Master’s Thesis IB&M
An integrated approach on how context matters: institutionalism and the influence
of host country corruption on subsidiary behavior
Author: Maikel Klein Swormink Student number: S2906716
Supervisor: J. Canello Co-assessor: E. Mendiratta
Date: 17-06-2019
2 ABSTRACT
Previous studies raised notable concerns for the neglected integration of contextual factors when investigating the headquarters-subsidiary relationship. Foreign subsidiaries face institutional complexities when they are operating in an environment that is fundamentally different from its home country context. This research endeavors for a sophisticated examination of the
relationship between host country corruption and subsidiary behavior. I draw upon a sample of 10,823 foreign subsidiaries, dispersed over 18 developing countries. An empirically comparative approach has been adopted. The results corroborate the importance of contextual factors when investigating the way foreign subsidiaries behave. Particularly when the subsidiaries are
operating in more corrupt environments. The findings suggest that host country corruption tends to strengthen the level of subsidiary autonomy. Contrary, corruption impedes foreign subsidiary initiative. The implications following from the results build upon the critical call for strategic-level reconsiderations trough the influence of environmental factors. In the grasp of a corrupt environment, subsidiary managers should understand the potential of self-inserted initiative-taking behavior.
3 TABLE OF CONTENTS
1. INTRODUCTION ... 5
2. LITERATURE AND HYPOTHESES ... 8
2.1 Subsidiary autonomy, subsidiary initiative and institutional theory ... 12
2.2 Host country corruption and subsidiary autonomy ... 15
2.3 Host country corruption and subsidiary initiative ... 17
3. METHODOLOGY ... 23
3.1 Sample and procedure ... 23
3.1.1 Orbis database... 24
3.1.2 Business Environment and Enterprise Performance Survey V and Middle East and North Africa Enterprise Surveys (BEEPS V and MENA ES) database ... 24
3.1.3 World Bank WGI database ... 26
3.1.4 Transparency International database ... 26
3.2 Methods ... 26
3.2.1 Dependent variables ... 28
3.2.2 Main independent variable ... 29
3.2.3 Control variables... 30
4. FINDINGS... 33
4.1 Descriptive Statistics ... 33
4.2 Correlation analysis ... 35
4.3 Multicollinearity and heteroscedasticity ... 38
4.4 Factor analysis ... 38
4.5 Hypotheses testing... 39
4.6 Robustness checks ... 44
5. DISCUSSION ... 46
5.1 Theoretical implications ... 46
5.1.1 Host country corruption facilitates subsidiary autonomy ... 46
5.1.2 Host country corruption as a disincentive for subsidiary initiative ... 47
5.1.3 Host country corruption and the relation between subsidiary autonomy and subsidiary initiative ... 48
5.2 Managerial implications for MNCs... 49
4
6. CONCLUSION ... 51
7. APPENDICES ... 53
8. REFERENCES ... 62
LIST OF FIGURES Figure 1 Graphical illustration of factors affecting subsidiary behavior ... 15
Figure 2 Conceptual model ... 22
Figure 3 Graphical illustration of all control variables included for analysis ... 57
LIST OF TABLES Table 1 Descriptive statistics... 34
Table 2 Correlation matrix ... 36
Table 3 Correlation matrix corruption measures ... 37
Table 4 Corruption BEEPS factor variable ... 39
Table 5 Multiple linear regression results host country corruption and subsidiary autonomy .... 40
Table 6 Logit regression results host country corruption and subsidiary initiative ... 42
Table 7 Hypotheses summary ... 45
Table 8 The initial 32 developing countries reduced to the 18 countries ... 55
Table 9 Number of subsidiaries per country ... 56
Table 10 Description of the 6 initial BEEPS V and MENA ES corruption measures ... 58
Table 11 Multiple regression results host country corruption and subsidiary autonomy ... 59
5 1. INTRODUCTION
Institutional differences can cause multinational corporations (MNCs) to intentionally distance its headquarters from its foreign subsidiary. The purpose is to protect itself against reputational damages when the subsidiary operates in a host country environment that is fragile or where stakeholder expectations are conflicting with those of the MNC's home country (Rabbiosi & Santangelo, 2018). The MNC is construed as a heterogeneous inter-organizational entity (Ghoshal and Bartlett, 1990), and if the HQ has insufficient knowledge concerning the network contexts of its foreign subsidiaries, the HQ is restricted in its potential for influential power and commitment (Holm, Johanson, & Thilenius, 1995). When foreign subsidiaries experience severe pressures to abide to local expectations and local industry levels of their institutional
environment, MNCs tend to elect reduced ownership levels in exchange for external legitimacy in the foreign institutional environment that their foreign subsidiaries are operating in (Chan & Makino, 2007). The right course of action for coordinating foreign subsidiaries remains a persisting inquiry for MNCs. Especially due to the presence of institutional complexities
embedded in the HQ-subsidiary relationship. The potential consequence of restricted influential power is generally always present owing to the continuous dynamics of the network structure in which the HQ-subsidiary relationship is ingrained. The dynamics of the MNC’s network
structure are shaped by the interaction with suppliers, customers, competitors, and other contextual factors (Holm, Johanson, & Thilenius, 1995).
Inter-subsidiary differences are forged through the inter-organizational network, subsidiary-specific characteristics, and external pressures (Forsgren, Pedersen, & Foss, 1999; Hamprecht & Schwarzkopf, 2014). These factors affect the way subsidiaries behave. Examples on internal factors include the influence of socialization, centralization, procedural justice towards subsidiary managers, and formalization (Gupta, Govindarajan, & Malhotra, 1999). Internal factors that cause deficiencies within the HQ-subsidiary relationship regarding the adoption of innovation can be addressed by internal organizational processes, such as intensifying
6 Especially when it concerns internal approaches such as intensifying communication as a means of HQ-subsidiary monitoring which fosters internal embeddedness. Increasing internal
embeddedness may be at the expense of the positive impact of external embeddedness on subsidiary importance. For instance, diluted external embeddedness through increased internal embeddedness impedes subsidiary importance to production development (Yamin & Andersson, 2011). External pressures drawn from the local environment can be marked as determinants of subsidiary strength (Forsgren et al, 1999), which imposes context specific demands on the foreign subsidiary. The social context is an important factor to consider in the HQ-subsidiary relationship (Hoenen & Kostova, 2015; Cavanagh, Freeman, Kalfadellis, & Cavusgill, 2016). A subsidiary with dynamic linkages established in the host country has access to local knowledge and resources, whereby a diverse context can put pressure on subsidiary creativity and
development (Dimitratos, Liouka, & Young, 2009). In such situations, subsidiaries may opt for a localized strategy to obtain legitimacy from its local stakeholders.
The process for MNCs to deal with diverse institutional environments can leave the subsidiary with the freedom to pursue autonomous actions. On the contrary, subsidiaries will be
discouraged to pursue entrepreneurial actions in high corrupt environments (Dutta & Sobel, 2016). The subsidiary may operate in an environment where there is considerable tolerance for corruption and bribery as opposed to the MNC's home country. In such occasions, MNCs face opposing stakeholder expectations. Challenges derived from dealing with diverse institutional environments can therefore result in different consequences for the way a MNC’s subsidiaries behave. While investigating the subsidiary's initiative process, it is important to consider
7 host country corruption and the level of subsidiary autonomy. Thereby, a decentralized
organizational decision strategy enables for more effective decision-making through better use of local information (Rabbiosi & Santangelo, 2018). Host country corruption tends to foster
subsidiary autonomy while it may have a distinct impact on subsidiary initiative as the subsidiary needs to deal with environmental complexities, such as corruption.
