“The effect of institutional distance on corporate attention: a study of Western MNE subsidiaries in advanced and emerging markets.”
By Tjeerd Jeroense
August 25th 2016
University of Groningen Faculty of Economics and Business
Masters’ Thesis IB&M Student number: 2784173 Supervisor: Dr. Rian Drogendijk
Abstract
With the changes in China’s institutional environment the question arises whether institutions affect the organizational dynamics of actors in that environment such as multinational
enterprises (MNEs). This research specifically investigated whether institutional distance affects the degree of corporate attention that a subsidiary receives from its’ headquarters. By using letters to shareholders from annual reports to measure headquarters attention, a research design emerged that allowed for efficient and effective data collection from readily available secondary databases. Binominal logistic regression provided evidence upon which the main hypothesis was partially supported. Knowledge distance and geographic distance affects the degree of received corporate attention, but these results appear to be inconsistent over time.
Table of Contents
1. Introduction and Central Research Question ... 2
2. Literature review ... 5
Institutional distance ... 5
Headquarters’ (corporate) attention ... 8
Impact Institutional Distance on headquarters’ attention ... 11
Conceptual model ... 14
3. Methodology & Data ... 15
Research structure ... 15
Sample ... 15
Data collection for quantitative research ... 17
The dependent variable ... 17
Independent variables ... 22 Controls ... 25 Data analysis ... 26 4. Results ... 27 Assumptions (model) ... 27 Regression results ... 29
5. Discussion & Conclusion ... 34
Models 2009 ... 34
Models 2014 ... 37
Comparing the years ... 38
Limitations ... 39
Recommendations for future research ... 41
Conclusion ... 42
REFERENCES ... 43
1. Introduction and Central Research Question
Attention is, according to the attention based view, the noticing, encoding,
interpreting and focussing of time and effort by organizational decision-makers on both issues and answers (Ocasio, 2011).
Organizational attention is the socially structured pattern of attention by decision makers within an organisation. It is essential to know how firms distribute and regulate the attention of their decision-makers in order to understand the firm’s behaviour. In turn, it is the comprehension of how firms behave that leads to understanding whether and when firms can adapt to a changing environment, adapt strategies and capabilities accordingly or fail to react on their competition (Ocasio, 1997). Thus, the concept of attention is important for a firm operating in a world of change.
This is a matter that firms of all sizes can relate to. The multinational enterprise (MNE) for instance, is a large firm; a collection of geographically and culturally dispersed subsidiary units that each controls a specific stock of resources. They are legally owned by and report to the corporate headquarters. Moreover, this network has a complex portfolio of markets, functions and businesses (Ambos et al., 2010, Bouquet & Birkinshaw, 2008).
These characteristics of the MNE entail that the management at the
headquarters has trouble in keeping its attention on the total set of issues, and it has to divide or prioritize its attention thereby choosing one issue over the other (Bouquet & Birkinshaw, 2008; Ocasio, 1997). The amount of attention each subsidiary gets depends on its weight & voice within the corporate system (Bouquet & Birkinshaw, 2008). Because headquarters cannot control everything, each subsidiary acts partly on its own behalf; each subsidiary is given a specific role (weight) in the MNE network and with it the subsidiary takes initiative (Bouquet & Birkinshaw, 2008).
Meanwhile it is expected from the subsidiary that this decision-making freedom is utilized to prolong its survival in the host environment, develop and that these actions are in line with MNE corporate strategy (Strutzenberger & Ambos, 2014).
required to manage the differences between their home and foreign institutions. The institutional distance between the home country and the subsidiary country influences the potential voice that a subsidiary possesses, and together with the strategic
importance of its host market they determine the weight of the subsidiary in the corporate network (Bouquet & Birkinshaw, 2008).
Not only do institutions vary from each other, they change over time as well. China, an emerging market with a vastly growing middle class, is reforming to western society institutions and their society is getting more individualistic (Delios & Xufei, 2010). For a Dutch subsidiary stationed in China and being a sole manufacturer of medium-tech consumer goods initially, this could result in a growing local market to sell its merchandise. Moreover, Ambos & Birkinshaw (2010) mentioned that China and India in particular are receiving more attention in general. Good performance for the subsidiary means outcompeting host firms, and that is only realised when
overcoming the costs of doing business abroad or liability of foreignness (LOF). LOF partly originates from the dissimilarities between institutions, a factor external to the MNEs firm boundaries (Zaheer, 1995). Outperforming the competition is key for the subsidiary to ensure its survival and show good performance and it partially builds on the resources from the corporate environment, thus they all fight and strive for
corporate attention (Bouquet & Birkinshaw, 2008).
institutional distance and corporate attention, therefore the following research question arises.
Does institutional distance have a positive or negative effect on the degree of corporate attention that majority owned subsidiaries receive from their headquarters in West-European MNEs active in emerging markets?
The central research question will be answered with the help of answers to the following research questions. (1). What is institutional distance and why is it
important for MNEs? (2). What is corporate attention, and what is its link with MNEs? (3) How does institutional distance influence corporate attention? (4) Does this influence change over time?
2. Literature review
Institutional distance
Institutions. Institutions are ‘the rules of the game’ and constrains created by humans that shape human interaction (Moore et al., 2015; North, 1992). They create stable structures that organize economic transactions and reduce transaction costs in uncertain and complex environments (Ando & Paik, 2013; North, 1992).
According to the NIE or New Institutional Economics perspective (North, 1992) institutions solve the problem of fostering transaction. Institutions differ from each other in that they each have a different composition along a cognitive, normative and regulatory dimension (Scott, 1995). Cognitive rules refer to societies’ information collection, processing and interpretation. These cognitive programs; frames, schemas and representations create the cognitive structures and the common social knowledge (Scott, 1995). Illustrating this with the concept of corruption, cognitive rules in a country define how unity and identity is formed among the country’s inhabitants and how this creates shared believes on corruption within the country. Normative rules determine the things that should be done in a country. They are embedded in the social norms and believes and in the assumptions of human nature and human behaviour of the countries’ inhabitants. Referring back to the corruption example again; normative rules determine whether corruption is frowned up on by society or accepted. Finally, regulative rules state what can be done in the country; these are the existing laws and rules. An example is the Foreign Corruption Practices Act in the United States (Scott, 1995) or an explicit country law prohibiting corruption.
Every country has its own set of cognitive, normative and regulative rules. The degree of dissimilarity among countries in this respect shapes the so-called Country institutional Profile (CIP) of each country. Institutional distance is defined as the difference between the CIPs of the home country and the host country (Kostova, 1999).
as well as the mode of entry into the new market (Kogut & Singh, 1988), and can form a resistance for implementing strategies abroad. This is seen in their study by the decrease in amount of investment activity by venture capitalists in an institutionally distant host country (Moore et al., 2015).
In her study, Zaheer (1995) explains why more institutional distance could results in more costs for the MNE. She finds evidence for a positive relationship between institutional distance and liability of foreignness (LOF). LOF is the particular cost of doing business abroad. These costs can arise from spatial distance,
unfamiliarity with the local environment, host country legitimacy costs and home country restrictions. They pose a disadvantage for the subsidiary in terms of competition with the local firms, because only non-local firms incur LOF costs. However, there are different ways in which MNEs try to overcome these costs.
