STRUCTURE, STRATEGY & ATTENTION: HOW DIVERSIFICATION INFLUENCES THE ATTENTION DISTRIBUTION OF HEADQUARTER EXECUTIVES

(1)

STRUCTURE, STRATEGY & ATTENTION:

HOW DIVERSIFICATION INFLUENCES THE

ATTENTION DISTRIBUTION OF HEADQUARTER

EXECUTIVES

Master’s Thesis, MSc Advanced International Business

Management & Marketing

University of Groningen, Faculty of Economics and Business

Newcastle University, Newcastle Business School

(2)

ACKNOWLEDGMENT

(3)

Master’s Thesis International Business Strategy Christoph Osterloh University of Groningen & Newcastle University

1

ABSTRACT

(4)

2

TABLE OF CONTENT

ACKNOWLEDGMENT ... 2 ABSTRACT ... 1 INTRODUCTION... 3 THEORY ... 10

2.1. The Attention-Based View of the Firm ... 11

2.2. Types and Components of Attention ... 13

2.2.1 Types of Attention ... 14

2.2.2. Components of Attention ... 16

2.3. Internal Factors Influencing Headquarters’ Attention... 17

2.4. External Factors Influencing Headquarters’ Attention ... 25

RESEARCH METHODOLOGY ... 29 3.2. Sampling ... 29 3.3. Variables ... 30 3.3.1. Dependent Variables ... 30 3.3.2. Independent Variables ... 32 3.3.3. Control Variables ... 36 3.3.4. Conceptual Model ... 37 RESULTS ... 39 4.1. Descriptive Statistics ... 39 4.1.1. Missing values ... 39 4.1.2. Factor Analysis ... 41 4.1.3. Correlation Analysis ... 43 4.2. Testing Hypotheses ... 44 4.3. Multicollinearity ... 50

4.4. Alternative Explanation of the Core/Non-Core Industry Effect ... 51

4.5. Industry Concentration ... 52

4.6. Hypotheses Summary ... 56

DISCUSSION ... 57

5.1. Internal Factors ... 57

5.2. External Factors ... 62

5.3. Limitations & Suggestions for Future Research ... 66

CONCLUSION ... 69

(5)

3

INTRODUCTION

Since the emergence of the modern multinational enterprise (MNE), researchers have tried to explain how these organizations behave, which factors influence their strategy and structure and what makes them act in a certain way (Williamson & Winter, 1993). Early research in this stream was dominated by the so-called economic theory of the firm (Teece, 1982), which describes MNEs’ behaviour as driven by availability of complete information, profit-maximisation and abundance of internal resources (Pierce et al., 2008). However, soon after the conceptualisation of the economic theory of the firm, researchers began to question its assumptions and consequently developed the so-called behavioural theory of the firm (Cyert & March, 1963). This theory does not assume that the MNE is able to comprehend all information available, does not try to maximise its profits, but tries to reach satisficing results and views internal resources as scarce (Pierce et al., 2008).

(6)

4

they focus their scarce attention on (Simon, 1976). The limit of the human cognitive capacity was described by Simon (1947) as so-called bounded rationality. Given this bounded rationality, researchers investigated the factors and circumstances that influence executive managers’ attention and how these factors direct managers’ attention towards certain topics and issues (Shepherd, McMullen & Ocasio, 2017).

In several studies, researchers such as Bouquet and Birkinshaw (2008a) and ul Haq, Drogendijk and Blankenburg-Holm (2017) investigated potential sources, factors and circumstances that influence executive managers’ allocation of attention within organizations. These researchers discovered that factors such as subsidiaries’ weight (that is the size and strategic importance of the subsidiary), voice (that is the bottom-up approach of initiative taking and profile building of subsidiaries) (Bouquet & Birkinshaw, 2008a), and their cultural and geographic distance (ul Haq et al., 2017) from headquarters, influence how much attention executive managers devote to these subsidiaries.

(7)

5

(Bouquet & Birkinshaw, 2008a: 595). These findings imply that the influence of corporate structure has not completely been researched yet.

In organizational research, the concept of corporate structure consists of several components, such as organizational design, level of integration and role of subsidiaries and their interactions with company headquarters (Robbins & Barnwell, 2006). However, one aspect of corporate structure that is highly interesting for the attention-based view of the firm is diversification of the organization in terms of having subsidiaries operating in different industries.

(8)

6

with high structural complexity to obtain and comprehend all information necessary to make strategic decisions (Quinn, 1980).

(9)

7

In the last decades, the organizational structure and strategy of MNEs has substantially changed, leading to investments and divestments in certain industries and business lines (Osterwalder & Pigneur, 2010). For instance, Daimler has completely changed its organizational structure and defines new businesses they want to focus their efforts on (Hubik, 2018). Since executive managers’ attention has the potential to enhance subsidiaries’ performance (Conroy & Collings, 2016) it is important and relevant to understand how executive managers’ attention is directed in a diversified MNE.

This exploratory study does not only try to close the abovementioned research gap and investigates an important field of study, it also unites two crucial dimensions of the

Figure 1.1. – Illustration of the diversity of stimuli executive managers have to process

IT Industry

Chemicals Industry Aeroplane Industry

(10)

8

attention-based view of the firm by studying cognitive and structural complexity as determinants of attention distribution within an MNE. Specifically, this study aims to understand the flow and distribution of attention in a diversified MNE to reveal other factors that influence headquarters’ attention beyond weight and voice of the single subsidiary (Bouquet & Birkinshaw, 2008a), or more general aspects, such as geographic and cultural distance.

Given the lack of research in this direction and its importance, the research question of this study is:

“How does diversification of a company influence the attention a subsidiary receives from the company’s headquarters?”

Since the concept of diversification is rather broad and has not been researched yet in relation to the attention-based view, two sub questions are developed to answer the main research question. These two sub questions will deal with the internal and external aspects and circumstances of a diversified MNE.