An integration of environmental factors provides a deeper understanding on the HQ-subsidiary relationship. Extant research calls for additional specific contextual factors that can benefit the level of subsidiary initiative (Cavanagh et al, 2016; Kostova, Nell, & Hoenen, 2016;
8 influence of subsidiary autonomy on subsidiary initiative. The research question is formulated as following:
“To what extent does host country corruption affect the level of subsidiary autonomy and subsidiary initiative?”
This research uses a sample of 10,823 foreign subsidiaries operating in the manufacturing of machinery and equipment industry for which data has been extracted from Orbis. The role of corruption is evaluated using data coming from three different sources, the Business
Environment and Enterprise Performance Survey V and Middle East and North Africa Enterprise Surveys (BEEPS V and MENA ES), the World Bank, and Transparency International. I build on existing literature by following the shift from internal factors that affect the HQ-subsidiary relationship to the addition of external factors. Furthermore, I discuss the concepts of subsidiary autonomy and subsidiary initiative, and how they are impacted by the environmental factor, host country corruption. The purpose is to acquire an integrated understanding on how host country corruption influences subsidiary autonomy and subsidiary initiative. Next, I evaluate how the influence of subsidiary autonomy on subsidiary initiative varies based on the level of host
country corruption. Hypotheses have been established accordingly. In the methodology section, I outline the comparative empirical approach that adds reliability to this research. The hypotheses will be tested, after which the results will be presented. Finally, the results are discussed to derive policy implications on how MNCs should refine strategic level decisions on subsidiary behavior. The implications particularly relate to MNCs that face opposing institutional pressures through a foreign subsidiary that is operating in an environment where corruption is perceived as a common business practice.
2. LITERATURE AND HYPOTHESES
The multinational corporation (MNC) is an organizational vehicle of the headquarters and geographically dispersed subsidiaries that enables for the transfer of internal codified- and tacit knowledge (Kogut & Zander, 1993). MNCs are a means to organizational efficiency as
9 knowledge recombination internally in the organization is key to the MNC’s performance (Kogut & Zander, 1993; Mudambi & Navarra, 2004). Codified knowledge can be spread relatively easy throughout the MNC as codified knowledge is characterized by ‘information’. Tacit knowledge is more holistic and may require strong company specific networking capabilities or unique national subsidiary competencies. The differences between the two types of knowledge therefore largely relates to articulation difficulties (Mudambi & Navarra, 2004).
As of the late 1980s, academic interest in the functioning of MNCs emerged strongly.
Particularly, the multinational subsidiary became a more central unit of analysis to understand the different strategic roles. New conceptualizations of the MNC were introduced (Birkinshaw & Morrison, 1995). Prior, interorganizational theory was the main avenue to evaluate the complex organizational structure of MNCs, with some adaptations to interorganizational theory along the way. The MNC was conceptualized as an internally differentiated interorganizational network embedded in a broader external network (Ghoshal, & Bartlett, 1990). However, when
considering subsequent factors that shape subsidiary behavior, external factors to the MNC became a more evident point for consideration within the debate on what shapes subsidiary behavior. Throughout this research, the term subsidiary behavior is adopted to capture both the concepts of subsidiary autonomy and subsidiary initiative.
While not losing track of the importance of internal factors, literature moved from an internal organizational approach to the consideration of external factors. For instance, the concept of institutional duality was introduced whereby the host country institutional profile and relational ties within the MNC network affect the adoption process of organizational practices by
10 same environment will employ similar practices. Isomorphism is a process of homogenization that forces distinct units operating in a certain environment to resemble to other units that face the same set of environmental conditions (DiMaggio & Powell, 1983). For the subsidiary, isomorphic pressures implies that subsidiary behavior is not only guided by pressures from formal institutions, but also based on normative and cognitive component (Hotho & Pedersen, 2012). MNCs can face isomorphic institutional pressures which push subsidiaries to conform to local stakeholders’ expectations. Isomorphic pressures push local companies to engage in behavioral norms that are more alike, which results in increasing compatibility of environmental characteristics (DiMaggio & Powell, 1983). Legitimacy is obtained by abiding to local practices and pressures from local stakeholders. If there is an increase in the regulatory, cultural, and economic distance between home- and host country, firms select a more pronounced strategy of local isomorphism (Salomon & Wu, 2012). The literature moved away from the dyadic HQ-subsidiary relationship to an integrative network of foreign subsidiaries (Bartlett & Ghoshal, 1987b; Ghoshal & Bartlett, 1990), within a broader network of external relations and
environmental forces that influence subsidiary behavior (Ghoshal & Bartlett, 1990). The geographically dispersed subsidiaries within the MNC face different environmental
contingencies due to an internal structure that is heterogenous and systematically differentiated (Ghoshal & Nohria, 1989).
An MNC’s environmental concerns, such as isomorphism, disable MNCs to exclude environmental considerations from subsidiary strategy-making. This resulted in novel
conceptualizations of the MNC towards a transnational form as of the 1980s. The unidimensional focus on either efficiency, learning, or responsiveness was no longer sufficient due to a complex set of environmental forces. Firms now needed multidimensional strategic capabilities to
outcompete rivals (Bartlett & Ghoshal, 1987a). Furthermore, the transition meant an
11 MNCs need to identify what patterns of integration or differentiation are most effective for large organizational systems to deal with the external environment. Differentiation is the segmentation of subsystems, such as R&D or marketing, whereby each subsystem develops specific attributes that relate to the demands imposed by the external environment. Integration is about the unity of effort among the different subsystems while aiming for the achievement of an organizational task (Lawrence & Lorsch, 1967). MNCs frequently face trade-offs between local responsiveness and global integration.
Through systematically differentiated responsibilities and tasks, multidimensional organizations seek to maximize profits through the right balance between local responsiveness and global integration (Bartlett & Ghoshal, 1987b). In order to get there, such organizations create sophisticated strategies including decisions on the level of centralized decision-making,
subsidiary independence, and mechanisms for coordination and cooptation. These can be defined as core organizational characteristics that transnational organizations utilize to address
challenges when operating in different contextual environments (Bartlett & Ghoshal, 1987b). Martinez and Jarillo (1989) introduce the notion for more pronounced multidimensional organizations by building on literature for an even more changing international competitive environment. Organizational structures are developed to fit the changing international
environment. Thereby, an evolutionary pattern can be identified which, for instance, resulted in rising popularity of informal coordination mechanisms as a reaction to changing environmental influences (Martinez & Jarillo, 1989). The essence of the need for transnational capabilities is the insufficiency of unidimensional strategic capabilities as they do not fit the concept of
isomorphism. Foreign subsidiaries face pressures from the MNC’s headquarter as well as the need to abide to local expectations. Organizations that had adopted a global strategic posture for the purpose of efficiency faced the challenge for national responsiveness and access to
innovative resources globally. Firms with local responsiveness as the dominant posture faced the challenge for global efficiency and improved worldwide knowledge creation and internal
12 In conclusion, the MNC with transnational/multidimensional capabilities is embedded in a network of external relations, including regulators, communities, and suppliers. (Ghoshal & Bartlett, 1990). The MNC is a geographically dispersed inter-organizational network, whereby the different national subsidiaries can have conflicting goals. Different MNC components relate to specific structural properties of the MNC’s external network. In the differentiated MNC network, no formal macrostructure exists that matches with all the organization’s heterogeneous environments. This has led to differentiated subsidiary coordination mechanisms to fit the varying subsidiary contexts. Internal managerial processes have changed as subsidiaries adopted new ways to respond to changing local conditions. The same holds for strategy formulation. Subsidiary-level strategy adjustments are needed to fit the functioning of the subsidiary with expectations of the local environment (Ghoshal & Bartlett, 1990). This network perspective on the MNC in the era where multidimensional strategic capabilities are a necessity highlights the importance of external forces that influence subsidiary behavior.
