll, 2006; Oliver, 1991; Scott, 2008; Suchman, 1995). By using an isomorphic adaptation strategy, the organization adapts its organizational practices to social expectations—for example, an MNC may choose to report more extensively on corruption in its annual reports if it perceives that this is an emerging practice in its industry or the country in which it operates. By using a moral reasoning strategy (Palazzo and Scherer, 2006; Scherer et al., 2013), organizations may enter into a discourse with focal stakeholders and societal groups about the acceptability of the organization’s status quo and behavior. For example, discussing ideas on how to fight corruption with peers in multi-stakeholder initiatives like the UNGC can be part of a moral reasoning strategy.

Third, companies may use a strategic manipulation strategy, which involves influencing how its constituents perceive the way in which the organization appears to benefit them (Oliver, 1991; Scherer et al., 2013). The key instruments associated with manipulation strategies are strategic public relations and impression management tactics that involve advertising campaigns, dissemination of (misleading) information, or lobbying (Fombrun, 2001; Oliver, 1991; Scherer et al., 2013). Arguing that organizations that use a manipulation strategy do not change the practices criticized by some of their stakeholders, scholars highlight the parallels of this strategy with the concept of decoupling (Scherer et al., 2013: 266; see also Palazzo and Richter, 2005; Weaver et al., 1999). “Decoupling” refers to a gap between formal structures and actual practices (Meyer and Rowan, 1977). The application of decoupling as a strategy to respond to conflicting institutional demands assumes that organizational structures and activities can be protected from evaluation (Meyer and Rowan, 1977: 357). The recent wave of disclosed corruption at MNCs involving large-scale investigations (Gebhardt and Müller-Seitz, 2011) suggests that protection from evaluation is not always possible.

Relating legitimacy strategies to corruption control types

To be able to determine which legitimacy strategy a firm applies, we need to identify and categorize organizational practices in the context of corruption. The following literature proved helpful in this regard. Lange (2008) recently merged two streams of literature—one on organizational and the other on corruption control—to conceptualize an “Organizational Corruption Control Circumplex” (OCC). Therein, he distinguishes social/cultural controls from administrative controls (see left column of Table 1). “Controls” in this context refer to organizational mechanisms to fight corruption. While the former type of controls is transmitted by values and norms in organizations, the latter type focuses on formal structures and processes. Regarding legitimacy strategies, we may roughly associate a focus on administrative controls with an “isomorphic adaptation” strategy, and a focus on social/cultural as well as interactive controls with a “moral reasoning” strategy. If, however, a company implements only a fractional set of administrative controls without any elements of social/cultural controls, a “decoupling” strategy seems likely to be at play (see Scherer et al., 2013).

OPEN IN VIEWER

Table 1.

Relating corruption control elements and legitimacy strategies.

Type	Corruption Control Elements	Legitimacy strategies
Social/cultural corruption controls	Leadership support • SC: high-level personnel as overseer (see also: Pfarrer et al., 2008) • CC: strategic integration and leadership support	Moral reasoning		Substantial influence
	Training • SC: effective communication• CC: training• See also OCC: self-controls
	Whistle-blowing procedure • OCC: vigilance controls; concertive controls (horizontal surveillance by coworkers)• CC: creation of a complaints procedure
Administrative corruption controls	Bureaucratic controls • OCC: formalized rules, policies (see also: Pfarrer et al., 2008), hierarchical and centralized structure, specialized positions• SC: compliance standards and procedures; due care in delegating substantial discretionary authority	Isomorphic adaptation	Decoupling(if implemented loosely and fractionally)
	Alignment of incentives • OCC: incentive alignments• CC: alignment of incentive structures (see also: Pfarrer et al., 2008)
	Monitoring and follow-up • SC: monitoring, auditing, reporting; enforcement and disciplinary mechanisms; response upon detection• OCC: punishment (measuring, monitoring and sanctions)• CC: evaluation
Interactive corruption controls	Collective Action• CC: quality of stakeholder relationships; level of participation in collaborative CC initiatives• Beyond OCC and SC

• CC = Corporate Citizenship Assessment Tool (Baumann-Pauly and Scherer, 2013).

• OCC = Organizational Corruption Control Circumplex (Lange, 2008).

• SC = Sentencing Commission Guidelines (USSC, 2010).

In addition, given that our article deals with corruption cases that included settlements with the SEC, we also considered the guidelines for “Effective Compliance and Ethics Program[s]” of the US Sentencing Commission (USSC, 2010: §8B2.1) and categorized them into social/cultural versus administrative controls (Lange, 2008). Finally, as we shall explain later, our empirical analysis revealed the need to further consider an interactive and/or voluntary type of organizational measure, for example, as depicted in the corporate citizenship assessment model (CC) by Baumann-Pauly and Scherer (2013). This type captures corruption controls that go beyond the elements of the previous two frameworks and focus on collaborative efforts of organizations with the intent to change anti-corruption practices not only within the organization but also in its institutional environment. As such, this type proved necessary to fully capture the organizational practices in the context of a “substantial influence” strategy that emerged from our analysis. Table 1 displays the synthesis of corruption controls resulting from the three presented frameworks and relates those controls to legitimacy strategies (left part of Table 1). Depicting the three types of organizational corruption controls in the left, column, we list key organizational corruption control elements of each type in the middle column; the original literature and practitioner sources (OCC, USSCG, CC) are listed as bullets beneath each element. Note that Pfarrer et al. (2008) barely discuss actual organizational corruption controls, except for single examples in the rehabilitation stage, which we also referenced in Table 1.