(11)

9

states: “How do internal, structural factors of a diversified MNE influence the attention

a subsidiary receives from the company’s headquarters?”

Secondly, the external, industry environment effect on headquarters’ attention flow towards subsidiaries is investigated. The external, industry-specific environment could contain factors that influence the distribution of executive managers’ attention. A diversified MNE, which operates in several different industries, has to cope with diverse industry-specific circumstances that might only affect subsidiaries operating in a certain industry. For instance, due to operating in a certain industry, some subsidiaries could be affected by hazardous environments with a potential for failures and uncertainty (Baker, Day & Salas, 2006), severe competition, unreliable demand conditions or surprising technological revolutions (Christensen, Raynor & McDonald, 2015). Understanding, analysing and anticipating various industry-specific environments poses high cognitive demand on executive managers, and thus potentially affect the distribution of attention in a diversified MNE. Therefore, the second sub research question of this study states: “How do external, industry-specific

factors of a diversified MNE influence the attention a subsidiary receives from the company’s headquarters?”

(12)

10

in the core industry receive more attention than subsidiaries operating in the peripheral industries of the same MNE. Thirdly, the paper expands current theory which is limited to the characteristics of an individual subsidiary to a broader understanding by considering the characteristics of a whole business line or industry present in an MNE network. This expands theories such as weight and voice (Bouquet & Birkinshaw, 2008a) of the individual subsidiary towards a more comprehensive illustration of how cumulative factors of structure and strategy influence executive managers’ attention. The thesis will reach the research objectives by researching the attention subsidiaries perceive to receive from their headquarters by using survey data from 151 foreign subsidiaries of Swedish multinational enterprises. Measuring attention at subsidiary level was chosen, since attention is something that is experienced by the receiving party and therefore has an influence on the behaviour of subsidiaries.

THEORY

(13)

11

2.1. The Attention-Based View of the Firm

The attention-based view of the firm proposes that the focus and distribution of executive managers’ attention influences organizational decisions, actions and firm performance (Abebe, 2012). By investigating how executive managers’ attention was influenced in certain situation, it is possible to examine the process of executive managers’ decision making and organizational actions. Ocasio (1997: 189), one of the leading researchers in the field of attention, defines attention as “the noticing, encoding, interpreting and focusing of time and effort by organizational decision makers”. In the context of the headquarter-subsidiary relationship, the concept of attention can be defined as “the extent to which a parent company recognizes and gives credit to a subsidiary for its contribution to the MNE as a whole” (Bouquet & Birkinshaw, 2008a: 279).

(14)

12

headquarter that destroy value at the subsidiary level.” (Conroy & Collings, 2016: 612). This thesis will solely focus on positive attention for two reasons: Firstly, according to Bouquet and Birkinshaw (2008b) it is difficult to differentiate between the concept of negative attention and control and monitoring behaviour of the MNE headquarters. Therefore, by focusing only on positive, value-enhancing attention the thesis will gain conceptual clarity by avoiding interferences and communalities between the concept of negative attention and control and monitoring behaviour. Secondly, in past studies, the focus has been mostly on negative, control-related attention. Therefore, this thesis is interested in the less often discussed positive, value-enhancing form of attention.

(15)

13

“managers live inside their frames and to a great extent don’t know what lies outside” (Hamel & Prahalad, 1994: 54).

2.2. Types and Components of Attention

(16)

14

In the following paragraphs, the different components and types of attention are introduced and described in more detail.

2.2.1 Types of Attention

Attention Perspective

The first type of attention Ocasio (2011) identified is the so-called attention perspective. According to ul Haq et al. (2017), the attention perspective is closely linked to the organizational strategy of the MNE. This means that the attention perspective serves “as a guideline for top managers to allocate their attention in line with the visions and goals of a specific organization” (ul Haq et al., 2017: 112). In the context of this study, this implies that diversified MNEs might have an attention

Attention Perspective Relative Attention Attention Selection Attention Engagement Supportive Attention Structural Attention Headquarter level (Types of Attention) Subsidiary level (Components of Attention)

(17)

15

perspective that favours and diverts more attention to certain industries, since they are perceived as more important to reach strategic goals.

Attention Engagement

Ocasio (2011: 1288) defined the second type of attention, attention engagement, as the “process of intentional, sustained allocation of cognitive resources to guide problem solving, planning, sensemaking and decision making”. According to ul Haq et al. (2017), this intentional focus of attention depends on headquarters’ knowledge about the local business operations and relates to the structures, procedures and communication routines that direct attention in an active way. In the context of a diversified MNE, this means that it represents the time and effort executive managers invest to understand issues and to follow certain stimuli. In other words, executive managers need to know and understand the business operations in several industries. This means they need to invest effort to gain knowledge about issues within each of these industries, which leads to higher cognitive demand. This can result in more attention devoted to industries the executive managers are knowledgeable about, since less effort is needed to understand and comprehend issues and stimuli in these familiar industries.

Attention Selection

(18)

16

knowledge and background of the executive managers severely influence which subsidiaries and industries receive their attention. Thus, subsidiaries which operate in industries that are not perceived as relevant to the overall business strategy and do not fit with the knowledge of the executive managers are likely to receive less attention.

2.2.2. Components of Attention

After having described the types of attention, this paper will now introduce the components of attention. According to Bouquet and Birkinshaw (2008a), the attention executive managers devote to their subsidiaries consists of visible, relative and supportive attention. Since this study investigates the perceived headquarter attention at subsidiary level, not all of these three components are completely applicable. Instead of visible attention, which illustrates the fact that the subsidiary is mentioned in official reports and documents from the MNE, structural attention is included in this study, since this component is more suitable for the subsidiary level of analysis (Drogendijk, ul Haq & Blankenburg-Holm, 2018).