2.1 Subsidiary autonomy, subsidiary initiative and institutional theory
Organizational structures that encompass strong diversity in levels of subsidiary autonomy and subsidiary initiative can be explained by determinants of the MNC’s headquarters, the subsidiary itself, and the host country that the subsidiary is embedded in (Simoes, Biscaya, & Nevado, 2000; Boojihawon, Dimitratos, & Young, 2007; Hamprecht & Schwarzkopf, 2014). MNCs opt for organizational structures whereby the parent firm assigns different levels of autonomy to its subsidiaries. Such organizational structures are embedded in transnational organizations forged by firm level factors as well as environmental complexities (Bartlett & Ghoshal, 1987b).
Interorganizational theory with (internal) organizational factors as the sole explanatory power for subsidiary behavior was reconsidered through a more integrated perspective by including
environmental forces. By drawing upon institutional theory, foreign subsidiaries face a higher pressure for local conformity if they experience greater corruption distance (Rabbiosi & Santangelo, 2014).
13 subsidiary within the MNC network. Organizational socialization helps to overcome subsidiary reluctance towards the adoption of innovation from the HQ. Such organizational processes within the MNC's HQ-subsidiary relationship contribute to a better integration of subsidiary initiative. For instance, intense HQ-subsidiary- and inter-subsidiary communication can facilitate integration of the subsidiary (Ghoshal, & Bartlett, 1988). Next to these networking and
organizational determinants, the local environment puts an isolated pressure on the
entrepreneurial characteristics of an MNC's subsidiary. Specific host country characteristics, such as the level of dynamism, are determinants for a subsidiary's entrepreneurial activities (Boojihawon et al., 2007). The local institutional environment drives subsidiary behavior.
In light of firm level factors as determinants for particular organizational structures, those are both defined by the HQ and the subsidiary itself. For instance, in terms of subsidiary autonomy, the level of autonomy can be predefined by the HQ or is a form of remuneration as it is earned by the subsidiary based on performance. Also, it can be based on the type of decisions that determine whether HQ approval is required (Cavanagh et al, 2016). Regarding the HQ role in defining the level of subsidiary autonomy, the HQ can stimulate subsidiary managers to act autonomous if the corporate strategy aims to dissociate the HQ from a foreign subsidiary. A matter of forced autonomy. This may occur when the HQ wants to protect itself against potential reputational losses (Rabbiosi & Santangelo, 2018). Next, characteristics specific to the subsidiary shape subsidiary autonomy. Subsidiary managers can take independent actions whereby they behave more autonomous to diminish inefficiencies or to speed up the decision-making process. This relates to assumed autonomy (Cavanagh et al., 2016).
The concept of subsidiary initiative involves risk-taking behavior, a proactive attitude,
14 Prior research reviewed extant literature on subsidiary initiative to ultimately display subsidiary initiative as a multi-dimensional construct (Dörrenbächer & Geppert, 2009; Strutzenberger & Ambos, 2014). Dörrenbächer and Geppert (2009) distinguish between the corporate (MNC) context, subsidiary-specific factors, and the context of the local environment to identify
determinants of subsidiary initiative. Strutzenberger and Ambos (2014) focus on the antecedents, implementation, and the outcomes of subsidiary initiative. The concept of subsidiary initiative can be approached as a process (Strutzenberger & Ambos, 2014). Subsidiary initiative is largely shaped by autonomous actions (Birkinshaw, Hood & Jonsson, 1998; Boojihawon et al., 2007) and pursuance of an entrepreneurial culture. An entrepreneurial atmosphere will foster risk- and initiative-taking behavior (Birkinshaw et al., 1998), which relates strongly to the overall
construct of subsidiary initiative as defined earlier. It is essential to acknowledge the specific context, including environmental forces, home country pressures, and the external network that the subsidiary is embedded in. This holistic approach explores the origins of subsidiary strategy making and highlights how subsidiary initiatives lead to a stronger overall impact of the
subsidiary (Strutzenberger & Ambos, 2014).
Based on the evolutionary approach and the comparative institutionalism approach, the HQ and the subsidiary are both main actors in shaping subsidiary initiative. Both actors need to deal with different environmental contingencies and their behavior is determined by the home- and host country institutional settings (Dörrenbächer & Geppert, 2009). Severe pressures from the host country institutional environment can trigger subsidiaries to undertake specific subsidiary initiatives. For instance, a high level of environmental pressures can ensure subsidiaries to adopt climate mitigation initiatives (Hamprecht & Schwarzkopf, 2014), whereby the subsidiary needs to abide to local expectations to obtain local legitimacy. Hence, there are environmental forces at stake which affect subsidiaries behavior. Lastly, business networking affects subsidiary behavior. For instance, higher levels of internal networking may result in higher intra-organizational
15 the impact of host country contextual factors, with special emphasis on the influence of host country corruption.
Figure 1 Graphical illustration of factors affecting subsidiary behavior
2.2 Host country corruption and subsidiary autonomy
Among the different external factors that affect the HQ-subsidiary relationship, corruption is expected to be particularly relevant. Corruption is the abusive exercise of public power for private gains that breaches the rules of the game. Corruption is grounded in an illegitimate nature with government officials that misuse the public office for personal benefit (Cuervo-Cazurra, 2016). First, corruption reflects the concerns of isomorphism. Following institutional theory, foreign subsidiaries experience a greater need for local conformity when the local corruption level varies more strongly from the home country corruption level (Rabbiosi & Santangelo, 2014). Due to severe corruption differences between the home and host environment, the HQ may distance itself from the foreign subsidiary operations (Rabbiosi & Santangelo, 2018).
16 Second, local corruption affects subsidiary strategy formulation. The institutionalization of corrupt practices relates positively to the pressure that the MNC’s subsidiary faces to engage in corruption locally (Spencer & Gomez, 2011). Hence, corruption is an evident representation for the impact of external factors on subsidiary behavior.
As discussed earlier, subsidiary autonomy may be predefined by the HQ, shaped by the
efficiency seeking subsidiary, or emerge gradually through subsidiary performance, whereby a distinction exists between assigned autonomy and assumed autonomy (Cavanagh et al., 2016). Host country corruption supplies concerns for the MNC, that can be different for the HQ in relation to the subsidiary. Host country corruption could alter the HQ’s perspective on the appropriate level of subsidiary autonomy or the subsidiary’s perspective for distinct reasons.
