The role of autonomy in subsidiary initiatives and development CavanaghAndrew 2017 The concept of subsidiary autonomy has featured prominently in recent subsidiary management research (Collinson & Wang, 2012; Gammelgaard, McDonald, Stephan, Tuselmann & Dorrenbacher, 2012; Gammelgaard, McDonald, Tuselmann, Dorrenbacher & Stephan, 2011; Pisoni, Onetti & Fratocchi, 2010; Tong, Wong & Kwok, 2012). Yet despite this, the notion is not yet fully understood. For example, studies by Miozzo and Yamin (2012), Balogun, Jarzabkowski and Vaara (2011), Gammelgaard et al. (2011), Sandvik (2010), Sargent and Matthews (2006) and Delany (2000) have claimed that instead of a subsidiary’s autonomy being derived entirely through assignment from the head office, it may be developed beyond the assigned levels by the subsidiary itself. However, none of these previous studies have explicitly differentiated between different types of autonomy. Additionally, Gammelgaard et al. (2012), Pisoni et al. (2010) and Young and Tavares (2004) have noted that further research may be needed to redefine subsidiary autonomy in relation to the different value functions, such as marketing and finance. This is because recent studies such as those of Collinson and Wang (2012) and Gammelgaard et al. (2011) have proposed that a subsidiary’s autonomy may vary across these value functions. This is in contrast with the traditional assumption held by the majority of previous studies (such as those of Couto, Vieira and Borges-Tiago [2005] and Takeuchi, Shay and Li [2008]) that a subsidiary’s level of autonomy would be consistent across all value functions. The relationships between subsidiary autonomy and related concepts such as authority (Chiao & Ying, in press; Gammelgaard et al., 2011; Gupta & Govindarajan, 1991; Mudambi, 1999; Young & Tavares, 2004) have also yielded conflicting results, and require further investigation. Tong et al. (2012), Young and Tavares (2004) and Brock and Paterson (2002) noted that in order to address these gaps in the extant literature, research must first establish an understanding on how the nature and extent of autonomy (including its different types) are perceived and negotiated by the subsidiary and the parent company. Gammelgaard et al. (2012), Gammelgaard et al. (2011) and Young and Tavares (2004) have argued that more research is particularly needed to investigate the role that the different types of autonomy play in influencing the process of subsidiary development (Balogun et al., 2011; Birkinshaw & Hood, 1998; Cavanagh & Freeman, 2012; Delany, 1998, 2000). Such an investigation must also consider the role of the different types of autonomy in the development of subsidiary initiatives, which are those actions by the subsidiary management that drive the subsidiary development process (Balogun et al., 2011; Birkinshaw, Hood & Jonsson, 1998; Cavanagh & Freeman, 2012). With this in mind, the current study’s broad research problem was ‘What role do the different types of autonomy play in subsidiary initiatives and role development?’ In order to address this research problem, the study drew upon agency theory (Jensen & Meckling, 1976; Saam, 2007), the network model of the MNE (Andersson & Forsgren, 1996; Ghoshal & Bartlett, 1990) and the decision process perspective (Bower, 1970; Burgelman, 1983a). In doing so, the study was able to answer the following research questions which emerged from the literature review: Research Question 1a: How do subsidiaries and headquarters understand autonomy? Research Question 1b: How do subsidiaries and headquarters negotiate autonomy? Research Question 2a: What role do the different types of autonomy play in the development of subsidiary initiatives? Research Question 2b: How do the different types of autonomy influence subsidiary development? The study employed a qualitative research design through the use of multiple case studies. Qualitative data was found by Singh (1981) and Pisoni et al. (2010) to be necessary in order to fully explore the complexities of subsidiary autonomy in detail. A case study was selected as the most appropriate research strategy for this current study as it investigated a contemporary issue, and involved ‘how’ research questions (Yin, 2009). The current study employed a multiple case study design based on 10 cases so as to gain more powerful inferences than could be obtained from a single case (Yin, 2009). The cases were selected using judgement (purposive) sampling, snowball sampling and opportunistic sampling techniques (Patton, 1990; Tharenou, Donohue & Cooper, 2007). Of the 10 cases, 6 involved interviews with the head office, 9 involved interviews with the subsidiary, and 5 incorporated the views of both the head office and the subsidiary. This enabled the study to address the research gap identified by Brock and Paterson (2002), Tong et al. (2012) and Young and Tavares (2004) through the development