Relative Attention

(19)

17

Supportive Attention

According to Bouquet and Birkinshaw (2008a), supportive attention reflects the idea that the MNE headquarters provide resources and knowledge to certain subsidiaries to enhance their value creation for the MNE. In the context of a diversified MNE, this implies that subsidiaries operating in a certain industry receive more supportive attention and hence can make use of more resources and knowledge.

Structural Attention

Structural Attention refers to the recognition given by MNE headquarters to subsidiaries by including them in strategic decision-making processes (Drogendijk et al., 2018). In diversified MNEs, this could mean that subsidiaries operating in a certain industry are more often included in the decision-making process than subsidiaries operating in other industries.

In the following paragraphs, this thesis will develop hypotheses regarding which internal and external factors have an influence on executive managers’ attention in diversified MNEs. The construct of attention is represented by the three components discussed in the previous paragraphs: relative, supportive and structural attention. These components should give a more fine-grained idea which aspects of the concept of attention are influence by the internal and external factors.

2.3. Internal Factors Influencing Headquarters’ Attention

(20)

18

Commonly, MNEs are conceptualized as networks of semi-autonomous units with differentiated resources (Ghoshal & Barlett, 1990). Hence, this paper assumes that in a diversified MNE resources held by subsidiaries differ depending on their industry. Different industries produce different stimuli. These stimuli catch executive managers’ attention and might even be institutionalized through organizational structures (Plourde, Parker & Schaan, 2014). Depending on their perceived importance for the MNE, information coming from certain subsidiaries or industries is treated with higher priority than from other subsidiaries or industries (Bouquet & Birkinshaw, 2009).

(21)

19

In this context, the concept of core industry and non-core industry is highly interesting. In a diversified organization it is possible to identify one or more core industries. Often, the core industry of a company marks the starting industry from which the organization has grown and expanded into other industries (Hnátek, 2015). As Hnátek (2015) explains, this core industry has special meaning for the MNE, since the presence of the organization in this industry represents the mindset and vision of its founder. Even when the founder has left the organization, this mindset and vision can still be present and influence the future actions of this organization (Harris & Ogbonna, 1999). In other words, the historical legacy of an organization leads to strategic inertia (Huff, Huff & Thomas, 1992) which affects what executives focus their attention on. This is in line with Bouquet and Birkinshaw (2008a: 580), who argue that headquarter executives tend to focus their attention “in the way they have always done”. The aspect of the historical legacy of the organization illustrates how the attention perspective favours stimuli from the core industry. Therefore, the attention perspective could explain that information coming from subsidiaries operating in non-core industries is regarded and categorized as less strategically important and hence these subsidiaries receive less attention from headquarters.

(22)

20

(23)

21

Figure 2.2. illustrates these findings. As the outcome of the attention perspective and attention engagement, subsidiaries operating in so-called core industries are likely to receive more attention from headquarters than subsidiaries operating in non-core industries of the same MNE. This is shown by the thicker arrows, which represent the attention flow from headquarters, towards the subsidiaries operating in the core industry of the MNE. This leads to the first hypothesis:

Core Industry

Industry Periphery (Non-Core Industries)

Figure 2.2.) Conceptual illustration of the flow of attention to core and non-core subsidiaries.

(24)

22 Hypothesis 1:

Subsidiaries operating in the core industry will receive more attention [1a)

relative attention / 1b) supportive attention / 1c) structural attention] from

headquarters than subsidiaries operating in non-core industries (industrial

periphery) of the same organization.

After laying out how the managerial mindset and frames influence the attention flow within diversified MNEs, the following paragraphs will deal with the argument that the weight of the assets of a whole business line or industry within the MNE network influence the flow of attention. This argument is built on Bouquet and Birkinshaw’s (2008a) finding that the weight of the individual subsidiary positively influences headquarters attention.

(25)

23

expanded concept called cumulative weight of subsidiaries could describe how the weight of a whole industry or business line directs and influences executive managers’ attention. Expanding the concept of weight allows to have a more comprehensive perspective on the factors influencing the flow of attention.

The concept of cumulative weight can be measured by analysing the intraorganizational concentration of subsidiaries according to their industry. This has been done in a similar manner in previous studies. For instance, Choi and Cowing (2002) analysed the relationship between diversification and concentration on economic performance for Korean companies. They propose to make use of the Herfindahl-Hirschman index to calculate the degree of concentration within an industry based on the number of subsidiaries of a company in this industry. Interestingly, they found that a larger concentration in an industry negatively influenced the overall profitability of the company (Choi & Cowing, 2002). Besides, both researchers did not test whether the higher concentration as an indication for cumulative weight leads to more attention from headquarters to those industries. The negative relation between concentration and profitability could have been the results of uneven spread of attention of the headquarters’ executives. In fact, it could be, that high attention towards the industries with a high concentration caused to overlook business opportunities in other industries.

(26)

24

Industry A

Industry D

Figure 2.3. illustrates this argumentation. In industry A the concentration of subsidiaries is the highest compared to the other industries within the MNE network. Therefore, subsidiaries operating in this industry receive more attention from headquarters since their cumulative weight is higher than the cumulative weight of the other subsidiaries operating in different industries. Thus, Choi and Cowing’s (2002) approach allows to expand Bouquet and Birkinshaw’s (2008a) notion of weight of the individual subsidiary to a whole business line within an MNE network. Hence, the second hypothesis states:

Industry B

Industry C

Figure 2.3.) Conceptual illustration of the flow of attention influenced by cumulative weight of an industry

(27)

25 Hypothesis 2:

The higher the concentration of subsidiaries in a certain industry, the more

attention [2a) relative attention / 2b) supportive attention / 2c) structural

attention] these subsidiaries receive from headquarter.

2.4. External Factors Influencing Headquarters’ Attention

After developing the hypotheses concerning the internal, structural effects on headquarters’ attention towards their subsidiaries, the following paragraphs will deal with the external, environmental industry effect on attention.

(28)

26

generates more stimuli that attract executive managers’ attention. They found out that this load of stimuli coming from the external environment often forces executive managers to focus their attention more on these external aspects than on internal issues.