Zaheer (1995) mentions the institutional view and the resource-based view; two different perspectives that could explain the strategies and reactions of
subsidiaries in their quest to overcome the liability of foreignness in their host environment.
The Institutional theory states that organizational practises differ among countries. MNEs subsidiaries will try to imitate the local practices, because the local firms are experienced actors in their institution. The local firms probably know the best way of manoeuvring through this, for exogenous firms’ complex and costly host environment (Zaheer, 1995). Furthermore, the new subsidiary is not the only one being oblivious to the environment, it also works vice versa. Little is known about the subsidiary by the host-country environment upon entry, and therefore stereotyping and misjudging MNEs by the host environment is evident. More effort and costs are required from the MNE to obtain legitimacy. One possible strategy for subsidiaries to increase legitimacy is to partake in CSR activities to show good citizenship
(Campbell et al., 2012; Zaheer, 1995). The desired effect of having more legitimacy is to minimize costs and thereby improving their performance. It is argued however that there is a limit to the level of institutional isomorphism that a MNE can reach, and that conforming to all normative, cognitive and regulative rules and values from multiple and conflicting sources is not at all possible (Kostova, Roth, & Dacin, 2012).
minimizing the institutional distance by means of mimicking others, they rather counteract the costs with more strategic resources from the MNE network, which other players in the market obviously will not have, these firm specific advantages or FSAs will thereby help them differentiating from the others. Subsidiaries capitalize on the fact that they are part of the MNE. Building upon earlier work Rugman &
Verbeke, (2001) argue that in the broadest sense the economic success of MNEs can be explained because they are able to create a non-location bound FSA, non-location in this case meaning that they can be easily transferred cross border. Non-location bound FSAs appear in two forms; it is either specific functional knowhow, or the capability to efficiently organise and disperse assets (Dunning, 1988; Dunning & Rugman, 1985; Rugman, 1981). In their study they develop a detailed framework of patterns for development and dispersion of FSAs within a MNE. Referring back to the issue of overcoming LOF with strategic MNE resources, it becomes evident that in their quest for support, subsidiaries can count on sources from headquarters, as well as sister subsidiaries in the MNE network. The desired outcome of these efforts is again, improved performance (Zaheer, 1995).
The challenge however for foreign firms is facing the difficulty in transferring organisational practices and knowledge (Kostova, 1999). When institutional distance is great, the institutions may conflict each other, making it difficult for a firm to transfer organisational practices (Zhang et al., 2014).
Headquarters’ (corporate) attention
Headquarters’ attention or corporate attention is the outcome of a process whereby headquarters in MNEs respond to stimuli in their environment. Ocasio (1997, 2011) focuses on the core concept of attention and uses this to explain how decision makers structure their attention in organizations. He posits that using a so-called attention based view could explain organizational adaptation. In this respect, the concept of attention is defined as: the noticing, encoding, interpreting and
focussing of time and effort by decision makers regarding issues and answers. ‘Issues’ are the problems, opportunities and threats seen by the decision makers. ‘Answers’ is the collection of action alternatives that decision makers undertake (Ocasio, 1997: 189). Firms are seen as systems that self regulate and distribute the attention of its decision makers by three situation- and context specific forces. First, the Focus of attention. On the individual level, decision makers decide by themselves, which issues and answers they attend to at a certain point in time, and those decisions determine the actions that they undertake eventually. Second is the level of situated attention. It concerns the influence of surroundings, the context where the decision makers find themselves in. For instance, another study found that inherent herd mentality
predicted that headquarters would follow the direction of the group. This means that headquarters will be influenced by what competitors do in the same region (Bouquet & Birkinshaw, 2009). Lastly, these two levels are overall determined by the structural distribution of attention in the firm. The focus of attention and situated attention are controlled by the allocation of available procedures and communication within the firm (Ocasio, 1997).
Bouquet & Birkinshaw (2008, 2009) link attention to MNEs and define headquarters’ attention as “The extent to which the parent company recognizes and gives credit to the subsidiary for its contribution to the MNC as a whole” (Bouquet & Birkinshaw, 2008:579). Headquarters’ attention can be seen as the positive result of a relationship between the headquarters and the subsidiary in which the headquarters seeks to identify and build on new ideas (Ambos & Birkinshaw, 2010), and is generally focussed on the most productive and efficient strategies in the MNE (Bouquet & Birkinshaw, 2009).
accordance with corporate goals, and by using resources made available to them (Ambos & Birkinshaw, 2010). In a perfect world, a subsidiary could utilize all the valuable corporate strategic resources from the headquarters and use them when it sees fit, to outcompete everyone and accomplish highest possible performance. However, unfortunately for the subsidiaries, the corporate strategic resources are scarce and not in total control of the subsidiaries. It is the task of the headquarters to divide its attention and supervision over its sub-units (Ambos & Birkinshaw, 2010; Ocasio, 1997; Plourde et al., 2014).
When dividing attention, the headquarters is subject to internal and external stimuli. Internally, it is influenced by the general reporting processes that exist within the firm and by individual lobbyists who persuade headquarters. External stimuli include: industry reports, the media and competitor intelligence (Birkinshaw et al., 2007).
Consequently, these strategic resources or headquarters’ attention is fought for by subsidiaries within the MNC through an internal market (Ambos & Birkinshaw, 2010). The subsidiaries can use two dimensions of their corporate relevancy to seek attention. These are the ‘weight’ and ‘voice’ of the subsidiary. Weight is the degree of importance that the subsidiary’s local market is to the MNE, and the strategic role of the subsidiary within the MNE network. Voice is the amount of profile building and initiatives that the subsidiary undertakes. Initiative taking is here defined as the self-initiated penetration of new markets and product development (Ambos & Birkinshaw, 2010: 582), and profile building is seen as the internal lobbying for role upgrading. (Ambos & Birkinshaw, 2010: 583), The amount of voice and weight are not fixed either. By using more voice, the subsidiary can influence its weight on the long term as well (Birkinshaw et al., 2007; Bouquet & Birkinshaw, 2008) When obtained, corporate attention provides the subsidiary with the assets required to create
subsidiary specific advantages which increases performance, which ultimately could increase the subsidiary’s importance in the MNC network, so subsidiaries are
constantly striving for more attention.
subsidiary abroad, so the ability to capitalize on subsidiary information is key here and influences the amount of corporate attention that a subsidiary receives. A
subsidiary that is geared totally on the host-environment will develop information that could be unusable for the headquarters in other parts of the MNE network so this subsidiary will receive less attention (Ambos & Birkinshaw, 2010; Birkinshaw et al., 2007).
Types of attention. As mentioned earlier, the ability for the subsidiary to create subsidiary specific advantages also depends on the amount of corporate
attention that it receives from the headquarters. This attention can take various forms: the amount of decision-making autonomy, the bargaining power over other actors, and the possession of critical resources.