To sum up, institutional theory scholars have acknowledged that legitimacy concerns arise particularly when routines fail and a transgression or crisis occurs (Ashforth and Gibbs, 1990; Meyer and Rowan, 1977; Scherer et al., 2013: 262). However, how such events trigger particular legitimacy strategies in the context of a transgression over time remains largely unaddressed (Pfarrer et al., 2008). Our aim is, hence, to close this gap by exploring how distinct elements of transgressions influence the dynamics, sequence, and combination of organizational strategies in the context of legitimacy loss and which factors play a crucial role along the reintegration process. Having shed light on these questions, we will theorize a generalized process model of organizational strategies in the context of legitimacy loss.

Research design and sampling

We applied a multiple dynamic case design to compare and interpret our findings, which cover the period 2012–2014, across our target firms (Yin, 1984), as well as over time and space (Gerring, 2007), and aggregated data where appropriate. In selecting our sample, we followed the sampling approach (see Strauss and Corbin, 1998) of choosing “extreme” cases wherein the phenomenon of interest (here, the disclosed corruption transgression) is “transparently observable.” On the basis of our sampling approach, we selected the following companies: Siemens, Daimler, and ABB. In all cases, we were able to interview firm representatives toward the end of the regulatory proceedings (the period in which the transgressors operate under the DPA or a similar settlement with public authorities, as defined above) and after the termination of these regulatory obligations.

We selected cases of MNCs implicated in recent transnational transgressions, which differed in key aspects, most notably concerning subsequent regulatory and social sanctions. Our assumption was that the analysis of different transgressions of multiple case firms would enable us to develop theory about the reintegration process with regard to the sequence, combination, and extent of applied legitimacy strategies of large, prominent, and powerful organizations facing heterogeneous expectations in their environment (see Doz, 1996; Pfarrer et al., 2008). At the same time, we aimed to control for factors that are unrelated to the transgression characteristics: we focused on firms from the high technology manufacturing sector, generally known to face moderate to high corruption risks (TI, 2011). Furthermore, we chose only MNCs headquartered in Switzerland or Germany. Both countries are perceived to have similarly low corruption risks (TI, 2012), have similarly strict anti-corruption legislation, and are at an advanced stage of implementing anti-corruption measures that derive from the OECD Convention against Corruption (see OECD, 2012). To gain a more global perspective on the corruption phenomenon, we expanded our data collection to several other countries. We focused on Asia Pacific, as this region includes several high-corruption-risk countries that have been involved in the corruption schemes of our transgressors (see TI, 2012). Furthermore, this region allows for smooth data access thanks to infrastructure and use of English language.

Case context

Siemens is among the leading technology companies worldwide, headquartered in Munich and Berlin, Germany, and operating with four sectors (energy, industry, infrastructure and cities, and healthcare—until a slight restructuring in September 30, 2014) in more than 190 countries (Siemens, 2013). Siemens has a total of around 370,000 employees and its annual turnover generally exceeds €80 billion (Siemens, 2015). According to the SEC’s complaint, “between March 12, 2001, and September 30, 2007, Siemens created elaborate payment schemes to conceal the nature of its corrupt payments, and the company’s inadequate internal controls allowed the conduct to flourish” (SEC, 2008b). Infractions covered various sectors and projects including, for example, metro transit lines in Venezuela, power plants in Israel, refineries in Mexico, mobile telephone networks in Bangladesh, and medical devices in Vietnam, China, and Russia (for a detailed list, see SEC, 2008a: 13–32). Siemens paid “more than $1.4 billion in bribes to government officials in Asia, Africa, Europe, the Middle East, and the Americas [and] the pattern of bribery […] was unprecedented in scale and geographic reach” (SEC, 2008b). The SEC complaint further specifies that the MNC used third parties to obscure the purpose for, and the ultimate recipient of, thousands of cash payments, which employees sometimes transported in suitcases across international borders. Like the transgression itself, also the institutional sanctions for Siemens were unprecedented. Table 2 provides a comparative overview of regulatory and social sanctions at our case firms.

OPEN IN VIEWER

Table 2.

Classification of transgression sanctions.

Company	Type of sanction	Magnitude of sanction
Siemens	Regulatory sanctions (fines and organizational compliance costs)	In December 2008, Siemens entered an agreement to settle the Foreign Corrupt Practices Act (FCPA)-related charges it had been called to pay, including the installation of an external monitor for a period of 4 years until December 2012 (SEC, 2008c). In a settlement with the World Bank in July 2009, Siemens committed to pay US$100 million over the following 15 years to support anti-corruption work (World Bank Group (WBG), 2009). The fines imposed by the US and German regulatory authorities amount to more than US$1.6 billion; the organizational costs for rehabilitation are expected to surpass this figure. Altogether, we can ascertain with confidence that fines and costs involved in this Siemens corruption transgression are unprecedented.
	Social sanctions (reputation loss and prominence of transgression)	As we learnt in the course of our interviews, the disclosure of the transgression was rather unique and had an immense effect on social sanctions: in November 2006, around 200 policemen and officials turned up at the Munich headquarters to confiscate material relevant to the corruption cases at Siemens and arrested Siemens employees of the communication sector (SEC, 2008a). This “Dawn Raid” is likely to have made the magnitude of the problem obvious to almost every Siemens employee around the world and triggered an unprecedented level of negative media coverage on corporate corruption (see RepRisk, 2013). Even more, closely after this event, Transparency International, Germany, the prominent normative anti-corruption institution, terminated Siemens’ membership in December 2006. Overall, we perceive social sanctions to be by far the highest among our case firms.
Daimler	Regulatory sanctions	The fines Daimler paid to settle the Securities and Exchange Commission (SEC) charges with the SEC and the Department of Justice (DOJ) amounted to around US$185 million in total. Furthermore, an external monitor was installed pointing out major organizational compliance deficits (Spiegel, 2011); hence, the organizational costs to achieve rehabilitation are likely to be substantial. Although fines and related costs at Daimler can be described as high, they do not reach the level of burden at Siemens.
	Social sanctions	The corrupt practices were disclosed by the SEC in March 2010 and the agreed Deferred Prosecution Agreement (DPA) lasted from March 2010 to March 2013, so the Daimler transgression erupted at least 1½ years after the Siemens transgression. This may also explain why Daimler was not as much in the spotlight as Siemens, reflected in the significantly lower level of media coverage on corruption risks at Daimler (RepRisk, 2013). In the Daimler case, there was also no big and sudden investigative act to trigger instant and massive public awareness. The level of social sanctions is thus rather moderate compared to Siemens.
Asea Brown Boveri Ltd (ABB)	Regulatory sanctions	To settle the SEC charges, ABB agreed to pay more than US$39.3 million. In this case, no external monitor was appointed, which together with the requirements listed in the DPA (DOJ, 2010) indicates that the organizational costs for rehabilitation are considerably lower than those of Siemens and Daimler.
	Social sanctions	The 3-year DPA agreed with the SEC started in September 2010, almost 2 years after the Siemens agreement; moreover, the scope of the corrupt practices was limited to a few countries. This may indicate the rather low media coverage on compliance risks, especially compared to the Siemens case (see also Table 3) supports our assessment: from 2006 to 2013, we found only 10 entries on corruption-related reputation risk concerning ABB in the RepRisk database, compared to 138 for Siemens and 40 for Daimler (RepRisk, 2013). In sum, social sanctions are lowest among case firms.