(29)

27

Particularly in dynamic and uncertain industry environments, it is difficult for executive managers to make predictions about the future and anticipate changes. Therefore, especially negative effects of such dynamic industry environments, which potentially affect firm performance, require executives’ attention (Shepherd et al., 2017). Aldrich (1979) describes these environments as more demanding as they force managers to consider multiple factors including competitors, suppliers and customers. Not surprisingly, negative information and threats from these environments catch executives’ attention more than information that indicates good news or opportunities (Jackson & Dutton, 1988). One reason why attention is triggered more by negative information and threats can be found in the aspiration literature. For instance, Washburn and Bromiley (2012) state that companies and their executive managers have certain aspirations, so they expect an outcome of an organizational action to satisfy their demands. As soon as performance falls below a certain benchmark, their aspiration expectation is not met, and hence executive managers are not satisfied. In turn, being not satisfied forces these managers to focus their attention on the issues at hand.

(30)

28

need to innovate and to develop new technological solutions (Dyer et al. 2014). In other words, it describes how certain executive managers can forecast technologies needed to successfully compete within the industry. In highly dynamic industries, there is a high need to innovate and it is difficult to predict which kind of technology will be used in the future. Industries vary in terms of their demand and technological uncertainty. For instance, industries, such as the retail and hotel industries must face high demand fluctuations and are therefore categorized as having high demand uncertainties, whereas their need to develop new, innovative solutions is rather low. This means that the retail and hotel industries face low technological uncertainties. Subsidiaries operating in the most uncertain industries, such as the pharmaceutical industry, have to face both demand and technological uncertainties (Dyer et al., 2014).

All these studies imply that subsidiaries operating in highly uncertain environments are likely to receive more attention from headquarters than subsidiaries of the same organization operating in less uncertain environments. This leads to the third and fourth hypotheses:

Hypothesis 3:

Subsidiaries in a network of a diversified organization receive more attention

[3a) relative attention / 3b) supportive attention / 3c) structural attention] from

headquarters when they are operating in an industry characterised by high

levels of demand uncertainty than subsidiaries of the same organization

(31)

29 Hypothesis 4:

Subsidiaries in a network of a diversified organization receive more attention

[4a) relative attention / 4b) supportive attention / 4c) structural attention] from

headquarters when they are operating in an industry characterised by high

levels of technological uncertainty than subsidiaries of the same organization

operating in an industry characterised by low levels of technological

uncertainty.

RESEARCH METHODOLOGY

3.2. Sampling

(32)

30

different companies that traditionally pursue international expansion by setting up subsidiaries in different host countries. Furthermore, the sample consists of companies which operate in a variety of different industries, which is particularly suitable in the context of studying diversification. Secondly, for the independent variables “Core/Non-Core Industry” and “Industry Concentration” and for the control variables, data will be gathered from the Orbis database assembled by Bureau van Dijk. Thirdly, the data concerning the independent variables “Demand Uncertainty” and “Technological Uncertainty” is taken from the Compustat database. Fourthly, data for the control variable “Cultural Distance” is taken from the Hofstede Institute (2018). In the following paragraphs, all variables included in this study are introduced and described.

3.3. Variables

3.3.1. Dependent Variables

The dependent variable in this study is “Attention received from headquarter”. As mentioned above, researchers found three underlying components of which the concept attention consists (Bouquet & Birkinshaw, 2008a, ul Haq et al., 2017). Therefore, in this study, statistical tests will be conducted with these three components as dependent variables. These components of “Attention received from headquarter” will be operationalized as following:

(33)

31

high) how much attention a subsidiary receives from the company’s headquarter with respect to (1) their own expectations, (2) other subsidiaries in the same geographic area, and (3) other subsidiaries with similar size.

2) Perception of Supportive Attention: This concept was initially developed by Bouquet and Birkinshaw (2008a) and is a three-item scale measured by a seven-point Likert scale (1 = not at all agree, 7 = completely agree). Questions the respondents answer deal with (1) the headquarter’s interest in gaining knowledge about the local business; (2) the headquarter’s interest in sharing best practices with subsidiaries; and (3) whether the headquarter has enough knowledge to understand information sent from the subsidiaries.

(34)

32 3.3.2. Independent Variables

Four independent variables have been chosen for this study: Demand Uncertainty, Technological Uncertainty, Core Industry/Non-core Industry and Industry Concentration.

(35)

33

The second variable is called “Technological Uncertainty” and is represented by the global industry R&D expenses as a percentage of the total revenue over 10 years. In other words, it measures how much money companies invest in researching new technologies in their specific industry (Dyer et al. 2014). Firms operating in industries characterized by high technological uncertainty spend a larger share of their revenue for R&D than firms operating in industries characterized by low technological uncertainty. Firms operating in industries with high technological uncertainty need to invest in innovation and new technology to keep up with their competitors. The values Compustat provides ranging from 0,05% of R&D expenses to almost 50% of R&D expenses from total revenue.

Core industry/Non-core industry: This binary variable will be developed by

comparing the core industry of the company and the industry of the subsidiary as listed in the Orbis database. This database makes use of NACE industry codes (Nomenclature statistique des activités économiques dans la Communauté

européenne). This system identifies the economic activities of companies and

(36)

34

company by mentioning the corresponding NACE sections. In Orbis, every MNE can have only one core industry, but a few primary and several secondary industries. To account for the fact that some companies have more than one core business, the paper will take the industries mentioned as core industry and as primary industries as their core industry. In previous research, Heikkilä and Cordon (2002) describe the distinction between core and non-core businesses. For instance, according to them, Nokia’s core business is the information technology industry, whereas the production of boots is a non-core business. According to Bureau van Dijk (2018), the assembler of the Orbis database, information allowing to categorize the MNEs in industries comes from more than 140 information providers, including stock exchanges, rating organizations such as Moody’s or governmental agencies.