Positive corporate attention received a lot of attention in the literature. An extensive researched multi dimensional construct was developed by Bouquet & Birkinshaw (2008). They look at attention as a forward looking and possibly value-enhancing construct consisting of three levels. The first shows the level of support a subsidiary receives from the headquarters, supportive attention, where discretionary resources (technology, best practices and people) are distributed that help the subsidiary in its development. Second, relative attention is the perceived attention from headquarters by subsidiaries in comparison to its sister subsidiaries. Lastly, visible attention shows the explicit recognition from the headquarters shown in the media e.g. annual reports.
Impact Institutional Distance on headquarters’ attention
After investigating the different streams of literature on the topics of
(corporate) attention in the MNE network and institutional distance, the basis is set to explore for a potential link between the two.
During the discussion of attention it was established that ethnocentrism within the minds of top management at headquarters can steer the direction of corporate attention. Specifically, it enables the subsidiaries in the vicinity of the headquarters in terms of institutional distance to receive more attention than the ones who are further away from the headquarters (Birkinshaw, 1998). Perlmutter (1969) shows that headquarters with an ethnocentric mind set believe that host-country ways are superior to foreign manners. There is more on control on subsidiaries and less listening to their point of view. Attention in terms of rewarding and punishing performance is more focussed on headquarters then on subsidiaries, and there is a clear one-way direction of communication towards subsidiaries. Since the fight against LOF require a subsidiary and headquarters to have a strong cooperative relation (Zaheer, 1995), one can argue that one-way communication (ethnocentrism) inhibits the corporate attention that subsidiaries seek to receive in order to overcome LOF and strive in their host-environment. Identities (national) are part of the
cultural/cognitive dimension of institutional distance (Scott, 1995). The MNE network consists of semi-autonomous subsidiaries in various countries (Ambos & Birkinshaw, 2010; Ambos et al., 2010). These countries can differ a lot among each other in terms of identity, so therefor subsidiaries can suffer more or less from the degree of ethnocentrism at the headquarters. Since the variety of identities among countries determine partly the institutional distance between those countries, the link with institutional distance hereby is formed. The difference in identities has an influence on the degree of corporate attention that the subsidiary gets. More
institutional distance (more ethnocentrism), means less corporate attention received by the subsidiary.
understood by the headquarters (Bouquet & Birkinshaw, 2008). Although the
literature here focuses on geographic distance specifically, which is part of measuring institutional distance (Berry et al., 2010) it is reasonable to posit that these problems regarding information transfer and communication will also intensify when other cognitive, normative and regulative differences (institutional distance) increases among countries. For example, information can be interpreted differently (cognitive and normative), and specific rules (regulative) can form a barrier for transferring information. Therefore an increase in institutional distance is argued to decrease the degree of corporate attention that a subsidiary gets.
This suggests a negative relationship between institutional distance and corporate attention.
However, the attention dispersion within the MNE network is more complex; subsidiaries are not identical and this diversity plays part in the attention dispersion as well. Headquarters are more drawn to threats opposed than to opportunities (Plourde et al., 2014). Furthermore not all subsidiaries have an equal strategic importance in the MNE network. There is more focus from headquarters on subsidiaries with a substantial ‘weight’ in the MNE network (Bouquet & Birkinshaw, 2008). The weight determines the strategic importance of the subsidiary via their local market and/or strategic role (Bouquet & Birkinshaw, 2008). Important subsidiaries that encounter challenging market conditions or have difficulties in achieving their mission thus have a heightened priority for the headquarters, since they create more risk for the MNE (threats) (Plourde et al., 2014). A selection of subsidiaries that fit the above
mentioned profile are the ones that face LOF, and it is argued that more institutional distance between headquarters and subsidiaries results in more LOF felt by these subsidiaries (Zaheer, 1995). Therefor an increase in institutional distance between the subsidiary and the headquarters will increase the LOF felt by this subsidiary and will draw more corporate attention from the headquarters.
institutional distance between the headquarters and subsidiary then results in more ‘voice’ used by the subsidiary, ultimately increasing the propensity of more corporate attention.
Finally, it is expected that the headquarters understands distant subsidiaries poorly and therefore these subsidiaries receive less attention. However, there are strategic options from the headquarters to strengthen the communication between these subsidiaries and allow for more attention. The more uncertainty felt by headquarters regarding the host environment of the subsidiary, the more organizational control is needed by headquarters to make sure the subsidiary is operating along the MNE roadmap. This control is created through staffing strategy. Specifically, the control is realized by moving more home country nationals (expats) to the subsidiary (Ando & Paik, 2013). These expats have strong links with
headquarters through personal connections (Plourde et al., 2014). Since corporate attention is build from an individual, situational and structural level (Ocasio, 1997), it is suggested that personal connections between headquarters and expats will at least influence the attention on the individual and situational level and thus draws corporate attention towards the subsidiary where the expat resides. Moreover, these ties help the subsidiary channelling its’ communication used for the profile building mentioned earlier (Plourde et al., 2014; Bouquet & Birkinshaw, 2008). More institutional
distance increases the uncertainty felt by headquarters (Zaheer, 1995), this uncertainty increases the use of expats stationed at the subsidiary (Ando & Paik, 2013), and these expats draw more corporate attention from the headquarters (Plourde et al., 2014).
Evidently, there are multiple arguments for positive as well as negative influences of institutional distance that steer corporate attention globally.
After a thorough review of the existing body of literature on both the concepts of institutional distance and corporate attention, the following hypothesis is considered appropriate for testing in this research. Furthermore, a conceptual model was
developed (see Figure 1).
FIGURE 1. Conceptual model
Since it is expected that institutional distance can affect corporate attention both positively and negatively, it is hard to predict the overall sign that this association should hold. The literature review did not bring clarity as to an overall impact that institutional distance would have on corporate attention. Therefore due to this uncertainty, the overall sign of this relationship has been left out, and merely an association is hypothesized.
As mentioned in the introduction, this research is quantitative of nature. The goal of the research is looking for evidence supporting the main hypothesis.
3. Methodology & Data
The design starts with a brief explanation of the nature of the study, the chosen sample, the variables of the design including the review of existing measurements in scientific literature and concludes with the method of data analysis that will be carried out.
The literature established that corporate attention could appear in several forms being; supportive, relative and visible attention (Bouquet & Birkinshaw, 2008). The choice was made to use only readily available, secondary data. The annual report of Philips NV was investigated to explore options for measuring the three forms of corporate attention. In order to make good judgements the company Philips NV was seen as a suitable candidate. It is the largest multinational in consumer goods and one of the largest multinationals of the Netherlands, (Economywatch, 2013;
Topforeignstocks, 2013) and the largest patent gainer of the Netherlands in 2014 (IPO, 2015). It was found after studying the annual report that visible attention would be the only possible collectable form of corporate attention with corporate media, and therefore the focus thereon was on this form of attention.
Research structure
The research is quantitative of nature since its purpose is to find evidence for supporting the main hypothesis.