Daimler is a multinational automotive corporation with headquarters in Stuttgart, Germany, and locations on every continent; it focuses on Europe, Asia, and North America. The company has five divisions, of which the two most important are Mercedes Benz Cars and Daimler Trucks (Daimler, 2013). Daimler’s annual turnover of over €100 billion (Daimler, 2013) exceeds that of Siemens, although the company has fewer employees (around 275,000) and operates in fewer countries. According to the SEC,

Daimler paid at least $56 million in improper payments over a period of more than 10 years. The payments involved more than 200 transactions in at least 22 countries. Daimler earned $1.9 billion in revenue and at least $90 million in illegal profits through these tainted sales transactions, which involved at least 6,300 commercial vehicles and 500 passenger cars. (SEC, 2010b)

Table 2 outlines the high regulatory and rather moderate social sanctions Daimler faced.

ABB is one of the leading firms worldwide in energy and automation technologies. The company has its headquarters in Zurich, Switzerland, and operates in more than 100 countries (ABB, 2013). ABB has around 146,000 employees and an annual turnover of around US$40 billion (ABB, 2013). On September 29, 2010, the SEC charged

ABB Ltd with violations of the Foreign Corrupt Practices Act (FCPA) for using subsidiaries to pay bribes to Mexican officials to obtain business with government-owned power companies, and to pay kickbacks to Iraq to obtain contracts under the U.N. Oil for Food Program. (SEC, 2010a)

According to the SEC, “ABB’s subsidiaries made at least $2.7 million in illicit payments […] to obtain contracts that generated more than $100 million in revenues for ABB” (SEC, 2010a). ABB faced lower regulatory and social sanctions compared to Siemens and Daimler as noted in Table 2.

Data collection and analysis

Following established guidelines (Glaser and Strauss, 1967), we analyzed our data during the collection process. To reduce biases wherever possible, we use interview as well as documentary data, both of which we triangulated by incorporating various perspectives (Eisenhardt, 1989; Flick, 1992) as summarized in Table 3. We conducted 40 interviews with representatives of the case firms and third parties such as lawyers, judges, auditors, nongovernmental organizations (NGOs), and officials in the MNC headquarters and Asia Pacific region. Interviews typically lasted between 45 and 90 minutes and were transcribed. We selected respondents initially based on their expertise and/or direct involvement in the reintegration process, and later added informants recommended during previous interviews. In addition, we analyzed 54 company (annual and/or sustainability) reports, policies, and codes of conduct; 16 documents relating to regulatory and criminal proceedings provided by the SEC, the DOJ, and the World Bank; and 207 press articles on our case firms provided by RepRisk AG (RepRisk, 2013) and other media. Table 3 lists the number of documents per type of source for each company. Given that our interview data spans 3 years (2012–2014), including repeated interviews with several actors, and our documentary data covers the period from 2003 to 2014, we have been able to analyze the organizational situation before, during, and after the respective corruption transgressions.

OPEN IN VIEWER

Table 3.

Data sources and corresponding perspectives in the analysis.

Data types	Siemens	Daimler	Asea Brown Boveri Ltd (ABB)	Third parties (lawyers, auditors, NGOs, officials)	Perspective (purpose: triangulation)
Interviews (Germany and Switzerland)	3	2	2 (2 informants each)	4 (+1 formal and several informal interviews with academics)	Corporate headquarters
Interviews (Australia)	2	1	1	3 (+3 formal and several informal interviews with academics)	Corporate global (low perceived corruption risk)
Interviews (Southeast Asia)	4	1 (2 informants)	2 (now at Siemens)	11	Corporate global (high perceived corruption risk)
Foreign Corrupt Practices Act (FCPA)-related documents	6	3	6	1	Regulatory
Company reports and policies	20	14	20	–	Corporate perspective (details on processes, structures, and performance)
RepRisk (4 September 2013) and other media	152	45	10	–	Public/societal perspective

At the beginning of our analysis, in 2012, we considered a variety of documentary sources (Yin, 1984), ranging from regulatory proceedings related to the FCPA to company reports and media reports, to form a picture of the corruption. On this basis, we developed the guideline for the first interviews at the companies’ headquarters, focusing on what characterized each transgression and on the initial reactions of the transgressors. As regulatory proceedings came to an end in the course of 2013, we gained further insights into the organizational processes of responding to legitimacy loss and observed corporate actions in “real time.” Such “real time” changes (e.g. in key personnel or the organizational structures) were addressed in subsequent interviews. We also drew on these first interviews and documentary data to conduct the first round of open coding (Corbin and Strauss, 1990; Glaser and Strauss, 1967).