Industry concentration: The concept of industry concentration follows the work of

Choi and Cowing (2002) to calculate the Herfindahl-Hirschman index based on the number of subsidiaries operating in a certain industry. These measures should indicate on which industry a company focuses its presence. The data for the calculation is taken from the Orbis database. Industry concentration is calculated as following:

𝐻𝑗 = (

𝑇𝐴𝑗 𝑇𝐴

)

2 Where:

Hj = Industry Concentration of industry j TAj = The number of subsidiaries in industry j

TA = The total number of subsidiaries in the MNE network

(37)

35

In addition, to allow for cross-company comparison, the concentration values will be standardized. This is done for one reason: large MNEs with many subsidiaries operating in a multitude of industries would receive small concentration values, whereas smaller MNEs with few subsidiaries operating in just two to three industries would receive high concentration values. Therefore, the actual MNE size influences the concentration values and hence, comparison with values from MNEs of a different size is not possible. Standardizing the concentration scores will allow to reflect the weight of a whole industry within an MNE network independent of the size of the MNE. The formula for the standardized concentration scores states:

𝐻𝑠 = (

𝐻𝑎

𝐻𝑗

) 𝑥 100

Where:

Hs = Standardized concentration score of an industry in an MNE network Ha = Average concentration of all industries present in the MNE network Hj = Industry Concentration of industry j

A comparable approach of standardizing values has been used in previous business research by Singh, Murty, Gupta and Dikshit (2009). They also suggest multiplying the scores with 10 or 100 to receive more visually manageable scores. This is also done in this study.

(38)

36 3.3.3. Control Variables

The study incorporates four control variables, namely Subsidiary Size, Number of Countries of the MNE, Number of Subsidiaries of the MNE and Cultural Distance, that may also have an influence on the levels of attention received by subsidiaries.

Subsidiary size: Like in Bouquet and Birkinshaw’s (2008a) research, subsidiary size is measured by the number of employees working in the subsidiary and represents the idea of weight a subsidiary has in an organizational network. Measuring the number of employees per subsidiaries gives an impression about the manpower MNEs devote towards a certain job or area of business. Larger subsidiaries have more weight and hence more power to attract executive managers’ attention. Therefore, larger subsidiaries receive more attention from headquarters.

Size of MNE: The size of an MNE is expected to have an influence on the received

attention from headquarters (Drogendijk et al., 2018). To account for different dimensions of firm size, two control variables will represent the concept of size of the MNE: the Number of Countries of the MNE and the Number of Subsidiaries of the MNE. Given the limited attention executive managers can devote to countries and subsidiaries, a higher number of subsidiaries and countries the MNE operates in would cause that subsidiaries would receive less attention.

Cultural Distance: The cultural distance between the subsidiary and the headquarter

(39)

37

distance between the subsidiaries and the home market in Sweden. The measurement will be based on Hofstede’s (1991) four initial cultural dimensions “Individualism – Collectivism”, “Power distance”, “Masculinity – Femininity” and “Uncertainty avoidance” and the two later developed dimensions “Long vs. Short-term orientation” and “Indulgence vs. Restraint” (Hofstede, 2011). Hofstede’s cultural dimensions have been widely criticized in research (cf. Leung, Bhagat, Buchan, Erez & Gibson, 2005; McSweeney, 2009; Tung & Verbeke, 2010). Particularly Shenkar’s (2001) critique questions the validity of Hofstede’s cultural dimensions. However, this thesis still decided to make use of Hofstede’s cultural dimensions, because they allow an operationalization of a very abstract concept such as cultural distance. Secondly, as Kirkman, Lowe and Gibson (2006) showed, Hofstede’s dimensions are widely used in literature and thus are accepted in research. It is expected that a larger cultural distance will lead to a lower degree of attention received from headquarters.

3.3.4. Conceptual Model

(40)

38

Figure 3.1. – Conceptual Model

(41)

39

RESULTS

4.1. Descriptive Statistics

The data analysed in this paper consists of 151 subsidiaries, which are located in 50 different countries and employ between 1 and 3.000 employees (Mean: 207,77; Standard Deviation: 438,16). These subsidiaries belong to 68 companies which operate between 3 and 116 subsidiaries (Mean: 34,72; Standard Deviation: 32,70). These companies operate in 11 different industries: Mining and quarrying; Manufacturing; Electricity, gas, steam and air conditioning supply, water supply; Sewage, waste management and remediation activities; Construction; Wholesale and retail trade; Repair of motor vehicles and motorcycles; Information and communication; Financial and insurance activities; Professional, scientific and technical activities; and Administrative and support service activities.

4.1.1. Missing values

(42)

40

Figure 4.1. – Missing Data Analysis

Data source # Variable/Indicator

Total Count Missing Count Percent Dr og e nd ij k e t a l. (2 0 1 8 )

1 Headquarter attention; own expectation 151 3 2,0

2 Headquarter attention; other subsidiaries same area 151 11 7,3

3 Headquarter attention; other subsidiaries similar size 151 10 6,6

4 Headquarter interest in gaining local knowledge 151 0 0,0

5 Headquarter interest in sharing best practice 151 3 2,0

6 Headquarter has enough knowledge to understand information

151 0 0,0

7 Take part in meetings with headquarter that affect subsidiary 151 0 0,0

8 Take part in meetings with headquarter that affect whole company

151 1 0,7

9 Executive managers regularly visit subsidiary 151 2 1,3

10 Employees of the subsidiary visit other subsidiaries 151 1 0,7

11 Subsidiaries host regular meetings with executive managers 151 2 1,3

Compustat 12 Demand Uncertainty 151 27 17,9 13 Technological Uncertainty 151 27 17,9 O rbis 14 Core/Non-Core Industry 151 22 14,6 15 Subsidiary Size 151 1 0,7

16 Number of countries of the MNE 151 6 4,0

17 Number of subsidiaries of the MNE 151 8 5,3

Hofstede Institute

18 Cultural Distance 151 2 1,3

(43)

41

variable “Core/Non-Core Industry” collected from the Orbis database. Here, the industry classification appears to be missing completely at random (Rubin, 1976). For instance, some industry classifications are equally missing for subsidiaries in developed or developing countries or for larger and smaller subsidiaries in terms of employees employed. Since the amount of missing data in the Drogendijk et al. (2018) dataset, which comprises all attention related variables, is rather small, a missing-data-bias appears to be highly unlikely. In conclusion, the missing values of this study appear to be at an acceptable level and hence should not negatively affect the validity of this study.