An increased regulatory, economic and cultural distance between the home- and host country stimulates a more pronounced strategy of local isomorphism (Salomon & Wu, 2012). The decision of subsidiary management to engage in host country corruption applies to such distinction between institutional environments. Spencer and Gomez (2010) found a positive relationship between the level of host country corruption and pressures perceived by the subsidiary that are derived from the local environment to engage in bribery. In line with
17 environment demands a localized strategy, with a decentralized decision-making approach and severe level of local embeddedness. Hence, if the subsidiary encounters higher levels of corruption compared to the MNC home country, it will result in a higher level of subsidiary assumed autonomy (Cavanagh et al., 2016).
Second, environmental complexities can be a burden for obtaining local legitimacy and to information processing. Decentralized decision-making supports efficiency by minimizing the associated costs and it diminishes discrimination and adverse treatment by local stakeholders (Rabbiosi & Santangelo, 2018). Next, subsidiary autonomy through decentralized decision-making can protect the MNC against reputational damages when the subsidiary is subject to a host country corruption scandal, as the MNC compartmentalizes its identity and blames the subsidiary for wrongdoing (Rabbiosi & Santangelo, 2018). In this case, it may be that the HQ assigns a higher level of autonomy to its foreign subsidiary (Cavanagh et al., 2016).
In line with facing opposing isomorphic pressures, operating in different institutional environments requires subsidiaries to comply with host country normative- and cognitive distinctiveness, even if this concerns tolerance for engagement in host country corruption (Rabbiosi & Santangelo, 2018). Hence, a direct positive impact of host country corruption on subsidiary autonomy is expected. With respect to host country corruption and subsidiary autonomy, the following hypothesis has been established:
Hypothesis 1. An increased level of host country corruption will result in an enhanced level of subsidiary autonomy.
2.3 Host country corruption and subsidiary initiative
18 al, 2007; Hamprecht & Schwarzkopf, 2014). Corruption is regarded to have a pronounced
external impact on subsidiary initiative due to the strong association with the concept of isomorphism and its implications on the HQ-subsidiary relationship as indicated before. For instance, corruption is particularly relevant as it generally inflicts domestic entrepreneurship (Dutta & Sobel, 2016).
First, while considering both direct and indirect effects of corruption, the total impact of corruption never demonstrates a positive impact on entrepreneurship (Dutta & Sobel, 2016). Dutta and Sobel (2016) measure entrepreneurship by the number of newly registered firms over a 7-year time period. Their results clearly indicate that corruption hurts entrepreneurship.
Corruption dilutes economic initiative which hampers local firms’ entrepreneurial performance. This issue starts with local actors that face difficulties to pursue entrepreneurial activities. As a result of the corrupt environment, starting a business encompasses dealing with environmental complexities that discourages local actors from entrepreneurial activity. Consequently,
opportunities for the subsidiary to acquire local firms logically wise decreases when there is a reduced number of newly registered firms. Hence, besides the direct negative impact of corruption on local actors’ entrepreneurial intentions, it also translates into impediments to foreign subsidiaries to pursue entrepreneurial activities. Ceteris paribus, corruption slows down a subsidiary’s entrepreneurial engine.
Second, on the national level, meager control of corruption is associated with lower levels of innovation and entrepreneurship, because inadequate efforts to control corruption hampers trust (Anokhin & Schulze, 2009). A subsidiary's entrepreneurial culture is, among others, reflected in its risk-taking behavior as an entrepreneurial culture has a strong and positive relationship with subsidiary initiative (Birkinshaw et al., 1998). Foreign firms operating in an institutional
19 innovative activity (Anokhin & Schulze, 2009). The institutional environment determines the nature and extent of the transaction costs. In line with institutional theory, if the subsidiary is facing institutional differences through strong divergence in the level of corruption, it will inflict the ability, and increase costs, to innovate (Chadee & Roxas, 2013). Host country corruption discourages innovation as local actors are more likely to act opportunistically (Dheer, 2017). The local providers of resources that support (subsidiary) innovation experience reliability and integrity issues if host country corruption is severe, especially in light of impersonal law
enforcement which makes it risky for subsidiaries to pursue innovative opportunities (Anokhin & Schulze, 2009).
Subsidiary initiative can be explained by determinants of the MNC’s headquarters, determinants from the subsidiary itself, and host country determinants (Simoes, Biscaya, & Nevado, 2000; Boojihawon et al., 2007; Hamprecht & Schwarzkopf, 2014). With respect to the latter, the concept of institutional duality results in different inter-subsidiary strategies when facing
opposing isomorphic pressures. The MNC needs to address opposing isomorphic pressures when operating in different environments, to obtain legitimacy across the different environments that it is operating in (Hotho & Pedersen, 2012). These different institutional environments bring out distinct effects on a subsidiary's creativity as well as its reaction to facilitate further development (Dimitratos et al., 2009). Regulatory quality, rule of law, and corruption have an inverse effect on firm innovation (Chadee & Roxas, 2013). Institutional voids, such as instability regarding peace and order, infringement of property rights, and unreliable legal or judicial systems to settle disputes and enforce legal contracts enhance the level of transaction costs and undermine a firm's ability to innovate. Consequently, corruption will have a negative impact on a subsidiary's
innovative performance (Chadee & Roxas, 2013). Based on the delineated argumentation above, the following hypothesis has been established.
20 2.4 Host country corruption, subsidiary autonomy, and subsidiary initiative
Subsidiary autonomy is strongly associated with subsidiary initiative, as subsidiary autonomy is a necessary condition for subsidiary entrepreneurship, whereby autonomy refers to the freedom to pursue entrepreneurial activities (Boojihawon et al., 2007). Subsidiary initiative can support a subsidiary’s full exploitation of organizational capabilities through decentralized decision-making (Delany, 2000). Thereby, decentralized decision-decision-making makes it easier for the
subsidiary to react to local stakeholder expectations. With other conditions remaining the same, subsidiary initiative leads to efficiency gains in the local market (Birkinshaw et al., 1998).
Moreover, subsidiaries with a high level of autonomy foster creation and diffusion of innovation, but subsidiary autonomy does not benefit the subsidiary adoption process (Ghoshal, & Bartlett, 1988). Subsidiary creation and diffusion of innovation reflects a proactive attitude with an entrepreneurial culture facilitated by subsidiary autonomy. Ceteris paribus, if the subsidiary becomes more autonomous, it will result in a higher level of subsidiary initiative. In the sections below, I investigate the relationship between subsidiary autonomy and subsidiary initiative while considering the presence of corruption.
21 The second argument is based on a shifted focus in the literature. Instead of the foreign
institutional environment that shapes the way MNCs perform business, the attention is now based on the MNC whose presence affects the foreign environment. The MNC affects the corrupt institutional environment. The MNC becomes more actively involved as it functions as a regulator (Kwok & Tadesse, 2006). The MNC introduces fresh perspectives on business
practices which promotes debates on better ways to conduct business and challenges the legitimacy of existing patterns. This is referred to as the ‘de-institutionalization’ of local firms’ existing organizational patterns (Kwok & Tadesse, 2006). However, efforts by the subsidiary that result in de-institutionalization may slow down the pace of implementing innovative actions due to uncertainties about potential obstacles and local resistance. Whereas the MNC initially assigns significant levels of autonomy to its foreign subsidiary, or the subsidiary assumes high levels of autonomy due to HQ’s environmental unfamiliarity, subsidiary efficiency seeking, or HQ-facilitated autonomy (Cavanagh et al. 2016), dealing with the complexities of the corrupt environment prevents the subsidiary from translating this freedom into the pursuance of innovative or entrepreneurial initiatives.