In the next step, we turned to academic literature to relate our data to theoretical concepts and frameworks. To develop our data structure, we adhered to recommendations for building grounded theory on the basis of a two-order code scheme (Corley and Gioia, 2004; Gioia et al., 2013; Ravasi and Phillips, 2011). According to this approach, first-order codes reflect as closely as possible the language used by the informant. In contrast, second-order codes are mainly based on the analysis and synthesis of first-order codes. For example, the repeated first-order references to a “no choice” and “strong emotional shift” situation (at Siemens) or (FCPA) monitor “pressure” and “intensity” (at Daimler) helped us identify dominant second-order themes like “legitimacy shock” or “external enforcement pressure” that we incorporated in our emerging data structure. Third, we discussed possible relationships among the themes that had emerged. This process relates to axial coding techniques (Strauss and Corbin, 1998). Figure 1 illustrates the relation between first- and second-order codes as well as the overarching themes that we subsequently synthesized.

OPEN IN VIEWER

Figure 1.

Data structure.

These data structuring procedures together with continuous discussions between the authors about reconfigurations of codings are expected to “lend the requisite rigor to the analyses” (Gioia et al., 2013: 22). Additionally, we took into account independent codings of selected data by student assistants, and discussed the emerging themes together with our interpretations in follow-up interviews with respondents from the case firms and incorporated their feedback (see Corbin and Strauss, 1990: 11). Our motivation for this step was to test our confidence in our findings and interpretations (see Gioia et al., 2013: 22).

Strategic shift toward substantial influence in the context of legitimacy loss

Before highlighting the individual case narratives, we now focus on the commonly observed strategic shift toward “substantial influence” after disclosed corruption. This post-transgression strategy emerged inductively from our data and is composed of the two codes “we do more than we have to” and “we want to make sure everybody else does so too.”

Strategy shifts and differences in the chronology, speed, and intensity of strategies

Daimler’s gradual response and external enforcement boost

Daimler faced a corruption transgression of similar (geographic) scope to the Siemens affair, involving high regulatory sanctions (see Table 2), yet pursued a significantly different reintegration process. Nonetheless, the MNC finally also shifted toward “substantial influence.” The analysis of sanctions at Daimler suggests that pragmatic legitimacy was at risk whereas the other two types of legitimacy were largely unthreatened. We outline in the following that, because the way of disclosing the transgression was far less geared toward the media and social sanctions were mild, Daimler had—and may also have needed—time to make its global workforce aware of the transgression.

At Daimler, corruption controls in the period (t_pre), that is, before the transgression was officially disclosed in a settlement with the SEC on 1 April 2010, were mainly bureaucratic (e.g. formalized rules and policies) and to a limited extent social/cultural or interactive (e.g. participating in voluntary initiatives like the UNGC). Daimler introduced further—at least basic—measures, as the process chart (Figure 3(b)) shows. Social/cultural corruption controls were fostered, for example, through a statement by Dieter Zetsche outlining that “no business deal can ever justify putting our company’s reputation at risk“ (translated into English, speech delivered at the general assembly, April 2006), as well as relevant training and a whistle-blowing procedure. Furthermore, Daimler increased its administrative corruption controls in 2005 by monitoring and following up cases of corruption (see GRI indicators SO2, SO3, and SO4 in Figure 3(b)). Considering that several of Daimler’s compliance controls had been previously applied elsewhere (for example, at ABB), we identify elements of an “isomorphic adaptation” strategy. Furthermore, Daimler applied already in t_pre elements of “moral reasoning,” for example, by participating in voluntary initiatives. However, the scope of transgressions disclosed at Daimler in t_pre also reveals that systemic corrupt practices continued well into this period. Hence, “decoupling” was a key strategy at that time. (“Until the year 2008 we did not do that in a quality that we considered and consider as necessary” (Daimler, Senior CM, November 2012).)

Although Daimler faced fewer regulatory and social sanctions during t_trans than Siemens (see Table 2), we noted in t_post that its compliance program in many aspects comes close to that of Siemens. It contained strong social/cultural controls, including a close integration of compliance in the board-level brief “Integrity and Legal Affairs” established in 2011, comprehensive anti-corruption training, and a whistle-blowing procedure. Interestingly, yet to a lesser degree compared to Siemens, we now also found narratives at Daimler with a certain personal and emotional connotation: “the firm has pushed to stigmatize bribery, even in countries where it is a common business practice. ‘It’s no longer just a gentleman’s crime,’ she [Hohmann-Dennhardt] said” (Ensign, 2013). Similarly, clear anti-corruption signals from and to the leadership now referred to notions such as “self-servicing mentality.” On the other hand, administrative controls include risk assessments, monitoring, and sanctions such as dismissals. Third-party bank accounts were significantly reduced, although value flow centralization was lower than at Siemens.

Observing the overlaps between the Siemens and Daimler compliance programs despite the significant differences in (social) sanctions led us to investigate the factors that may have influenced the transgression response process. After consulting further literature on contingencies, press coverage, and personal interviews, we learned about the powerful role of an externally imposed FCPA monitor exceeding by far the role of regular auditors. For example, the monitor can request access to all sorts of compliance-related documents, interview people in various hierarchical positions, and visit foreign subsidiaries (Hartmann, 2012; Schwarz, 2011).