The most serious problem with missing data occurs for the concentration analysis. The concentration analysis requires almost complete availability of the industry classifications for all subsidiaries of the MNEs investigated. However, for some of these MNEs the industry classifications for a substantial part of their subsidiaries is missing. Therefore, the tests of hypothesis 2, regarding the effect of industry concentration on attention, could not be done reliably. Thus, the concentration analysis is excluded from the regression analysis and will be conducted in isolation. The results are shown at the end of the results section.

4.1.2. Factor Analysis

(44)

42

indicators. The indicators capture three factors that explain 64,96% of the variance. The factors found correspond with the components of attention outlined in the theory section of this paper: “Relative Attention”, “Supportive Attention” and “Structural Attention”. Figure 4.2. shows the factor loadings of each indictor on the respective factor.

Figure 4.2. – Factor Analysis

# Indicator Factor 1: Relative Attention Factor 2: Supportive Attention Factor 3: Structural Attention

1 Headquarter attention; own expectation -0,712 2 Headquarter attention; other subsidiaries same area -0,893 3 Headquarter attention; other subsidiaries similar size -0,917

4 Headquarter interest in gaining local knowledge 0,814 5 Headquarter interest in sharing best practice 0,797 6 Headquarter has enough knowledge to understand information 0,784

7 Take part in meetings with headquarter that affect subsidiary 0,664 8 Take part in meetings with headquarter affecting whole company 0,779 9 Executive managers regularly visit subsidiary 0,544 10 Employees of the subsidiary visit other subsidiaries 0,663 11 Subsidiaries host regular meetings with executive managers 0,489

Cronbach’s Alpha 0,863 0,799 0,728

(45)

43

The three factors found will be treated as dependent variables. They are computed by summing the respective indicators.

4.1.3. Correlation Analysis

After analysing the underlying structure of the dataset, all variables used for testing the hypotheses are listed in figure 4.3. and described in terms of their mean, standard deviation and correlation.

Figure 4.3. – Correlation Analysis

# Variable Mean s.d. 1 2 3 4 5 6 7 8 1 Relative Attention 3,48 1,45 2 Supportive Attention 4,79 1,34 0,446** 3 Structural Attention 4,15 1,25 0,455** 0,481** 4 Demand Uncertainty 62,40 19,11 0,039 0,060 -0,017 5 Tech. Uncertainty 2,855 3,377 0,029 0,134 0,048 0,570** 6 Subsidiary Size 207,77 438,16 0,102 0,039 0,141 0,132 0,136 7 Number of countries 21,79 24,78 0,157 -0,078 0,013 -0,035 -0,007 0,205* 8 Number of subsidiaries 113,87 131,66 0,167 0,000 -0,011 0,056 -0,028 0,358** 0,712** 9 Cultural Distance 3,88 2,29 0,118 -0,024 0,022 0,271** 0,166 0,039 0,238** 0,151 ** Correlation is significant at the 0,01 level (2-tailed); * Correlation is significant at the 0,05 level (2-tailed).

(46)

44

customers’ demand and to stay ahead of the competition. This leads to technological and demand uncertainty for the same industries. Both variables will be used in a regression analysis separate from each other. However, in the full model, that contains all independent variables, “Demand Uncertainty” and “Technological Uncertainty” could cause multicollinearity issues. Potential issues caused by multicollinearity are investigated at the end of the results section. Lastly, significant correlation was found between the control variables “Number of countries”, “Number of subsidiaries” and “Cultural Distance”. Also, these relationships make sense, since a company that is operating in more countries is likely to have more subsidiaries and the cultural distance is higher given the amount of foreign countries it is operating in. However, the correlation between “Number of countries” and “Number of subsidiaries” is extremely high. This could weaken the explanatory power of the regression models. Therefore, “Number of subsidiaries” will not be included in the regression analysis.

4.2. Testing Hypotheses

First an Independent Samples T-Test is conducted to receive an indication whether subsidiaries in the core industry receive more attention than subsidiaries in the industrial periphery.

Figure 4.4. – Independent Samples-T-Test

(47)

45

The results of the Independent Samples T-Test shown in figure 4.4. reveal that subsidiaries operating in the core industry of the company receive more relative, supportive and structural attention than subsidiaries operating in the industry periphery. This gives a first indication of the flow of attention towards core industry subsidiaries. However, for the test of the first hypothesis, also the other variables need to be included and controlled for.

(48)

46

Figure 4.5. – Regression Analysis; Dependent Variable: “Relative Attention”

1 2 3 4 5 Core/Non-Core 0,229* (0,271) 0,252* (0,287) Demand Uncertainty -0,020 (0,008) -0,029 (0,009) Tech. Uncertainty -0,008 (0,040) -0,065 (0,046) Subsidiary size 0,085 (0,000) -0,027 (0,000) 0,068 (0,000) 0,066 (0,000) 0,022 (0,000) Number of countries 0,118 (0,005) 0,188* (0,006) 0,089 (0,006) 0,091 (0,006) 0,156†_(0,006) Cultural Distance 0,098 (0,055) 0,149† (0,058) 0,181* (0,065) 0,176* (0,062) 0,171† (0,064) Adjusted R2 _0,018 _0,069 _0,018 _0,017 _0,050 R2_Change _0,018 _0,051* _-0,051 _-0,001 _0,033* F value 1,283 3,088* 1,183 1,175 1,945† Number of observations 128 113 108 108 107