of an in-depth, multiple-perspective understanding of subsidiary autonomy. The data for the study was collected through semi-structured, in-depth interviews with senior managers at both the subsidiary and head office level within the selected cases. This data was supplemented by the addition of documentary evidence that was collected from newspaper articles, industry journals and company websites. Once the data had been collected through the interviews and documentation, the study used a template approach to data analysis (Crabtree & Miller, 1992). This approach involved four steps: ‘immersion in the data’, ‘coding’, ‘creating categories’ and ‘identification of themes’ (Green et al., 2007). The findings of the current study make a number of key theoretical contributions to the field. First, the findings for Research Question 1a led to the development of a revised, comprehensive definition for subsidiary autonomy. This extended the key previous definitions of Ambos, Andersson & Birkinshaw. (2010), Brooke (1984), Chiao and Ying (in press), Gammelgaard et al. (2012), Tong et al. (2012) and Young and Tavares (2004), and refined those of Najafi-Tavani et al. (2012) and O’Donnell (2000) by differentiating between autonomy and authority. The findings for Research Question 1a also extended the network model of the MNE (Andersson & Forsgren, 1996; Ghoshal & Bartlett, 1990) in two key ways. First, the findings recognised the need for the model’s application at an interdepartmental, rather than organisational, level. Second, the findings identified that local environment factors will not only affect subsidiary behaviour, but will more specifically force the subsidiary to employ or develop autonomy. The findings for Research Question 1b then identified and applied terminology for two separate and distinct types of autonomy: assigned and assumed. Assigned autonomy was found to be delegated to the subsidiary by the head office, and falls within the subsidiary’s authority. Assumed autonomy, on the other hand, is developed by the subsidiary independently, and exceeds its authority. These findings thus addressed the research gap identified by Tong et al. (2012), Young and Tavares (2004) and Brock and Paterson (2002) in relation to the precise origins of a subsidiary’s autonomy, and the resultant types of autonomy. The findings for Research Question 1b also had a number of implications for agency theory (Arrow, 1985; Jensen & Meckling, 1976; Saam, 2007). First, they refuted the agency theory assumption of different risk preferences by revealing that subsidiaries are often less risk averse than the head office. Second, the findings refined agency theory's goal conflicts assumption to include three additional causes of agents acting against their principal’s direct instructions. These causes include: the subsidiary being forced to assume its own autonomy on account of indifference or lack of interest from the head office; the subsidiary assuming its own autonomy in order to maximise the efficiency and effectiveness of an operation; and the head office allowing or facilitating the extension of a subsidiary’s autonomy beyond its authority. Third, the findings refined agency theory by recognising the difference between ‘real’ and ‘stated’ goals of the principal, and in doing so challenged the assumption that an agent’s failure to follow its assigned charter will result in agency problems. The findings for Research Question 2 (a & b) then identified assumed autonomy as necessary to not only generate subsidiary initiatives, but also to lead to successful subsidiary development. In doing so, the findings addressed the research gap identified by Gammelgaard et al. (2012), Gammelgaard et al. (2011) and Young and Tavares (2004). In order to facilitate this process, the findings revealed that a low level of subsidiary-head office communication in the formative stages of initiative creation is essential to their longer-term viability. In addition, head office endorsement of these initiatives was found to be imperative if they are to successfully translate into subsidiary development. This in turn was found to require a strong subsidiary-head office relationship. The findings for Research Question 2 (a & b) also extended the network model of the MNE by specifically noting that a subsidiary may only develop its role to a position of equality or even superiority vis-à-vis the head office through the use of assumed autonomy. Lastly, the findings for Research Question 2 (a & b) had three significant implications for the decision process perspective (Bower, 1970; Burgelman, 1983a). First, the study extended the application of the perspective to the subsidiary development context. This led to the replacement of Burgelman’s (1983a) original terms ‘induced’ and ‘autonomous’ strategic behaviours with the more appropriate terms of ‘assigned autonomy’ and ‘assumed autonomy’ respectively. (...)