This entire research uses solely secondary data. The data is obtained by performing desktop research through readily available databases and by scanning media to shareholders. Information gathering from these sources is reliable as well as cost- and time effective.
Sample
markets and in particular India & China, therefore this research focuses on firms that all have subsidiaries in emerging countries.
Initially, the plan was to take one of the largest Dutch multinationals, and explore the corporate attention distribution to all of her subsidiaries. However in the face of lacking information using a variety of measures, the choice was made to use a set with multiple multinationals using one proxy for measuring. Therefore the loss of information due to the increased focus is counterbalanced with more firms.
Orbis, (Orbis, 2016) was used to make the selection of MNEs. It is a readily available database accessible through the Rijks University of Groningen, and covers a vast collection of company information. The final collection of MNEs was
systematically chosen on the basis of a few criteria. First, only Dutch listed firms in emerging markets (BRICS) were selected, because of earlier mentioned reasons. Second, firms with a minimum of 500 staff and with more than five subsidiaries were chosen, because of minimum fixed settings in the database. Third, the only
subsidiaries that were taken into account where the ones where the MNE had fifty or more per cent ownership, to capture most of the subsidiaries with decision making power regarding their ‘voice’ and usability of corporate attention. Fourth, the set was composed out of firms operating solely in the manufacturing industry. Subsidiaries abroad of manufacturing companies could also create product adaptations and
produce for their local market (Plourde et al., 2014), thus increasing the propensity of attention dispersion. This resulted in a sample of 23 Dutch listed multinationals with a total of 2599 subsidiaries.
Data collection for quantitative research
The quantitative research contains one dependent variable, four independent variables and three controls. They will be discussed below in this order.
The dependent variable
The focus here lies on visible attention, because visible attention, the explicit recognition of attention in corporate media, is the only type of attention that could be observed while scanning the annual report of Philips NV and it is predicted that this will be the case for many other annual reports from MNEs in the sample. Using annual reports as data sources, and letters to shareholders in particular, follows the studies of Ocasio (2011) and Plourde et al. (2014). The link between letters to shareholders and attention follows from the literature review in Ocasio (2011). Subsequently, Plourde et al. (2014) found a way to link subsidiaries with headquarters’ attention. They use a technique for measuring the visible positive
corporate attention dichotomously through data gathered from letters to shareholders. Their measurement is clear and not complex and provides a dependent variable that can be used for statistical inference. In the pilot study on the Philips NV annual report we found that many subsidiaries were not mentioned at all, just like Plourde and colleagues (2014) observed in their study with their observed MNE. Since this situation led them to choose this technique, in this similar situation in this thesis it is opted to use the same technique as well. The dependent variable, corporate attention, is measured by reading letters to shareholders from annual reports. Thereby, the visible attention granted to a subsidiary of each MNE in the letters to shareholders is considered to be evidence for granting corporate attention by the headquarters to the subsidiary following (Ocasio, 2011). If attention is given this is indicated with a ‘1’ and otherwise a ‘0’ is given, therefore when a subsidiary was given attention multiple times this still only counted as one.
The dichotomous dependent variable is a newly created dummy variable in this thesis and was formed in two steps. First, the letters to shareholders were scanned on attention topics. Second, the collected attention topics were allocated to MNE’s subsidiaries whenever this was appropriate.
(non-) subsidiary specific attention. Then the summary of each MNE for each year was compared with each other and a general summary was formed containing the most prominent attention topics for all the MNEs for each year (see Table 1 for the attention topics frequencies of each year).
TABLE 1. Prominent attention topics
Attention topic in letter to shareholders Frequency 2009 Frequency 2014
3rd party recognition for exceptional performance 2 6
Attention toward performance segments, and product (groups) 8 12
Attention performance/markets for geographies bigger than countries (world regions)
7 16
Closing subsidiaries 2 1
Comparing performance with industry averages/ competitors 2 5
Credit agencies S&P/ Moody’s 1 0
Crisis/ economic downturn 15 0
‘Emerging market’ or ‘high growth economies’ 2 1
Financial management reorganization 2 0
Greenfield investment 2 5
Important partnerships, customers or competitors (meso-environment) 6 7
Interest for macro factors 1 4
New product launches 5 2
New R&D/innovation investment 3 2
Recent Acquisitions, Joint ventures 5 9
Reference to stock exchange or regulations 0 5
Reorganization 11 3
‘Sourcing’ investments in emerging markets 2 0
Allocating attention. By interpreting the relevance of the attention topic for the MNE as acknowledged by the headquarters, the actual corporate attention for each subsidiary could be observed. The only clear identification of subsidiaries in the Orbis dataset was for location in countries and cities and very occasionally, subsidiary names. Therefore attention to subsidiaries was acknowledged when they were
mentioned by their name individually, by their country, city or by their world region. This resulted in four groups of attention topics that were allocated geographically on different levels to subsidiaries by headquarters; MNE or subsidiary activities, Meso-relations, Performance recognition and New, Acquisitions/Joint ventures & Green field investment. The coding for the attention allocation, based upon the attention topics and geographic areas, is displayed in Table 2 and Table 3 respectively.
The letter to shareholders of Philips NV in 2014 contains examples of each of the four attention topic groups. It was mentioned that the facility in Cleveland (US) was investigated because of US law infringement, and therefore received special corporate attention. Attention for Meso-relations was coded based on the special attention of partnerships by Philips NV in 2014 with specific hospitals all over the world. Acknowledgement of performance was found when Philips NV referred to the underperformance of its consumer lighting business unit in Europe. Lastly, regarding the final group, there was a reference found to a completed acquisition (Vulcano) and a refurbishment centre (greenfield) built in the Netherlands. Moreover, DSM NV mentioned attention towards recent acquisitions in North America (2009), and a new factory (greenfield) in the United States in (2014). For joint ventures specifically, NXP NV (2009) had a reference in its letter to a joint venture with Trident
case it is seen as positive corporate attention. However whenever a subsidiary ceases to exist or leaves MNE control, and this is mentioned in the letter to shareholders, then this is not seen as attention. This is highlighted with (B) in Table 2 and Table 3. In 2014 Neways N.V. closed two facilities, and the headquarters mentioned the cities involved in this closure. Although mentioned, this is not seen as attention that would support the subsidiary since the subsidiaries ceased to exist.
Also the merely stating of a subsidiary as being inventory to the MNE ((C) in Table 2 and Table 3), does not count as attention. Some headquarters had the
tendency to treat subsidiaries as inventory that had to be ‘checked off’ rather than getting explicit attention. This could be due to the function of corporate media as promotion for external stakeholders (investors) (Ambos & Birkinshaw, 2010). This was shown in many letters to shareholders throughout the dataset in various degrees. However, this was first recognized in a lesser degree while scanning TomTom N.V. in 2014, where they claimed to be largest telematics provider of Europe. This provided no clue regarding headquarters’ attention towards any of its European subsidiaries while it is in fact performance mentioned and a location mentioned.