†_{p <0,10; *p<0,05; **p<0,01; ***p<0,001 (one-tailed if hypothesised) (Standard error in parenthesis)}

(49)

47

Figure 4.6. – Regression Analysis; Dependent Variable: “Supportive Attention”

1 2 3 4 5 Core/Non-Core 0,325*** (0,248) 0,281** (0,265) Demand Uncertainty 0,047 (0,007) 0,004 (0,008) Tech. Uncertainty 0,129† (0,038) 0,046 (0,045) Subsidiary size 0,064 (0,000) 0,020 (0,000) 0,123 (0,000) 0,112 (0,000) 0,062 (0,000) Number of countries -0,107 (0,005) -0,045 (0,005) -0,135†_(0,006) _-0,135† _(0,006) _{-0,075 (0,006)} Cultural Distance -0,008 (0,052) 0,038 (0,054) -0,008 (0,060) -0,018 (0,058) -0,001 (0,059) Adjusted R2 _-0,009 _0,087 _-0,004 _0,011 _0,063 R2_Change _-0,009 _0,096** _-0,083 _0,015 _0,052* F value 0,579 3,886** 0,889 1,308† 2,286* Number of observations 138 121 116 116 115

(50)

48

Figure 4.7. – Regression Analysis; Dependent Variable: “Structural Attention”

1 2 3 4 5 Core/Non-Core 0,285** (0,233) 0,300** (0,245) Demand Uncertainty -0,078 (0,006) -0,096 (0,007) Tech. Uncertainty 0,007 (0,035) -0,029 (0,041) Subsidiary size 0,159* (0,000) 0,106 (0,000) 0,265** (0,000) 0,252**(0,000) 0,206* (0,000) Number of countries -0,046 (0,005) 0,047 (0,005) -0,034 (0,005) -0,024 (0,005) 0,032 (0,005) Cultural Distance 0,036 (0,047) 0,066 (0,051) 0,088 (0,057) 0,061 (0,055) 0,085 (0,055) Adjusted R2 _0,003 _0,081 _0,038 _0,033 _0,103 R2_Change _0,003 _0,078** _-0,043† _-0,005 _0,070** F value 1,153 3,637** 2,150† 1,982† 3,187** Number of observations 138 120 115 115 114

(51)

49

indicates that the distinction between “Core/Non-Core Industry” adds the most explanatory power to the models.

Since the variable “Core/Non-Core Industry” appears to have the most significant influence on all dependent variables tested in this paper, a regression analysis with only the variable “Core/Non-Core Industry” and the respective dependent variables is performed.

Figure 4.8. – Regression Analysis; Independent Variable: “Core/Non-Core Industry”

Relative Attention Supportive Attention Structural Attention

Core/Non-Core 0,205 (0,257) 0,334*** (0,223) 0,286** (0,218)

Adjusted R2 _0,034 _0,104 _0,076

F value 5,157 15,669*** 11,305**

Number of observations 118 126 125

(52)

50

Since the independent variable “Technological Uncertainty” appears to be marginally significant when regressed on the dependent variable “Supportive Attention”, a regression analysis with only the independent variable “Technological Uncertainty” and the dependent variables is executed. This investigates whether the control variables might dampen the effect of the independent variable on the dependent variables.

Figure 4.9. – Regression Analysis; Independent Variable: “Technological Uncertainty”

Relative Attention Supportive Attention Structural Attention

Technological Uncertainty 0,048 (0,152) 0,133†_(0,154) _{0,029 (0,039)}

Adjusted R2 _-0,006 _0,010 _-0,008

F value 0,278 2,177† _0,097

Number of observations 120 121 113

Figure 4.9. shows that only for the dependent variable “Supportive Attention”, the model becomes marginally significant with “Technological Uncertainty” as marginally significant independent variable. Compared to model 4 of the regression analysis with “Supportive Attention” and all the control variables, this model has a slightly lower explanatory power. The other models with “Relative Attention” and “Structural” Attention” as dependent variables are still insignificant.

4.3. Multicollinearity

(53)

51

thereby lies below the common threshold of 10 and the more conservative threshold of 4 (O’Brien, 2007). In addition, the Condition Indices for all models were calculated. For all models, these indices never exceed a value of 12, which lies below the threshold of 15 (Grewal, Cote & Baumgartner, 2004). Thus, it is possible to conclude that multicollinearity is not an issue in this analysis.

4.4. Alternative Explanation of the Core/Non-Core Industry

Effect

To test whether the significant difference of attention core and non-core industry subsidiaries receive could also be explained by other factors, an Independent-Samples-T-Test was performed with several variables that could also explain this effect. The results of this analysis are depicted below:

Figure 4.8. – Independent Samples-T-Test; Test variable: “Core/Non-Core Industry”

# Variable t df Sig. (1-tailed) Mean Core s.d. Core Mean Non-Core s.d. Non-Core 1 Demand Uncertainty 1,551 121 0,062 65,13 18,38 59,81 19,25 2 Technological Uncertainty 3,225 121 0,001 3,92 3,96 2,01 2,60 3 Subsidiary Size 2,072 126 0,020 263,57 496,93 125,13 241,24 4 Number of countries -2,238 124 0,014 14,89 18,76 24,19 25,87 5 Cultural Distance 0,539 126 0,270 3,81 2,31 3,60 2,10

(54)

52

countries they are operating in. This leads to the following conclusions: The “Subsidiary Size” could be a potential alternative explanation for the industry effect found. “Subsidiary Size” reflects the concept of subsidiary weight (Bouquet & Birkinshaw, 2008a). However, in the previous regression analyses, “Subsidiary Size” lost its importance when the variable “Core/Non-Core Industry” got added to the models. This indicates a better fit of the variable “Core/Non-Core Industry” in the models. Hence the initial explanation that the industry of the subsidiaries influences the amount of attention they receive still holds. Moreover, the finding that core-industry subsidiaries have to face more technological uncertainty is very interesting. This observation entails that the concept “Technological Uncertainty”, which was previously found to be only in one case marginally significantly, plays a subtle role in influencing the amount of attention subsidiaries receive. However, these findings are tentative as they are only based on these exploratory analyses and were not tested strictly.