Third, corruption can reward unproductive firm behavior as unmerited rights and contracts are granted to firms that engage in bribing and corrupt activities which disadvantages firm efficiency and innovativeness. Corruption provides firms the option to buy-out of costly requirements in rigorous environments (Rodriguez, Uhlenbruck, & Eden, 2005). Subsidiary autonomy, either assigned by the HQ or assumed by the subsidiary, may therefore not convert into proactive, innovative, and/or initiative-taking actions.
22 complexities when performing business. The positive impact of subsidiary autonomy on
subsidiary initiative may not be as pronounced, as dealing with the environmental complexities will refrain the subsidiary from pursuing entrepreneurial activities. Additionally, the argument holds that unmerited rights and contracts may be granted to subsidiaries that engage in corrupt activities, which results in an abated tendency for firm efficiency and innovativeness (Rodriguez et al., 2005). Thus, host country corruption will have a negative effect on the relation between subsidiary autonomy and subsidiary initiative, as the subsidiary is wasting resources in dealing with a corrupted environment.
Hypothesis 3. The positive impact of subsidiary autonomy on subsidiary initiative is mitigated by an increasing level of host country corruption.
Figure 2 depicted below provides a graphical illustration of this research.
23 3. METHODOLOGY
3.1 Sample and procedure
The study of subsidiary behavior has been conducted on a sample of firms operating in the manufacturing of machinery and equipment industry. The decision to focus on this sector is twofold. The decision is motivated by the fact that MNCs operating in this industry tend to be highly innovative, and often face challenges due to the global integration vs local responsiveness controversy. With respect to subsidiary initiative, a study on factors affecting innovative output gathered data on the number of innovations across industries in the US in 1982. Of the 8,074 innovations reported by the US Small Business Administration, 4,476 related to the
manufacturing industry (Acs & Audretsch, 1988). In a study by Boojihawon, Dimitratos, & Young (2007) on multinational subsidiary entrepreneurial culture, the authors focus on the advertising industry as local- vs global responsiveness pressures are at place (Boojihawon et al., 2007), which requires the subsidiary to abide to local desires while not losing track of integration in the MNC's global environment. A trade-off exists between time- and cost benefits from global efficiency and fully customizing to local desires to match with local expectations. In the
manufacturing of machinery and equipment industry, firms are pressurized by innovative developments.
24 developing countries should empower the results of the empirical analysis on how this affects subsidiary behavior. The primary database containing firm-level data is Orbis. This information has been integrated with country-level data on corruption and other environmental factors coming from three different data sources: The Business Environment and Enterprise
Performance Survey V and Middle East and North Africa Enterprise Surveys (BEEPS V and MENA ES) database, the World Bank, and Transparency International. By using this integrated approach, I aim to conduct a comprehensive empirical analysis which allows to provide an all-encompassing overview of the relationship between host country corruption and subsidiary autonomy and -initiative.
3.1.1 Orbis database
Orbis was created by a combination of around 160 separate sources, next to hundreds of Bureau van Dijk's own sources. The Orbis database provides firm-level data on approximately 275 million companies across the globe operating in a variety of industries. The database enables for analysis of company-specific information, such as financials, intellectual property, and corporate structure (Bureau van Dijk, 2019). The Orbis database includes 140,307 companies spread across the globe for the machinery and equipment industry. For each of the 29 countries for which data is available (gathered over the period 2012-2015) by the BEEPS V and MENA ES database regarding the independent variable, the number of observations has been analyzed per country for the dependent variables included. The countries included for analysis derived from the Orbis database had to satisfy the criteria of data availability for at least ten companies per country for the dependent variables. For the final dataset, 10,823 foreign located subsidiaries were included from which data had been derived from the Orbis database. Prior to the empirical analysis, the dataset derived from the Orbis database was subject to refinements to conduct proper analysis. The data cleaning process for the Orbis database can be found in Appendix A.
3.1.2 Business Environment and Enterprise Performance Survey V and Middle East and North Africa Enterprise Surveys (BEEPS V and MENA ES) database
25 2015. In total, the database presents firm-level data for 41 developing countries. Due to missing data on the independent variable, 9 countries have initially been deleted. As data for Russia was gathered in the period 2011 to 2012 and data for Cyprus and Greece in 2016, these countries were omitted for analysis as well. Thus, the database provides data for 29 developing countries that are relevant for analysis.
26 3.1.3 World Bank WGI database
Data for the independent variable, host country corruption, is acquired from the BEEPS V and MENA ES database, as well as from the Transparency International dataset, and the World Bank WGI dataset. The World Bank is comprised of the International Bank for Reconstruction and Development and the International Development Association, which consists of five institutions. The World Bank provides financial and technological benefit to developing countries around the globe (World Bank Group, 2019a). The World Governance Indicators provided by the World Bank is a set of individual and aggregate governance indicators for six dimensions, gathered over the period 1996 to 2017 for over 200 countries (World Bank Group, 2019b). The data on host country corruption from the WGI World Bank dataset is based on the ‘control of corruption estimate’ for which data has been gathered for the year 2015. All countries that will not be included for analysis have been deleted from the dataset. Consequently, the World Bank WGI has been merged with the master dataset.
3.1.4 Transparency International database
Transparency International is an organization that provides aid to witnesses and victims of
corruption by fighting against bribery and the abuse of public power (Transparency International, 2018). The corruption perception index is based on the perceived level of corruption in the country’s public sectors. Countries are ranked based on the national corruption level, with a value ranging from 0 to 100. The final dataset was prepared by adding the Transparency International dataset to the master dataset.
3.2 Methods
The empirical analysis is implemented using a two-step approach. In the first part, the impact of internal and external factors on subsidiary autonomy is assessed using a multiple regression analysis. In selecting an applicable multivariate statistical method, I first specify a foreign subsidiary’s level of autonomy as a log-linear function of a set of independent variables. The model of the multiple log-linear regression analysis has been denoted as following:
27 Where,
Y = 𝑒β0 + 𝑋β1 + 𝐶β2 + 𝑍β3 + 𝐶𝑜𝑟𝑟β4 + 𝜀
Where Y represents the autonomy level of the foreign subsidiary, vector X is a set of firm specific variables, vector C is a set of country-specific variables, Z represents country- and industry fixed effects, and Corr reflects host country corruption. The error term is denoted as ε. The scalar β4 (corruption coefficient) is consulted to test Hypothesis 2. The base of the natural log is reflected by e. A one-unit change in X results in a change for Y by 100 ∙ (𝑒β𝑋 – 1)%. The established models can be compared through the adjusted R2 value. The adjusted R2 value explains how well the variation of the dependent variable is explained by the independent variables.
In the second part, the influence of host country corruption on subsidiary initiative is evaluated using a logit regression approach. The influence of internal and external factors on subsidiary initiative will be examined. Furthermore, the model includes the interaction term between subsidiary autonomy and host country corruption. The third hypothesis is tested using the following advanced specification: The logit regression model is specified as follows:
Inno = 1𝐼𝑛𝑛𝑜>0,
(2) Inno = δ + Xβ1 + Cβ2 + Zβ3 + Corrβ4 + Autoβ5 + (Auto x Corr)β6 + ε
28 increase in the continuous predictor variable, the odds that the dependent variable is positive (value of 1 for subsidiary initiative) increases by the odds ratio of factor X.