Our analysis revealed that long during t_trans, corruption controls were implemented rather slowly and insufficiently. As the Spiegel (2011) reports about information regarding the strictly secret “First Follow-up Monitor Report” by Louis Freeh (FCPA monitor at Daimler):

The Daimler-internal investigations of suspicious cases are too slow (…) Daimler does not scrutinize the efficiency of the initiated processes (…) The enterprise has not prevented “unethical behavior of managers” consistently (…) At the moment, [Daimler] would not fulfill the criteria of the US stock exchange supervision SEC.

Such references to slow yet insufficient change at Daimler point toward a combination of “decoupling” and “isomorphic adaptation.” We further noted that “Daimler has not changed its leadership team. Daimler played it differently [than Siemens]” (Lawyer, Gibson Dunn, November 2012). The reasoning on why to abstain from a massive leadership exchange involves considerations of fairness and open-ended communication with the Daimler workforce reflecting elements of “moral reasoning”:

[M]any leaders grew up and studied in a world in which the perception existed that in some countries making business without facilitation payments, to name it that way, is not possible (…) I think it is fair to first inform and enable them to cope with this new context. (Senior CO, Daimler HQ, November 2012)

That is, the transgression was partly attributed to the educational background of managers—foreign corruption had long been tax-deductible in Germany (Glynn et al., 1997)—and the perception of corruption as a common practice in several countries worldwide.

ABB’s ¹ cyclical and partly inconsistent responses to repeated minor transgressions

Our analysis of ABB’s corruption controls took a different course compared to the two other cases: our ABB interview partners focused mainly on ABB’s present state of corruption controls, that is, the period t_post marked by the end of the DPA in September 2013. In this post-transgression period, we identified several strong and comprehensive social/cultural controls, notably training based on a case study of real (anonymized) past incidents of corruption, a sound and offensively communicated whistle-blowing procedure (including an information desk in the HQ lobby) and strong leadership support (including a separate board position). Significantly, we also noted personal emotional narratives at ABB in this period. On the one hand, our content analysis of ABB’s personal and written communication revealed that ABB refers mostly to integrity, rather than compliance. On the other hand, our conversations with two senior ABB representatives in 2012 and 2013 diverged several times from a purely rationalized question-and-answer process toward more personal and emotional reflections of the respondents, for example, as regards the (ethical) role of corporations, the distinction between good and bad and corrupt and noncorrupt behavior. During these interviews, our respondents used notions such as “being tired of rules,” “the individual personal aspect” of integrity, and consequences in case “you screw up.” However, apart from these strong social/cultural corruption control narratives and certain interactive controls—for example, membership in TI, the UNGC, and the PACI—we were not able to obtain detailed information on administrative controls. The partially vague and ambiguous responses from our ABB interviewees, together with the assessments from third parties, suggest that ABB’s administrative controls are less comprehensive compared to its social/cultural controls. As the CO of an industry peer puts during our interview in December 2013:

In the control system [referring to administrative controls], I believe, ABB has weaknesses today (…) We compare notes on this issue (…) and ABB also openly communicates that they did not invest as much in the implementation of a control system …

A gap between the ABB headquarters perception and our own, as well as third-party observations about the company’s handling of the transgression, is evident in the processes of risk assessment, monitoring, following up of incidents, and sanctions. For example, ABB mainly relies on annual surveys and the audit department to conduct corruption-risk assessments (Siemens has special “compliance” assessment teams in addition to the audit department). Furthermore, ABB’s reporting on these issues is less comprehensive compared to the other case firms in t_post. In sum, ABB displays, on the one hand, elements of “substantial influence” and “moral reasoning.” On the other hand, the gap between the image of ABB as a leader in anti-corruption management, which our interviewees painted, and the gaps between ABB’s administrative controls and the industry’s best practices that we perceive, and an industry peer and other external experts corroborated, may suggest that certain forms of decoupling were present after the second settlement.

In t_trans, the period marked by the SEC settlement including a 3-year DPA starting in September 2010, ABB faced lower regulatory sanctions and almost negligible social sanctions (see Table 2) compared to Siemens. This suggests that none of the three legitimacy types was severely threatened. Notably, in contrast to Daimler and Siemens, ABB was not scrutinized by an external monitor, while lawyers, consultants, and auditors conducted more limited investigations than at Daimler and especially at Siemens. Overall, it cannot be said that external authorities left “no stone unturned” in ABB’s case. Consequently, instead of establishing a system that would help it achieve full transparency about every business process, ABB chose the possibly less resource-intensive approach of preventing corrupt behavior by attempting to change the mindset of its employees.

While respondents from all three firms mentioned the powerful role of the SEC, ABB interviewees downplayed the effect of the 2008–2010 transgression disclosure and settlement on the company’s anti-corruption strategy. They admitted that there was a whole list of organizational requirements that the company had to fulfill in order to comply with the FCPA. Furthermore, considering that their rhetoric revealed a confidence that the company would not have to introduce drastic changes indicates the application of an “isomorphic adaptation” strategy. From t_trans onward, ABB pursued this strategy by orienting itself more to the practices already implemented at competitors like Siemens: A comparison of the slogans of each company’s compliance program (Siemens: “Prevent, Detect, Respond”; ABB: “Prevent, Detect, Resolve”) supports this impression. While our respondents could not specify when and why these slogans were conceived, our document analysis revealed that Siemens has used its slogan in public reports since 2009 (e.g. Annual Report, 2009: 59), whereas the first ABB document we found using this slogan was published on 1 April 2012 (“The ABB integrity program,” 2012: 3). Similarly, Siemens’ slogan “Only clean business is Siemens business—everywhere—everybody—every time” (see Moosmayer and Winter, 2011: 4, indicating that the slogan originated in a Löscher speech between 2007 and 2008) is later echoed in ABB’s code of conduct: “We expect this [acting ethically and with integrity] of every single ABB employee, in every location, every day” (ABB Code of Conduct, 2012: 4). References in the DPA such as “following discovery of the bribery, ABB (…) voluntarily and timely disclosed to the (…) SEC the misconduct” and “ABB Ltd conducted a thorough internal investigation of that and other misconduct” (ABB Deferred Prosecution Agreement, 2010: 4) suggest a termination of “decoupling.” However, the identified gap between ABB and third-party perceptions leaves us uncertain as to whether some elements of “decoupling” may have persisted at ABB in this period (see Figure 3(c)).