TABLE 2. Coding table for dependent variable (attention topic)
Attention topic categories Attention granted (example) Attention not granted (1) MNE or subsidiary
activities
All subsidiaries in area mentioned related to activities from subsidiaries or MNE as a whole. (Philips suspended the production of a facility in Cleveland US due to law infringement by this subsidiary (Philips N.V., 2014))
A, B, C
(2) Meso-relations All subsidiaries in country of origin of specific customers, competitors and partners mentioned (Top management Philips acknowledges partnerships with hospitals (Philips N.V., 2014))
A
(3) Performance recognition All performance by subsidiaries recognized by the MNE in the area mentioned (Underperforming division of Philips in Europe (Philips N.V., 2014))
C
(4) New Acquisitions/Joint ventures & Greenfield investment
The newly acquired company (Vulcano for Philips N.V. (2014) or North America by DSM N.V. (2009)), newly created Joint venture (Trident for NXP N.V. (2009)), or greenfield investment (setting up refurbishment centre in the Netherlands for Philips N.V. (2014), new factory in Iowa for DSM N.V. (2014)).
If not named specifically but instead the city or country, all subsidiaries in that city or country were expected to receive attention.
A
(A) All other subsidiaries, (B) When mentioning is due to discontinuation of MNE operations of that facility (Neways N.V., 2014), (C) Just stating subsidiaries as ‘MNE inventory check-up’ (TomTom N.V., 2014)
Since geographic identification was important for coding and measuring the dependent variable, it was essential that the geographic scope in the letters to
TABLE 3. Coding table for dependent variable (geographic area)
Geographic area Attention granted Attention not granted
Attention towards Europe All Subsidiaries in EU28 + Belarus, Switzerland, Norway, Serbia, Ukraine (Philips N.V., 2014)
A, C
Attention towards North America All subsidiaries in the United States and Canada (Philips N.V., 2014)
A, C
Attention towards Asia All subsidiaries in Russia, Japan, China, India, Mongolia, North Korea, South Korea, Indonesia, Pakistan, Sri Lanka, Philippines, Taiwan, Hong Kong, Brunei, Malaysia, Papua New Guinea, Singapore, Thailand, Vietnam (NXP Semiconductors N.V., 2014)
A, C
Attention towards Middle East All subsidiaries in United Arab Emirates, Georgia, Israel, Kuwait, Kazakhstan, Oman, Qatar, Saudi Arabia, Turkey (Nedap N.V., 2009)
A, C
Subsidiary mentioned specifically by name
Specific subsidiary mentioned (Nedap N.V., 2014)
A, B, C
Attention towards Western Europe
Subsidiaries in Netherlands, Belgium, Luxembourg, Denmark, Sweden, Norway, Finland, Iceland, UK, Ireland, France, Germany, Spain, Portugal, Italy, Austria, Switzerland (DSM N.V., 2014)
A, C
Activities, meso-relations or performance in specific country or city
All subsidiaries in that particular country or city. Hong Kong and Taiwan not part of China,
India does not include Sri Lanka. (Berry et al., 2010; Hofstede, 2001; Philips N.V., 2014)
A, B, C
Activities, meso-relations or performance in ‘emerging market’
All Subsidiaries in (BRICS) countries. Brazil, Russia, India, China and South Africa (Philips N.V., 2009; Adu et al., 2015)
A, C
(A) All other subsidiaries, (B) When mentioning is due to discontinuation of MNE operations of that facility (Neways N.V., 2014), (C) Just stating subsidiaries as ‘MNE inventory check-up’ (TomTom N.V., 2014)
Independent variables
Focussing on only the widely used cultural distance measure of Kogut & Singh based on the Hofstede index does not provide the most accurate and complete picture of the three pillars of institutional distance; normative, cognitive and
al., 2013) measure institutional distance as a whole by including the World Bank’s six governance indicators. These indicators are: accountability, political stability,
government effectiveness, regulatory quality, rule of law, and corruption control. Other studies include De Jong et al., (2015) who follow a new approach that incorporates a measure from Hakanson & Ambos, (2010) combined with a separate improved Kogut & Singh based measure for cultural distance (De Jong et al., 2015) developing measures capturing geographic distance, economic distance and other intra country differences. Moore et al., (2015) relies on the World Competitiveness Yearbook index for measuring the regulative and normative dimension, and on a standardised Kogut and Singh index for the cultural-cognitive dimension (Moore et al., 2015). Both use multiple measures to create an overall view of institutional distance (De Jong et al., 2015; Moore et al., 2015).
Yet another study (Zhang et al., 2014) used primary instead of secondary data to measure the three dimensions of institutional distance. This measure was not since this thesis relied on secondary data.
All studies share a clear cultural element based on the Kogut & Singh index from Hofstede. No different exactly are Berry et al., (2010) who created a
Cultural Distance. Cultural distance was measured with the original Kogut &
Singh (1988:422) formula because of its widespread use. The formula can be found in Figure 2 in the appendix. The country used as reference is The Netherlands. Where Berry et al. (2010) measured the index based on World Value Survey data, this thesis measured the index with the original dataset from Hofstede because of its reliability in previous studies. Hofstede’s dimensions of culture dates back to 1980 and covered a set of 77 countries. In the following years, the dataset has been extended with more cultural distance scores from other studies to enlarge the list. Whenever a country culture score is from an augmented study other than the original 77 from Hofstede himself, it had been highlighted in the database. The data is collected from the
Hofstede Centre that draws upon (Hofstede, 2001) for its data. Cultural distance refers to the normative and cognitive distance dimensions since it reflects the differences in cognitive structures, social norms, and values and believes between countries just like the Hofstede country values (Scott, 1995). It is considered a variable that was fixed over time in this thesis.
Knowledge Distance. Following Berry et al. (2010), knowledge distance is
measured by taking the ratio of all filed patents in the United States by a country over the total population in millions of that country in a particular year. The data is sourced from the United States patent and trademark office (USPTO, 2016) and is used in conjunction with data from The World Development Indicators database (WDI, 2016) from The World Bank. Unlike Berry et al. (2010), the amount of scientific articles are not included, as it does not necessarily contribute to firms’ knowledge. Measuring the amount of patents connects with the cognitive and normative dimension of
institutional distance since it reflects the differences in cognitive structures and the underlying societies’ information collection, processing and interpretation (Scott, 1995), and with the role of schools in educating on societies’ norms and believes. This variable was expected to change over time.
Administrative Distance. This dimension measures differences in
bureaucratic patterns based on colonial ties, religion, language and legal system. The scores for the relevant countries are taken directly from the Berry et al. (2010) study. It reflects institutional distance on the regulative dimension and normative dimension since it involves differences in legal structures and therefore rules. Moreover,
Geographic Distance. Lastly, geographic distance measures the great circle
distance (in kilometres) between the countries. The scores are taken directly from the Berry et al. (2010) study. It is included because according to Berry et al. (2010), geographic distance increases the costs of transportation and communication. This in turn reflects the degree of a countries’ ability to stay connected with other countries in terms of trade. It relates to the cognitive, regulative and normative dimension, since it relates to the process of collecting information for cognitive structures (availability of information), therewith the formation of norms and shared believes is affected and different laws and rules ought to be developed (Scott, 1995).