The comparative nature of this research lays within the adoption of data on the independent variable, host country corruption. I add methodological value by comparing the BEEPS V and MENA ES database with the World Bank database, and the Transparency International database. All databases provide measures for the independent variable, host country corruption. First, I run a multiple regression analysis by relying on the BEEPS V and MENA ES database for data on host country corruption, with subsidiary autonomy as the dependent variable. Next, I will redo the multiple regression analysis by adopting the World Bank data, and consequently the Transparency International data for host country corruption. Second, I run a logit regression analysis with the corruption variable from all three databases as the independent variables, and subsidiary initiative as the dependent variable. Lastly, a logit regression has been performed to exhibit the influence of subsidiary autonomy on subsidiary initiative, while accounting for the corruption level. Results will point out whether there is an opportunity for generalizable conclusions on the relationship between host country corruption and subsidiary
autonomy/initiative. Further substantiation of the robustness checks can be found under
‘robustness checks’ in the ‘findings’ section. The comparative approach adheres to the reliability of this research.
3.2.1 Dependent variables Subsidiary autonomy
In equation 1, subsidiary autonomy is measured by the percentage direct control of the global ultimate owner. I hold the assumption that a lower direct ownership percentage by the global ultimate owner over the subsidiary indicates that the subsidiary is more autonomous, since the global ultimate owner is more distant from its subsidiary in terms of involvement. The firm-level data for subsidiary autonomy has been derived from the Orbis database.
29 subsidiary autonomy is approached as a factor in a broader network of elements that contributes to the evolution of the MNC (Johnston & Menguc, 2007).
Subsidiary initiative
In equation 2, subsidiary initiative is measured by the number of patents held by the subsidiary. This choice is justified by product innovations, new technologies, specialized resources in innovation, and entrepreneurial actions which are all marked as indications of subsidiary
initiatives (Ambos et al., 2010; Birkinshaw, 1997; Dörrenbächer & Geppert, 2009; Hamprecht & Schwarzkopf, 2014). Subsidiary initiative has for instance been measured by investments in R&D (Ambos et al., 2010) or product innovations (Birkinshaw, 1997). In line with intuitive reasoning, such proxies fit the measure of number of patents. Firm-level data has been derived from the Orbis database.
3.2.2 Main independent variable Host country corruption
The BEEPS V and MENA ES database has been consulted for aggregated firm-level host
country corruption scores. Six aggregated firm-level host country corruption measures have been included, which were reduced to two factors after running a factor analysis.The host country corruption factor variables are continuous variables with a value ranging from approximately –2 to 2.
Second, the World Bank provided country-level corruption scores. The corruption scores are based on the level of ‘control of corruption’. The control of corruption estimate captures “perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as "capture" of the state by elites and private interests” (World Bank Group, 2019b). Control of corruption is a continuous variable, with values ranging from –2.5 to 2-5. -2.5 implies that control of corruption is lowest whereas a value of 2.5 indicates best control of corruption.
30 the perceived level of corruption in the country’s public sectors (Transparency International, 2018). Values have been assigned that range from 0 to 100, where a lower value reflects a poorer perception on a country’s control of corruption.
3.2.3 Control variables
Corporate-, subsidiary-specific-, and environmental factors all affect subsidiary initiative (Dörrenbächer, & Geppert, 2009), whereby the corporate context is based on the MNC's home country. I include both firm level- and country level control variables.
Firm level variables - Subsidiary
First, I control for subsidiary age. An increasing level of experience in the business and the local environment will enable for more initiative-taking opportunities (Ambos, Andersson, &
Birkinshaw, J. (2010). Subsidiary age, measured in years since establishment, is assumed to relate positively to the level of subsidiary autonomy (Ambos, Asakawa, & Ambos, 2011). Besides, older subsidiaries tend to be more innovative as a consequence of increased levels of autonomy (Minbaeva, Pedersen, Björkman, Fey, & Park, 2003). Subsidiary age will be measured by the date of incorporation, for which data has been acquired from the Orbis database.
Moreover, subsidiary size is included as a control variable. Subsidiary size can have a direct impact on subsidiary initiative as smaller subsidiaries tend to have a stronger propensity towards innovation. Small subsidiaries face simpler decision-making processes and there are less internal transaction costs involved (Hamprecht & Schwarzkopf, 2014). Next, larger subsidiaries may rely less on knowledge from other units within the MNC, resulting in a higher level of subsidiary autonomy (Minbaeva et al., 2003). Subsidiary size is measured by the fact whether the subsidiary is listed, delisted, or unlisted and data has been acquired from the Orbis database.
31 Subsidiary leadership will be measured by the number of directors and managers. Data has been gathered from the Orbis database.
Fourth, I account for subsidiary performance. Resources are required to take initiatives naturally (Ambos et al., 2010), whereby well-performing subsidiaries can have eased access to such resources. Furthermore, subsidiaries that perform better are likely to be granted with more autonomy (Ambos et al., 2011). Subsidiary performance will be measured by the subsidiary's profit margin for which data has been acquired from the Orbis database.
The last subsidiary internal factor that I control for is skill level through average cost of
employee. I assume that the average cost of employee can indirectly proxy for R&D intensity of the subsidiary. This is built on the assumption that R&D intensive firms face higher average costs of employee. R&D intensive firms face higher costs to compensate for the skill level of its employees. Moreover, larger firms spend relatively more on R&D (Shefer & Frenkel, 2005), which is processed on to higher wages due to skilled employees in R&D. This results in higher average costs of employee. The level of subsidiary innovativeness affects the subsidiary's tendency to draw upon local knowledge resources (Frost, 2001). It this leads to increased
uncertainty and higher transaction costs, lower levels of entrepreneurial activity is a likely result (Anokhin & Schulze, 2009). The data has been derived from the Orbis database.
External variables
With respect to external factors that affect subsidiaries’ behavior, I first account for inter-subsidiary competition; a MNC’s internal competitive pressures that the inter-subsidiary is
experiencing. The number of intrafirm competition is a factor that influences the implementation of subsidiary initiative (Strutzenberger & Ambos, 2014). Within the transnational MNC,
32 been consulted for the data, whereby inter-subsidiary competition will be measured by the
number of companies in the corporate group.
Second, I control for the level of informal competition in the host country. Subsidiary activities can be constrained by the nature of the local environment in which it operates (Birkinshaw & Hood, 1998). Due to isomorphism, the normative and cognitive component guides subsidiary behavior in addition to regulative pressures (Hotho & Pedersen, 2012). In an environment with a severe presence of informal competition, subsidiary behavior is likely to be more strongly influenced by normative and cognitive pressures. This increases uncertainties as the regulative practices that the foreign subsidiary can rely on are poorly enforced which hampers initiative taking behavior. Furthermore, information about institutions is a necessity in order to reduce liability of foreignness. Information on national institutions is a form of tacit knowledge as it is probably encapsulated in social structures (Gammelgaard, 2012). Access to information may be more difficult in case of severe informal competition. Consequently, subsidiaries in
environments with eased access to information due to low informal competition may be less constrained to act proactively. Informal competition is measured by the degree to which local firms argue that informal sector competitors are an obstacle to their operations. Data has been obtained from the BEEPS V and MENA ES database.