A central reason behind ABB’s introduction of considerable corruption controls in the period (t_pre) prior to the SEC settlement and DPA became apparent only after the second round of interviews and analysis: an earlier settlement between ABB and the SEC in 2004. The smaller scale settlement (bribes of US$1.1 million and a penalty of US$10.5 million, compared to the 2010 settlement involving at least US$2.7 million in illicit payments and a penalty of US$39.3 million) and the early date may explain why we did not initially notice it. Nevertheless, our analysis (see Figure 3(c)) shows that (in t_pre) ABB was indeed the first of our three case firms to install certain corruption controls, especially social/cultural ones as outlined above, which points to strategies of “moral reasoning” and “(substantial) influence.” Considering that the SEC found ABB’s integrity program deficient on several counts in the 2010 settlement (“ABB, through its officers, agents and subsidiaries, failed to detect and prevent the illicit payments revealing a lack of effective internal controls” (SEC, 2010a: 19)), “decoupling” is likely to have persisted at ABB after the first settlement (2004).

“Substantial influence” resulting from high legitimacy loss and emotional consternation

In contrast to the suggestions of previous research that companies facing an unprecedented transgression—like Enron, WorldCom, or Tyco—may not recover due to the enormous stigma that “first movers” suffer (Goffman, 2009; Pfarrer et al., 2008), our study reveals that “first movers” may indeed be able to regain legitimacy by substantially influencing their environment. Particularly, we argue that high levels of legitimacy loss, and especially a legitimacy shock (center of Figure 2), that is, a radical loss of legitimacy due to exceptionally high regulatory and social sanctions, increases the chances that transgressors will make wholesale and radical changes. Our findings further suggest that conventional reintegration models (Bertels et al., 2014; Pfarrer et al., 2008) may fail to explain accurately the process by which transgressors respond to legitimacy loss or even shock.

On the one hand, the reintegration literature tended to view organizational responses to legitimacy loss as primarily and continuously stakeholder driven, suggesting that organizational responses closely follow stakeholder claims as evolving over the different reintegration stages (Pfarrer et al., 2008). Similarly, the literature on institutional determinism (e.g. DiMaggio and Powell, 1983) views organizations as passive actors, which isomorphically adapt to and conform with the expectations in their institutional environment. In contrast, the “substantial influence” strategy identified here sheds light on firm-driven responses that proactively shape and influence these stakeholder or institutional demands through substantial organizational actions. That is, the organization defines a new benchmark for the field, which stakeholders then acknowledge and take for granted as new standard. This latter view of organizational actors complements initial arguments on proactivity through “technical actions,” that is, “ones that have the potential to address the causes of wrongdoing and thus attract the media and stakeholders’ attention to the internal processes of a firm” in the crisis and impression management literature (Zavyalova et al., 2012: 1080). Not only did the substantial (or “technical”) actions at Siemens attenuate the negative effect of its own wrongdoing (Zavyalova et al., 2012), but the unprecedented extent of these actions even influenced its peers, the public, and regulators to perceive Siemens as a new compliance benchmark.

On the other hand, previous models pictured MNCs as organizations consisting of rational top management and staff and characterized by well-thought-out decisions based on all relevant facts and options. Pfarrer et al. (2008), for example, describe the outcome of the “discovery stage” as the “facts of the transgression” (p. 735). Similarly, although the reintegration process that Bertels et al. (2014) describe is more nuanced (outlining, for example, the escalation paths of regulatory pressure), the authors primarily consider rational factors and largely ignore the emotional (personal) factors that determine the transgressor’s response to a transgression (see legitimacy shock legend of Figure 2). Also with reference to rationality, a central assumption in institutional theory is that organizations follow “rationalized myths,” that is, taken-for-granted rules or demands in its institutional environment, and hence pursue strategies of decoupling or isomorphism (here: isomorphic adaptation) (DiMaggio and Powell, 1983; Meyer and Rowan, 1977). However, in the context of (high) legitimacy loss, we theorize that not only the “rationalized myths” or “facts of a transgression” (e.g. monetary fines and regulatory sanctions), but also—if not predominantly—the personal and emotional involvement of the organization’s members explain the “substantial influence” strategy of our transgressors. On those grounds, we argue that emotions (Baumeister et al., 2007) are key to conceptualizing how transgressors reintegrate themselves into their environment after a transgression.

Creed et al. (2014) provide an analysis of the mechanisms through which shame, a self-directed emotion, affects people’s willingness to participate in institutional processes. Recognizing that maintaining social bonds is a “crucial human motive” (Scheff, 1990: 4), the authors argue that shame signalizes that social bonds are at risk and triggers efforts to preserve these bonds (Baumeister et al., 2007). We build on the idea that shame can act as a means of intersubjective surveillance and self-regulation and suggest that personal emotions like shame trigger the quest for substantial organizational change to secure or re-establish social bonds. The public-oriented raid by 200 police officers and officials at the Siemens headquarters in Munich, and to a certain extent the prominently published critique of Daimler’s FCPA monitor, can thus be regarded as instances of shaming. Previous reintegration literature has viewed shaming exclusively from the stakeholder perspective and treated it as a later stage in the reintegration process situated after “discovery” and “explanation” (Braithwaite, 1989; Pfarrer et al., 2008: 739). We complement this literature by highlighting that shaming may have an even bigger effect on organizational responses if situated at the very beginning of the transgression, and by explaining the effect of shaming by focusing on the firm (rather than stakeholder) perspective.