Controls
There are some effects that are expected to influence the association between institutional distance and corporate attention that need to be controlled for or
otherwise they would lead to a miscomprehension of the data.
India/China. Ambos & Birkinshaw (2010) mention that special attention is
given to these countries by headquarters. Therefore it is expected that subsidiaries operating in these countries will receive more attention just because they are located in China or India. It has been established that these emerging markets China & India receive more attention from headquarters, because top management has high hopes of gaining huge profits there (Ambos & Birkinshaw, 2010). For Western MNEs this directs the corporate attention towards their distant subsidiaries, and away from the subsidiaries enjoying the proximity argument normally. This is a dummy variable that will equal ‘1’ if a subsidiary is from China or India, and “0” otherwise.
Market Size. Corporate attention is given to those markets where money can
be made and lost (Plourde et al., 2014), therefore a next logical step is to control for the size of the market that a subsidiary is finding itself in. Market Size could be interpreted different ways, however it was chosen to use the following study from Dib et al. (2016), measuring the market size of a country as the sum of GDP of that
international $, the log is taken to smoothen the data. The data is sourced from The World Development Indicators database (WDI, 2016) from The World Bank.
Company Size. There is some difference between the MNEs in the set in
terms of size. For example, some firms have more subsidiaries than others in the set; therefore the data could be drawn heavily on the set of subsidiaries from these MNEs. Since every MNE in the set is equally important in this research, it has been therefore opted to control for this effect as well. The total of subsidiaries of each MNE in the set is summed up, after which a new variable is created that assigns every subsidiary with its’ parent total frequency of subsidiaries. This variable is then used in the regression as a control for the company size.
Data analysis
Binary logistic regression. Since the dependent variable is a binary categorical variable, current research used a customised version of multiple
regression: binary logistic regression. After checking if the assumptions are met, all data was tested at a confidence level of 95%.
Two regressions. There is quantitative information about two sets of (in)dependent variables, therefore the opportunity was taken to do two regression analyses, one for each year. Both regressions will have similar variables each measured with their own data.
China/India indicator. This variable was analysed with the group ‘0’ as reference group.
Direction of the relationship
4. Results
This section presents the results of the data analyses belonging to the quantitative research.
Assumptions (model)
The assumptions for a binary logistic regression were checked by using Statistics for SPSS (Field, 2009)
Assumption 1: linearity with log odds.
We assume in logistic regression that linearity exists between the continuous independent variables and the logit of the outcome (Field, 2009:273). Therefore, the main focus here is the significance of the interaction term (between the continuous independent variables and their individual logs). For 2009 the assumption was met. However, for 2014 the interaction terms of Administrative Distance and Knowledge Distance were significant. This means that for 2014 the results should be interpreted with care. See Table 4 in the appendix for the statistical evidence regarding this assumption.
Assumption 2: independence of errors. Since the independent variables jointly should form institutional distance, it is expected that they have a high association among each other. However, a Durbin Watson test was conducted to support this reasoning. For both years, 2009 and 2014, the test statistic equalled a value below 1, which indicates a high dependence on each other. See Table 5 in the appendix for the statistical evidence regarding this assumption.
Assumption 3: multicollinearity. For both 2009 and 2014, The VIF of all variables is lower than ten, and have a tolerance higher than .1, therefore it is concluded that there is no multicollinearity present. See Table 6 in the appendix for the statistical evidence regarding this assumption.
Missing values. Any subsidiary that had a missing value along at least one of the variables has been omitted from the analysis.
Correlation table
TABLE 7. Correlation table 2009
** is significant at a 95% confidence level,* is significant at a 90% confidence level, mean (M), standard deviation (SD), indicator variable ((i))
TABLE 8. Correlation table 2014
** is significant at a 95% confidence level, * is significant at a 90% confidence level, mean (M), standard deviation (SD), indicator variable ((i))
Variables M SD 1 2 3 4 5 6 7 8
1. Corporate Attention (i) .23 .42 1
2. Cultural Distance 1.87 1.50 .14** 1 3. Knowledge Distance 100.67 100.98 -.14** -.36** 1 4. Administrative Distance 8.59 11.10 -.08** .13** .11** 1 5. Geographic Distance 3977.79 4064.62 -.04 .52** .13** .22** 1 6. China/India (i) .08 .27 .36** .47** -.30** -.14** .26** 1 7. Market Size 28.08 1.46 .17** .13** .50** -.11** .28** .39** 1 8. Company Size 324.38 197.40 -.30** .09** -.03 .05* .16** .03 -.003 1 Variables M SD 1 2 3 4 5 6 7 8
1. Corporate Attention (i) .43 .50 1
Below in Table 9, a frequency table was setup to give extra information regarding the descriptive results for the indicator variables.
TABLE 9. Frequency table indicator variables
Regression results
In this paragraph, the results of the regression analysis are presented per year individually. Model 1 contains controls only. Model 2 includes controls and the independent variables. A comparison of both years’ results, the differences between the models, as well as a discussion on the signs of the independent variables will be presented in the discussion later.
Regression results 2009. The results for 2009 are shown in Table 10. All results were tested at a 95% confidence level.
As can be seen from Table 10, the results for model 1 (step 1) exhibited a prediction correctness increase of 3.7% over step 0. Which means that the inclusion of the controls helps in correctly predicting the outcome value of the dependent variable. The controls: Market Size, China/India and Company Size thus have significant exploratory power in the regression of 2009. In line with the expectations from the literature (Plourde et al., 2014), a higher market size increased the probability getting Indicator Variable Event occurring
(1)
corporate attention, however it was insignificant with a p-value of .06. China/India was a very significant predictor of the propensity for receiving corporate attention (p<.01), and was positively associated. This was also as expected (Ambos &
Birkinshaw, 2010). Company Size was also significant and was negatively associated with the probability of getting attention. The more sister subsidiaries in the MNE network, the less attention a subsidiary would probably receive.
Model 2 in Table 10 exhibited a prediction correctness increase of 4.2% in step 1 over step 0, which means that the inclusion of the controls and independent variables together were meaningful in predicting the outcome value of corporate attention. This provides evidence for supporting the main hypothesis. Moreover, Cultural Distance and Administrative Distance had positive signs, but Knowledge Distance was negative. The coefficient of Geographic Distance was too small for interpreting its sign correctly, thus by following the standardised coefficients
displayed for 2009 in Table 12 it was interpreted that Geographic Distance carried a negative sign.
This shows that several institutional dimensions carry different signs, like expected. Unfortunately however, Cultural Distance and Administrative Distance were insignificant, this result will be discussed in the next section. Since not all dimensions of institutional distance have been found significant, the main hypothesis is therefor partially supported. The statistics show that an increase in geographic distance and knowledge distance would decrease the probability of receiving corporate attention in 2009. The controls showed little change between the models. Though Market Size, which was insignificant in model one became significant in model 2, and thereby all controls were significant. All controls still carried the same sign from model 1. The Nagelkerke R squared slightly increased from .30 in model 1, to .33 in model 2.