Third, I control for Ease of doing business. Corrupt activities such as bribing can function as a solution to high transaction costs by reducing lengthy administrative procedures and ill
33 Next, I account for the level of host country political stability. Meager levels of political stability will discourage foreign subsidiaries to engage in risk-taking behavior. Poor political stability will result in higher levels of uncertainty which increases transaction costs (Chadee & Roxas, 2013). The measures for political stability have been derived from the World Bank WGI dataset, with a value ranging from approximately –3 to approximately 2.
Fifth, I control for the national innovation level. As this research adopts the institutionalism perspective as the main theory for evaluative purposes, I control for host country innovation levels due to the concept of isomorphism. Firms have a higher propensity towards innovation if they operate in a more innovative environment, given the assumption that distinct firms
operating in a similar environment resemble to other units that face the same set of
environmental conditions (DiMaggio & Powell, 1983). The national innovation level will be measured by the innovation efficiency ratio, derived from the Global Innovation Index.
Lastly, the models include a set of host-country and subsidiary industry fixed effects to account for unobserved factors associated with both the host country and the industry in which the subsidiary is operating. Subsidiary industry is measured by the NACE industry classification code. Data on the subsidiary industry and the host country that the subsidiary is located in has been acquired from Orbis. A graphical overview of all control variables matched with the broader component of interest can be found in Appendix C, figure 3.
4. FINDINGS
4.1 Descriptive Statistics
Table 1 depicted below demonstrates descriptive statistics on the variables included for analysis. The variables subsidiary size, subsidiary performance, and average cost of employee have the strongest impact on the number of observations included for analysis, due to a relatively high number of missing values. A value of 0 has been assigned to 5,166 subsidiaries whereby the global ultimate owner has a direct ownership percentage of 100 percent over the foreign
34 not have complete ownership, which indicates the assumption of higher levels of subsidiary autonomy. The dummy variable, subsidiary initiative, represents 684 subsidiaries that possess at least 1 patent and 10,139 subsidiaries that do not possess any patents. The number of subsidiaries per country are displayed in appendix B, table 9. The majority (approximately 60 percent) of all the 10,823 subsidiaries included for analysis are based in Turkey or Ukraine. The remainder of the sample subsidiaries are dispersed over the other 16 developing countries. After checking for normality, log variables have been created for subsidiary age, inter-subsidiary competition, and subsidiary autonomy. This is further enumerated in the ‘Orbis data cleaning process’ section in Appendix A. Lastly, all variables have been checked for outliers. As severe outliers were detected for average cost of employee, inter-subsidiary competition, and informal competition, the observations with extreme values have been replaced. A further description on checking for outliers can be found in Appendix A.
Table 1 Descriptive statistics
Variable Obs Mean Std.Dev. Min Max
Subsidiary autonomy 7849 .966 1.502 0 4.615 Subsidiary initiative 10823 .063 .243 0 1 Corruption BEEPS 10823 0 1 -2.4 2.516 Control corruption WB 10823 -.158 .598 -.98 1.294 Control corruption TI 10823 42.684 11.682 27 70 Subsidiary age 8100 2003.063 11.536 1771 2019 Subsidiary size 2468 2.527 1.995 0 9.118 Subsidiary leadership 10823 1.518 1.364 0 50 Subsidiary performance 3724 5.842 20.967 -98.337 100
Average cost of employee 1902 15.103 10.66 .013 116.172 Inter-subsidiary competition 10802 1.079 1.094 0 7.396
Ease of doing business 10823 67.941 4.77 58.99 80.54
Political stability 10823 -.836 1.149 -1.962 .979
Informal competition 10805 .906 .236 .314 1.275
35 4.2 Correlation analysis
In table 2 below, the correlation matrix has been displayed. Most interesting is the correlation between the corruption measurements. As the factor variables are based on the six corruption measurement variables from the BEEPS V and MENA ES database, it is interesting to see how these corruption measurement variables correlate with each other, as well as with the control of corruption variables from the World Bank and Transparency International.
37 (15) Innovation
efficiency host
0.049 -0.008 0.531 -0.534 -0.514 -0.016 0.045 0.004 0.012 0.187 -0.141 -0.537 -0.461 0.135 1.000
Table 3 Correlation matrix corruption measures
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Corruption government officials 1.000
(2) Corruption local officials 0.941 1.000
(3) Corruption parliamentarians 0.990 0.936 1.000
(4) Costs corruption 0.051 0.309 0.041 1.000
(5) Unofficial payments taxes 0.664 0.865 0.655 0.592 1.000
(6) Informal payments construction -0.545 -0.782 -0.533 -0.703 -0.955 1.000
(7) Corruption BEEPS 0.985 0.911 0.985 -0.088 0.620 -0.483 1.000
(8) Control Corruption WB -0.804 -0.897 -0.785 -0.471 -0.898 0.844 -0.746 1.000
4.3 Multicollinearity and heteroscedasticity
I tested for multicollinearity to analyze to what extent a corruption measurement variable from the BEEPS V and MENA ES database can be linearly predicted by another corruption measure with a considerable degree of accuracy. The results indicate that five out of the six corruption measurement have a variance inflation factor (vif) above 10. In consequence of assumed multicollinearity, a factor analysis has been conducted.
I tested for multicollinearity after all regression analyses for the influence of host country corruption on subsidiary autonomy and subsidiary initiative. Furthermore, I tested for
multicollinearity for the influence of all three corruption measures derived from three distinct databases. The results indicated that in all cases the vif was at least 1.007, whereas the maximum was 5.236. This is significantly lower than the threshold of 10, which confirms that
multicollinearity is not an issue.
Whether it is fair to assume constant variance, I account for heteroscedasticity. To verify for the presence and effect of potential heteroscedasticity, I conducted a robust standard errors model for the influence of host country corruption on subsidiary autonomy. However, no significant
deviation compared to the results of the default standard errors model was identified.
4.4 Factor analysis
The host country corruption variable derived from the BEEPS V and MENA ES database is based on six corruption measurements as displayed in table 4. To establish a unified measure for the host country corruption variable from the BEEPS V and MENA ES database, I have
conducted a factor analysis. The purpose of the factor analysis is to establish an
39 Table 4 Corruption BEEPS factor variable
Corruption measure Factor Variable name
(1) Corruption government officials Factor 1 Host country
corruption BEEPS
(2) Corruption local officials Factor 1
(3) Corruption parliamentarians Factor 1
(4) Costs corruption Factor 2 Host country
corruption BEEPS2
(5) Unofficial payments taxes Factor 2
(6) Informal payments construction Factor 2
Note A description of the corruption measures can be found in Appendix D, table 10.