Previous research focused on how emotions spread collective corruption within an organization (Smith-Crowe and Warren, 2014). In contrast, our findings outline how emotions can be used to trigger efforts against organizational corruption. We propose that such emotions are crucial to overcome rationalized routines that may have led organizations to continuously apply decoupling and isomorphic adaptation strategies by activating personal attention, motivation, and conviction for substantial change among organizational members. The legitimacy shock at Siemens, by triggering a strong emotional shift from pride to consternation, enabled the general acceptance of radical change. Metaphor-laden narratives such as “change the course or sink” fostered the emotional involvement of the workforce. Similarly, ABB aimed to secure its employees’ personal and emotional involvement by strongly focusing on a value- and principle-based integrity approach rather than a “tiring” rule-based compliance approach. Especially after the monitoring critique, Daimler also tried to reach the “hearts and minds” of its managers and employees by changing their “mentality” and “stigmatizing” corruption.

Varying temporalities in organizational responses: the role of a legitimacy shock versus regulatory sanctions

The Siemens response was marked by a sudden and substantial change in organizational strategy right after the MNC was hit by the legitimacy shock. Daimler, especially before the monitor critique, and ABB after its second transgression followed a much slower and more gradual (Daimler) and cyclical (ABB) reintegration path (see differently curved arrows in Figures 2 and 3(a) to (c)), which is more in line with conceptual reintegration models: Pfarrer et al. (2008) distinguish four reintegration stages: discovery, explanation, penance and rehabilitation.

According to this model, during the “discovery” stage, Daimler waited until most facts had been collected and disclosed, and abstained from a sudden response. In the “explanation” and “penance” stages, Daimler took the time to reflect on why the transgression happened and how the organization should respond, whereas Siemens had practically skipped large parts of the “discovery” and the entire “explanation” stages. We thus propose that in the absence of a legitimacy shock, for example, if “only” pragmatic legitimacy is at risk whereas cognitive legitimacy (the taken-for-granted acceptance of the organization) remains unchallenged, several aspects of previous reintegration models (Bertels et al., 2014; Pfarrer et al., 2008) hold. In this case, transgressors may judge that they can afford to wait until they have all the facts before they seek and provide appropriate explanations, enter into regulatory agreements by way of reparations, and implement organizational changes in line with these agreements, following the four stages (discovery, explanation, penance, and rehabilitation) Pfarrer et al. (2008) identified.

Furthermore, we suggest that in the absence of a legitimacy shock, continued regulatory pressure plays a crucial role in triggering transgressor responses. Repeated or escalating external pressure may be necessary for transgressors to develop comprehensive controls and implement leading practices in the fight against corruption, as the model put forward by Bertels et al. (2014) suggests. Our findings, especially those on Daimler, identify external compliance monitors as very effective—though overlooked in recent research—means of triggering a much higher dynamic in organizational change through escalating pressure. The Daimler case in particular illustrates that the presence of an external monitor during the process of regaining regulatory legitimacy influenced significantly the transgressor’s reintegration strategy. This is also of interest to practitioners and scholars concerned with corporate social responsibility (CSR) initiatives or standards. In the context of schemes like the UNGC, our study shows that in situations where the transgression and the transgressor’s power are not extreme (Daimler as opposed to Siemens), imposing external and stringent monitoring on the disgraced members may facilitate and accelerate their learning progress (Schembera, 2016). While the UNGC so far lacks resources to install direct monitoring (Rasche, 2009), one pragmatic solution would be to establish a fund to collect the necessary resources for appointing teams that monitor UNGC members.

Finally, our findings in the ABB case show that regulatory proceedings like the DPA that are closed only after the successful implementation of a set of agreed organizational corruption controls are effective in cases of low transgression prominence. After its first transgression, ABB was only obliged to pay monetary fines, and the company failed to implement a holistic set of corruption controls. After the second transgression, which was again of low prominence, US regulators and ABB entered a DPA with a list of required corruption controls, which turned out to be effective for ABB to realize more holistic organizational changes.

The two extremes of strategic manipulation: decoupling versus substantial influence

Previous legitimacy research falls short in distinguishing our transgressors’ strategy shift over time, and instead would refer to all strategies as “(strategic) manipulation” (see, for example, Child and Rodrigues, 2011; Oliver, 1991; Scherer et al., 2013; Suchman, 1995). A main reason for this limitation is that previous research refers to manipulation strategies mainly as a public relations tool that organizations employ to manipulate or co-opt the perceptions of critical stakeholders, without actually modifying the practices that these stakeholders criticize (Oliver, 1991: 159; Scherer et al., 2013: 266). To capture the commonly observed strategy shift, we thus propose to distinguish between two extreme variants of strategic manipulation: “decoupling” and “substantial influence” (see Figure 4). Both variants contain an organization’s attempt to manipulate its environment. However, the former involves no or minimal change of actual organizational practices, whereas the latter involves a very comprehensive change of actual organizational practices. “Substantial influence” can hence be defined as an organization’s radical and active effort to (1) manipulate public perceptions and (2) change its own organizational practices. Figure 4 illustrates the difference between the two variants.