Regression results 2014. The results for 2014 are shown in Table 11. All results were tested at a 95% confidence level.
In model 1 for year 2014 the results showed an increase of correct prediction of 7.0% in step 1 over step 0. All three controls (Market Size, China/India and
which was unexpected, this will be discussed in the next section. The Nagelkerke R squared was 0.10.
The independent variables and the controls in model 2 increased the correct prediction by 12.8% when comparing step 1 with the prediction in step 0. As in 2009, the signs of the regression coefficients where not consistent among the four
independent variables; Cultural Distance had a positive sign, but the result was still insignificant. But there were also differences; unlike in 2009 Knowledge Distance was now positively associated with the probability of corporate attention and was significant, Administrative Distance was negatively associated and insignificant. Finally, Geographic Distance was significant as well, however just as in 2009 the regression coefficient was too small for interpreting the sign correctly with
confidence. With the results of the standardised coefficients for 2014 in Table 12 it was seen as negative.
TABLE 10. Regression results 2009
Model 1: controls, model 2: controls + Independent variables, ** significant at 95% confidence level. Sample size ((N)), regression coefficient (B), Wald statistic (Wald), Odds ratio ((Exp(B)))
TABLE 11. Regression results 2014
Model 1: controls, model 2: controls + Independent variables, . ** significant at 95% confidence level. Sample size ((N)), regression coefficient (B), Wald statistic (Wald), Odds ratio ((Exp(B)))
Independent variables Corporate Attention 2009 (N) Model 1: 2176 Model 2: 2146
1 2
Main effects B(SE) Wald Exp(B) B(SE) Wald Exp(B)
Cultural Distance .09(.06) 2.14 1.10 Knowledge Distance -.01(<.01) 6.74** 1.00 Administrative Distance .01(.01) 1.23 1.01 Geographic Distance .000(.000) 26.12** 1.00 Controls Market Size .08(.05) 3.18* 1.09 .34(.07) 21.12** 1.40 China/India 2.85(.22) 171.89** 17.35 2.38(.30) 65.00** 10.81 Company Size -.01(<.001) 210.24** 1.00 -.01(<.001) 188.12** 1.00 Nagelkerke R-squared .30 .33
Prediction Correctness (improvement over step 0) 81%(+3.7%) 81.4%(+4.3%)
Independent variables Corporate Attention 2014 (N) Model 1: 2566 Model 2: 2526
1 2
Main effects B(SE) Wald Exp(B) B(SE) Wald Exp(B)
Cultural Distance <.01(.05) .006 1.00 Knowledge Distance <.01(<.001) 61.06** 1.00 Administrative Distance -.01(.01) 1.38 1.00 Geographic Distance .000(.000) 44.62** 1.00 Controls Market Size .42(.03) 166.93** 1.53 .28(.05) 29.82** 1.32 China/India -.89(.16) 29.54** .41 .56(.23) 6.18** 1.76 Company Size -.002(<.001) 60.80** 1.00 -.001(<.001) 43.78** 1.00 Nagelkerke R-squared .13 .19
For the purpose of speculating on the overall sign of the hypothesized
relationship, the regression coefficients were standardised; some of the variables had coefficients very close to ‘0’ as seen from Table 10 and Table 11. Geographic Distance even had a coefficient of .000, neither clearly negative nor positive.
However, by standardising the coefficients a different picture emerged. As seen from Table 12 below, Cultural Distance and Administrative Distance are both insignificant for both years like in Table 10 and Table 11 so the focus will be laid on the other dimensions. For 2009, the odds for receiving corporate attention decreased (.76) when the variable Knowledge Distance was increased with one standard deviation, and a similar increase in Geographic Distance also led to a decrease (.61) on the odds. This posits a negative effect of institutional distance on corporate attention.
Surprisingly contrasting for year 2014, the odds for getting corporate attention increased strongly with every standard deviation increase in Knowledge Distance (1.80), however the odds still decreased with every standard deviation increase of Geographic Distance (.61). The positive effect of Knowledge Distance seems to be bigger than the negative effect of Geographic Distance, therefore this posits an overall moderate positive effect of institutional distance on the degree of corporate attention. The results showed that both years where very different from each other in terms of regression output and topics mentioned in the letters to shareholders. This is reflected in Tables 1, 10, 11 and 12. A more detailed discussion regarding these differences is situated in the following section.
TABLE 12. Standardized coefficients
Independent variable Betas Exp(b)
5. Discussion & Conclusion
This section states the discussion of the results. Together with the help of the prominent attention topics used in the coding procedure for both years, possible explanations and interpretations for the results are formed. This section will be concluded with the limitations and recommendations for future research and the conclusion.
Models 2009
For the first model, the inclusion of all controls increased the correct prediction. However, although Market Size was expected to have an effect on corporate attention, it was insignificant. It is an interesting result, signalling the need for more detailed future research regarding the relationship between the size of
markets and corporate attention. China/India was significant. If a subsidiary resided in China or India, the probability of receiving more corporate attention increased, which was as expected. The attention topics in 2009’s letters to shareholders (Table 1) showed a large portion of attention to the financial crisis and to the process of reorganisation. This could underpin the finding that subsidiaries in China and India would receive more attention; perhaps the MNEs were trying to move production to Asia to realize cost savings. Furthermore, a large portion of attention went to special segments and (new) products within the company not specifically linked to
subsidiaries. It could be that some promising segments and (new) products were from China & India. Company Size reveals that larger networks have a negative impact on attention for subsidiaries. There is just so much attention that headquarters can give and the larger the group is the lower chances are for the individual subsidiary.
showed different results. The remaining independent variables Geographic Distance and Knowledge Distance had negative signs, which supported the ethnocentric argument established from the literature review; headquarters are more focussed on institutions similar to theirs’ when distributing attention, headquarters in 2009 could thus have been focussing on similar knowledge levels, or geographically close subsidiaries. But also the existence of an information transfer barrier is a possibility; more geographic distance implies that more effort was necessary at least in terms of practical costs to transfer information from subsidiary to headquarters and vice versa. This was restricting corporate attention. Moreover, a difference in knowhow of technology used in the knowledge transfer between the subsidiary and headquarters could have been a restricting factor as well. Also, the costs to protect intellectual property rights or acquiring them could have been too high in 2009 and less important than capitalizing on subsidiaries that had similar knowledge levels which therefore received more attention. However, the data collection required to acquire this specific information was beyond the scope of the research.
Also from the literature review it was established that there are arguments regarding subsidiary- or corporate activities that were able to counteract an
dealt with the financial crisis by sticking to procedures and protocols that were best known to them (ethnocentrism) and this ethnocentrism favoured close subsidiaries over distant ones. Nevertheless, it is interesting to see that subsidiaries in China or India, as shown from Table 10 were still promising and drew attention from the headquarters. It could be that this was for profit seeking reasons, however also reorganization with the goal of cost savings could have been a motivation.