4.5 Hypotheses testing
In table 5, I depicted the regression results for the influence of host country corruption on subsidiary autonomy by adopting the host country corruption measure from the BEEPS V and MENA ES database. Model 1 shows the impact of subsidiary internal control variables on subsidiary autonomy. Model 2 includes all subsidiary external control variables. Model 3 considers the impact of all control variables on subsidiary autonomy, including the fixed effects of subsidiary industry and host country. All control variables have significant effects except for subsidiary leadership and ease of doing business. Lastly, model 4 adds the host country
corruption variable. The model has an adjusted R2 of approximately 0.073. This points out that the model explains approximately 7.3% of the total variation. The influence of host country corruption in model 4 has a coefficient of 0.107 (p<.1). Hence, the null hypothesis can be
rejected as the results confirm hypothesis 1. A one-unit change in host country corruption results in a change for subsidiary autonomy by 100 ∙ (𝑒β𝐶𝑜𝑟𝑟 – 1)%. If there is a 1-unit increase in host country corruption, it multiplies the expected value of subsidiary autonomy by 𝑒 𝛽. e is a
mathematical constant for the natural logarithm with a value of approximately 2.718. A one-unit increase in host country corruption will affect the level of subsidiary autonomy by 100 ∙
40 Table 5 Multiple regression results host country corruption and subsidiary autonomy
(1) (2) (3) (4)
VARIABLES Model 1 Model 2 Model 3 Model 4
Host country corruption_BEEPS 0.107*
(0.0625) Subsidiary age -0.00516* -0.00327 -0.00533* -0.00533* (0.00282) (0.00284) (0.00286) (0.00286) Subsidiary_size 0.0699*** 0.0721*** 0.0533** 0.0533** (0.0213) (0.0258) (0.0269) (0.0269) Subsidiary leadership 0.0346 0.0469* 0.0462 0.0462 (0.0247) (0.0282) (0.0286) (0.0286) Subsidiary performance -0.00723*** -0.00734*** -0.00608*** -0.00608*** (0.00194) (0.00194) (0.00193) (0.00193)
Average cost of employee 0.00113 0.00892** 0.00909** 0.00909**
(0.00352) (0.00392) (0.00397) (0.00397)
Inter-subsidiary competition -0.0685** -0.0595* -0.0595*
(0.0346) (0.0347) (0.0347)
Ease of doing business 0.0441*** -0.00348 -0.0399
(0.0162) (0.0293) (0.0252)
Political stability -0.392*** -0.285* -0.0697
(0.104) (0.160) (0.124)
Informal competition 0.220 1.084*** 0.875***
(0.181) (0.329) (0.282)
Local innovation efficiency -2.118*** -6.504*** -4.569***
(0.713) (1.810) (1.162) Industry subsidiary - - Host country - - Constant 10.94* 5.779 17.31*** 18.37*** (5.667) (5.774) (6.067) (6.127) Observations 1,506 1,493 1,493 1,493 R-squared 0.031 0.045 0.087 0.087 Adjusted R-squared 0.028 0.038 0.073 0.073
41 Table 6 demonstrates the logit regression results for the influence of host country corruption on subsidiary initiative, whereby the host country corruption variable is derived from the BEEPS V and MENA ES database. The subsidiary specific firm-level control variables have been included in isolation in model 1. Model 2 reflects the influence of subsidiary external control variables on subsidiary initiative. All controls variables are considered in model 3. Subsidiary leadership, subsidiary performance, average cost of employee, informal competition, and local innovation efficiency do not show significant results. Model 4 adds the host country corruption variable to the analysis. Model 5 presents the results for the influence of subsidiary autonomy on subsidiary initiative, while model 6 adds the moderating effect of host country corruption. This is reflected by the interaction term between subsidiary autonomy and host country corruption. Derived from model 4, host country corruption BEEPS has an odds ratio of 0.505 (p<.05). This implies that a one-unit increase in host country corruption leads to a 0.505-fold increase of subsidiary
Table 6 Logit regression results host country corruption and subsidiary initiative
(1) (2) (3) (4) (5) (6)
VARIABLES Odds ratio Odds ratio Odds ratio Odds ratio Odds ratio Odds ratio
Subsidiary initiative
Host country corruption_BEEPS 0.505** 0.449**
(0.160) (0.162) Subsidiary age 0.988** 0.984*** 0.988** 0.988** 0.990* 0.990* (0.00480) (0.00590) (0.00566) (0.00566) (0.00583) (0.00585) Subsidiary_size 1.641*** 1.787*** 1.725*** 1.725*** 1.729*** 1.728*** (0.0992) (0.140) (0.141) (0.141) (0.152) (0.152) Subsidiary leadership 1.009 1.046 1.086 1.086 1.092 1.092 (0.0317) (0.0567) (0.0629) (0.0629) (0.0690) (0.0689) Subsidiary performance 1.003 1.001 1.001 1.001 0.996 0.996 (0.00748) (0.00806) (0.00810) (0.00810) (0.00832) (0.00832)
Average cost of employee 1.025*** 1.013 1.007 1.007 1.006 1.006
(0.00906) (0.0122) (0.0134) (0.0134) (0.0140) (0.0140)
Inter-subsidiary competition 0.788*** 0.785*** 0.785*** 0.798** 0.799**
(0.0557) (0.0577) (0.0577) (0.0718) (0.0718)
Ease of doing business 1.105** 0.821 1.853*** 0.738 1.956***
(0.0522) (0.197) (0.420) (0.192) (0.470)
Political stability 2.108* 4.289* 0.352 5.372* 0.270
(0.851) (3.479) (0.298) (4.648) (0.240)
Informal competition 1.346 3.681 0.00466** 7.394 0.00249**
(0.659) (8.344) (0.0112) (17.66) (0.00630)
Local innovation efficiency 2.374 349.3 6.217e+06** 118.4 1.527e+07**
43 Industry subsidiary - - - - Host country - - - - Subsidiary autonomy 1.077 1.065 (0.0784) (0.111) c.Subsidiary_autonomy#c.Corruption_ BEEPS 0.984 (0.0998)
Constant 1.219e+08* 1.085e+08 4.052e+11 0 1.638e+13 0*
(1.185e+09) (1.341e+09) (7.957e+12) (0) (3.386e+14) (0)
Observations 1,845 1,828 1,735 1,735 1,416 1,416
4.6 Robustness checks
To accredit to the empirical contribution of this research, I rely on three distinct databases for data on host country corruption. Three separate corruption variables have been derived from these three databases. The enumerated regression results of table 11, Appendix E, demonstrate consistency for the influence of host country corruption on subsidiary autonomy. Model 1 includes all control variables. Model 2 includes the control host country corruption WB variable, whereas model 3 adds the control host country corruption TI variable instead. Model 2 and 3 have an adjusted R2 of 0.073, which demonstrates that approximately 7.3 percent of the variation of the dependent variable is explained by the independent variables. The host country corruption WB variable in model 2 shows a coefficient of -0.512 (p<.1). This confirms the initial results inferred from table 5. Similar results are exhibited in model 3. The coefficient of host country corruption TI is -0.034 (p<.1). Hence, the two alternative measures for host country corruption support the results exhibited in table 5. When there is a 1-unit increase in control host country corruption WB or control host country corruption TI, the expected value of subsidiary initiative is multiplied by 𝑒 𝛽. A one-unit change in host country corruption results in a change for subsidiary autonomy by 100 ∙ (𝑒β𝐶𝑜𝑟𝑟 – 1)%. More specifically, if control of host country corruption (WB) increases by 1 (less corruption), subsidiary autonomy changes by 100 ∙ (2.718−0.512 – 1)%. This implies a decrease of approximately 40.01 percent in the level of subsidiary autonomy if control host country corruption increases by 1. When relying on
Transparency International for control of host country corruption, subsidiary autonomy decreases by approximately 3.34 percent for a one-unit increase of control host country corruption. Given
the fact that the World Bank and Transparency International corruption measures relate to the control of corruption, the negative Beta coefficient indicates a positive relationship between host country corruption and subsidiary autonomy. If there is more corruption (less control of
corruption), subsidiary autonomy increases.