Given these differences between the two strategic manipulation variants, we propose to abstain from referring to strategic manipulation in general and, in order to avoid ambiguity, instead to specify whether decoupling or substantial influence is at play. The presented findings support our initial suggestion that in recent legitimacy literature (Scherer et al., 2013), strategic manipulation (Oliver, 1991; Suchman, 1995) can be mostly equated with a “decoupling” strategy (Meyer and Rowan, 1977; see left side of Figure 4). Both strategy concepts are based on the assumption that organizations achieve legitimacy by constructing a facade of organizational structures and processes that reflects institutional myths (shaped, for example, by expectations concerning proper business practices) without adjusting their organizational practices to match the projected image.

We furthermore suggest that the other observed extreme variant of strategic manipulation should be labeled and defined as “substantial influence” (see right side of Figure 4). While elements of this strategy do appear in previous studies—for example, Nike’s decision to disclose the names of its supply-chain partners (Scherer et al., 2013: 268)—such earlier works focused on cooperation with civil society organizations and described this approach as a “proactive moral reasoning” strategy. However, in contrast to “proactive moral reasoning,” the “substantial influence” strategy, as we described and defined it, does not involve open-ended discourses between organizations and their stakeholders, but clear and predefined targets. In sum, distinguishing between the strategies of decoupling and substantial influence allows to contribute to the organizational reintegration (e.g. Pfarrer et al., 2008) and legitimacy (Oliver, 1991; Scherer et al., 2013) literatures by capturing a holistic typology of organizational strategies in the context of legitimacy loss.

Multiple versus single legitimacy strategies: what works when?

Theoretical assumptions on the use of legitimacy strategies vary greatly and we perceive the need to clarify the conditions under which the different assumptions hold true. The contradictions between different approaches—namely, the resource-based view (Christmann, 2000; Hart, 1995), discourse ethics (Stansbury, 2009), and institutional theory (Delmas, 2002; Schaefer, 2007)—give the impression that a corporation only applies one legitimacy strategy at a time. Palazzo and Scherer (2006) argue that choosing between different legitimacy strategies in a globalized economy does not work as smoothly as it once did (see Oliver, 1991). Scherer et al. (2013), therefore, suggest that in heterogeneous environments that place conflicting demands on corporations, a favorable way to maintain or regain legitimacy may be to combine different and potentially contradictory legitimacy strategies.

Our findings on Daimler and ABB reveal that nowadays transnationally operating organizations indeed tend to apply a mix of different legitimacy strategies in order to manage the heterogeneous anti-corruption expectations they face. Both firms combined strategies of “isomorphic adaptation” with “moral reasoning.” Combining “decoupling” with other legitimacy strategies turned out to be problematic, however. ABB temporarily combined elements of “decoupling” even with the other extreme variant of strategic manipulation “substantial influence.” While Scherer et al. (2013) suggested that contradictory legitimacy strategies may be applied simultaneously in complex environments, the reoccurrence of regulatory sanctions at ABB illustrated the limitations of such an approach. The Siemens case, in contrast, indicates that in complex institutional environments, a radical shift from “decoupling” to “substantial influence” appears more promising with regard to avoiding additional regulatory sanctions.

Boundary conditions and future research

A common question in case study research concerns the transferability of findings. We consider a key strength of our study that it analyzes multiple cases from which not only common themes across cases emerged but also fundamental differences; this combination arguably provided us with a good basis for deriving more general suggestions on the relation between legitimacy loss and organizational response. While we studied MNCs as a prototype case of organizations responding to legitimacy loss in a complex environment, our findings may well be transferable to other large and public organizations facing similarly heterogeneous demands. On the one hand, considering that our case firm’s slightly varying industry context was not a major factor in explaining organizational responses, generalizations may also hold across industries. On the other hand, considering that multinational organizations like FIFA (Fédération Internationale de Football Association), IAAF (International Association of Athletics Federations), or UCI (Union Cycliste Internationale) are also large powerful actors facing high public prominence and heterogeneous (anti-)corruption demands, certain elements of our findings may be also transferable to “non-profit” organizations.

As regards the type of transgression, because bribery was a particular common, yet not the only, type of corruption observed at our case firms, and the particular type of transgression helped little in explaining organizational responses, our findings may hold as well for other cases of legitimacy loss. Caution, however, seems necessary when deriving generalizations of our model for organizations that are not originated in low corruption risk countries as studied here. For example, due to potentially different social norms and/or transparency standards (i.e. risk of transgression disclosure and information exchange with Western authorities), a Chinese MNC may experience less of a legitimacy loss in its home country after transgressions similar to the Siemens or Daimler case. Consequently, future research may focus on responses to legitimacy loss of organizations originated in such perceived high-corruption risk countries.

We further acknowledge possible limitations with regard to the global level of analysis that we applied in our study. While we aimed to include a global perspective in our study—by obtaining data from various MNC locations worldwide—our focus was more on the central and common processes in the transgressors’ global strategy change. It was beyond the scope of this article to scrutinize in depth differential strategy shifts across MNC subsidiaries (Kostova and Roth, 2002). Our level of analysis was hence predominantly global rather than local. Future research is needed to zoom in on these differential processes across the various locations of multinational organization, for example, by looking at different types of network structures, value systems, and power relations.

To conclude, by empirically analyzing organizational strategies in the context of legitimacy loss, we found a common shift from “decoupling” to “substantial influence.” Our findings go beyond the existing literature in that they show how emotions enable organizations to deviate from rationalized routines or “myths” to realize substantial firm-driven change that goes beyond institutional or stakeholder expectations. We further outlined that there seems to be no prototype or ideal way of responding to legitimacy loss; rather, several transgression characteristics such as previous sanctions, prominent disclosures, and external enforcement pressure determine the chronology of strategies and speed of strategy shifts. Several studies have theorized this response process from different angles (Bertels et al., 2014; Pfarrer et al., 2008). However, future research is needed to empirically validate or specify existing models as well as focus on the particular challenges of multinational organizations operating in complex heterogeneous environments.