Since both significant predictors carried the negative sign it is apparent from Table 12 that the standardised results over the year 2009 showed a negative effect of institutional distance on corporate attention. The other (insignificant) predictors: Cultural Distance and Administrative Distance, given the size of their coefficient now could not have changed the overall sign if they were significant. The overall
relationship seems to be negative. This could happen because the explained variance of Cultural Distance and Administrative Distance could have been already captured in the effect of other dimensions like Knowledge Distance or Geographic Distance, forcing Cultural Distance and Administrative Distance to be more obsolete and therefor making them less significant i.e. insignificant. As seen from the correlation tables (Table 7 and Table 8) and setting the boundaries according to (Field, 2009:57), it is observed that the correlation between Cultural Distance and Knowledge Distance was moderate to large, and the correlation between Cultural Distance and Geographic Distance was large. This could support the notion that the effect of Cultural Distance was already captured in either Knowledge Distance or Geographic Distance.
However, the correlations of Administrative Distance with Knowledge Distance and Geographic Distance were small, therefor it is less plausible that the effect of
Administrative Distance was captured in other dimensions. The reason for its
insignificance could probably be found elsewhere. In Table 12 it can be observed that Geographic Distance is a very strong predictor while it was expected that Geographic Distance would have widespread effects on all the institutional dimensions;
Models 2014
The model 1 in 2014 looked very similar to model 1 from 2009, except for the control China/India which had a negative sign instead of positive, and all the controls were significant this time. This shows that also in 2014 the controls helped predicting the dependent variable outcome better than without the controls. The negative sign for China/India was not as expected. This suggests that although large markets were getting the attention, China or India were mostly not one of them. It could be that some (unattractive) country characteristics unique to China and India could not offset the benefits of being a large market. Future research should investigate this in more detail.
In model 2 the sign of China/India changed from negative to positive again. It is unsure why the inclusion of the independent variables in model 2, led to this sign change. One reason could be that because Market Size’s coefficient (decreased) changed a lot from model 1 to model 2, this could have led to a change in the coefficient of China/India as well, but due to this uncertainty more research in necessary in the future the investigate this phenomenon. The inclusion of the
independent variables improved the correct prediction thus their are found relevant for the model. The controls were all significant, with Market Size being positive and Company Size being negative. As with 2009, Cultural Distance and Administrative Distance were insignificant, while Knowledge Distance and Geographic Distance were still significant. Furthermore, contrasting with 2009, the Knowledge Distance coefficient was positive while in 2009 it was negative. This shows that the knowledge base of a country measured in the amount of patents filed, formed a basis in 2014 for the positive effect of institutional distance instead of a negative; a bigger knowledge gap between the headquarters and subsidiary meant more attention from the
headquarters. It could be that headquarters were seeing threats in these knowledge gaps, or that the headquarters listened to the ‘voice’ of these subsidiaries, or that expats were able to sustain a good communication link between these subsidiaries and headquarters. Geographic Distance had a negative coefficient, just as in 2009.
Geographically close subsidiaries got more attention than those far away. Perhaps Western MNEs were paying attention to subsidiaries that were
on; having more knowledge than the host country was probably beneficial, but it could also be that spill overs from countries with superior knowledge were attractive.
A quick look at Table 12 shows that the positive effect of Knowledge Distance was bigger than the negative effect of Geographic Distance. A positive effect of institutional distance on the degree of corporate attention is therefore expected. This positive effect even remains after taken into account the effects of the other
(insignificant) predictors: Cultural Distance and Administrative Distance. This suggests that the ‘attention inhibiting’ ethnocentric argument or information transfer barrier as proposed for 2009, was lower than the attention drawing arguments.
In 2014, recognition of performance was a prominent attention topic (Table 1), but this attention was accredited to specific segments and product lines; The
subsidiaries in specific individual countries were not mentioned in reference to performance evaluation nor were individual subsidiaries. Moreover, Table 1 shows that performance and demand was mentioned on world region level, as well as a focus on specific meso-relationships, recent acquisitions and joint ventures. Table 1 also shows that MNEs in 2014 had attention for awards and prestige; the MNEs valued their reputation a lot. Keeping a steady performance without drastic organisational change and some market expansion was therefore key in 2014.
Western MNEs had increased attention for large markets and that included China and India (Table 11). Since performance and reputation were prominent attention topics in 2014 (Table 1), perhaps activities by subsidiaries in China and India promoted these goals thereby making it attractive for headquarters to increase its exposure further to Chinese or Indian way of business. Specific business
relationships, joint ventures, Greenfield investments and acquisitions (also prominent attention topics in 2014 (Table 1)) in large markets including China and India could perhaps have provided the attractive market expansion and/or knowledge gaps that MNEs were looking for.
Comparing the years
evident that all significant independent variables were negatively associated with corporate attention in 2009, proposing a negative overall relationship, but one independent variable (Knowledge Distance) was extremely positive in 2014
proposing an overall positive relationship in 2014. It can be said that the signs of the coefficients of the independent variables measuring institutional distance are
inconsistent between 2009 and 2014. The data collected within this 5-year time frame showed that from one year to another, some measures of institutional distance have a positive effect on the probability of receiving corporate attention, and some measures have negative effects. These positive and negative effects seem to interact with each other thereby influencing the overall sign of the hypothesized relationship. This implies that institutional distance is more complex in predicting corporate attention, however this goes beyond the scope of this research. More research is required to make stronger inferences regarding the overall sign of the relationship, and more research of longitudinal nature is necessary to investigate potential time effects.
Limitations
Of course this research has some limitations as well.
The dichotomous coding of attention suggested by Plourde et al. (2014) did not capture the double entries of attention in a letter to shareholders. This could falsely create the view that subsidiaries are on the same level of attention on the corporate radar. For example, an acquisition or joint venture, mostly a one-time event in one country, is very specific. But could be an enormous point of attention, while a decline in market share in Europe is seen as attention to multiple countries containing perhaps multiple subsidiaries each.
Only Dutch MNEs in the industry sector have been chosen, and only 25 MNEs with subsidiaries in emerging markets (BRICS).
The sample of 2009 was significantly smaller than that of 2014 due to the lacking of five MNEs and the resulting missing values.
the specific activities, segments and product groups. A large part of attention could then not been attributed to an individual subsidiary and stayed quantitatively unnoticed. Performance mentioned in terms of segments and product lines were among the highest attention sources in the letters to shareholders. Globalization, and specifically the cooperation between subsidiaries on knowledge sharing and
responsibility for certain tasks within the MNE, let decision makers think about segments and product groups rather than location. Perhaps there should be more focus on measuring attention along activities and products stretching country borders in order to capture and study the more complex attention structures in MNEs.
Letters to shareholders are reflections of top management over specific business years and future plans. It should represent the company direction according to the corporate governance system, but letters to shareholders are part of the annual report, an investor oriented corporate media and is used as promotion of the company towards external stakeholders (Ambos & Birkinshaw, 2010), and is written in the view of one or two person(s). They are not all of the same size and differ in the amount of detail in which they are written. The consistency of the measure can therefor be questioned. This was observed in the diversity of the letters to shareholders.