4.5. Industry Concentration

(55)

53

1) Attention data is available for three or more subsidiaries from the same company in the dataset

2) The subsidiaries from the same company operate in at least two different industries

3) There is industry data available for at least 85% of all subsidiaries of the MNE

Conditions 1) and 2) are designed to make sure that the headquarter attention subsidiaries receive can be compared according to the differences in industry concentration. Condition 3) was chosen due to the impact missing data has on the results of concentration. A high share of missing data has the potential to deteriorate the construction of a concentration variable.

(56)

54

Figure 4.10. – Concentration Analysis

MNE Industry (NACE) Concentration Standardized Concentration Relative Attention Supportive Attention Structural Attention A J – Information and Communication 0,09 197,2 3,00 5,50 4,5

G – Wholesale and Retail trade 0,037 81,07 4,67 6,00 5,80

C –Manufacturing 0,018 39,44 5,00 5,00 4,80

F – Construction 0,001 2,19 3,67 5,67 5.60

B G – Wholesale and Retail trade 0,137 152,97 3,00 2,33 2,00

C – Manufacturing 0,111 123,94 6,00 6,33 5,40

C J – Information and Communication

0,669 121,9 3,33 6,33 4,80

M – Professional, scientific and technical activities

0,008 1,46 - 4,00 2,80

D C – Manufacturing 0,16 117,65 2,00 5,67 4,00

M – Professional, scientific and technical activities

0,16 117,65 4,67 4,67 4,60

E C – Manufacturing 0,226 165,23 3,00 4,83 4,10

F G – Wholesale and Retail trade 0,237 152,55 4,67 3,00 2,40

C - Manufacturing 0,117 75,31 3,54 4,17 4,00

G C - Manufacturing 0,179 150,27 4,33 5,33 3,20

(57)

55

weight of the whole business line. On the other hand, the same MNE has only very few subsidiaries operating in Professional, Scientific and Technical Activities. Here it would be expected to find less attention received from headquarters, given the lower cumulative weight of subsidiaries in that industry.

A correlation analysis of the measurement for standardized concentration and the three dependent attention variables should show whether there is a relationship between “Industry Concentration” and the three attention vartiables.

Figure 4.11. – Correlation Analysis with Industry Concentration

# Variables Relative Attention Supportive Attention Structural Attention

1 Industry Concentration 0,123 0,249 0,118

** Correlation is significant at the 0,01 level (2-tailed); * Correlation is significant at the 0,05 level (2-tailed).

(58)

56

4.6. Hypotheses Summary

After analysing the data and testing the hypotheses, the conclusions are summarized in figure 4.12.

Figure 4.12. – Summary of hypotheses

Hypothesis Test Results Confirmed/

Not Confirmed

1a Core/Non-Core Industry

- Subsidiaries operating in the core industry of the MNE do not receive more relative attention from headquarters than subsidiaries operating in non-core industries (industrial periphery) of the same organization.

Not Confirmed

1b - Subsidiaries operating in the core industry of the MNE receive more supportive attention from headquarters than subsidiaries operating in non-core industries (industrial periphery) of the same organization.

Confirmed

1c - Subsidiaries operating in the core industry of the MNE receive more structural attention from headquarters than subsidiaries operating in non-core industries (industrial periphery) of the same organization.

Confirmed

2a Industry Concentration

- No significant positive relationship between (2a) relative attention, (2b) supportive attention and (2c) structural attention with industry concentration was found.

- Insignificance could have been caused by the lack of data

Not Confirmed

2b 2c

3a Demand Uncertainty

- Subsidiaries in a network of a diversified organization do not receive more (3a) relative attention, (3b) supportive attention, (3c) structural attention from headquarters when they are operating in an industry characterised by high levels of demand uncertainty than subsidiaries of the same organization operating in an industry characterised by low levels of demand uncertainty.

Not Confirmed 3b 3c 4a Technological Uncertainty

- Subsidiaries in a network of a diversified organization do not receive more relative attention from headquarters when they are operating in an industry characterised by high levels of technological uncertainty than subsidiaries of the same organization operating in an industry characterised by low levels of technological uncertainty.

Not Confirmed

4b - Subsidiaries in a network of a diversified organization receive more supportive attention from headquarters when they are operating in an industry characterised by high levels of technological uncertainty than subsidiaries of the same organization operating in an industry characterised by low levels of technological uncertainty.

Confirmed

4c - Subsidiaries in a network of a diversified organization do not receive more structural attention from headquarters when they are operating in an industry characterised by high levels of technological uncertainty than subsidiaries of the same organization operating in an industry characterised by low levels of technological uncertainty.

(59)

57

DISCUSSION

After presenting the results of the data analysis, the paper will now discuss the implications of the findings. In the following paragraphs, firstly the implications of the findings regarding the internal, structural factors are evaluated, followed by a discussion of the findings regarding the external, environmental factors influencing headquarters’ attention distribution.

5.1. Internal Factors

In the theory section, this thesis proposed that the managerial mindsets, cognitive frames and the cumulative weight of an industry within an MNE have an influence on the attention distribution of executive managers. The analysis revealed that subsidiaries operating in the core industry of an MNE receive more supportive and structural attention than subsidiaries, which operate in non-core industries of the same MNE. In specific, core-industry subsidiaries receive more information and the headquarter shares best practices with them. Moreover, core-industry subsidiaries are more involved in strategy-development and decision-making processes that affect the whole MNE. In addition, headquarter executives have more often face-to-face contact with employees from core-industry subsidiaries.