___________________________________________________________________________ Master Thesis:
Organizational structure of MNEs: A determinant for financial performance?
___________________________________________________________________________
Julia Johanna Kronenberger Student number: 1071223
Discipline of study: M.Sc. in Business Administration -International Management Submission date: January 31, 2015
Supervisor: Dr. Lori DiVito Second supervisor: Dr. Ilir Haxhi
Statement of Originality
This document is written by Student [Julia Johanna Kronenberger] who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2
1. Introduction ... 4
2. Literature Review ... 4
2.1. Different types of organizational structures... 4
2.1.1. Functional Structure ... 5 2.1.2. Divisional Structure ... 6 2.1.3. Matrix Structure ... 8 2.1.4. Hybrid Structure ... 9 2.2. Underlying Theory ... 11 2.2.1. Information Asymmetry ... 11 2.2.2 Agency Theory ... 11
2.3. The Multinational Enterprise ... 13
2.4. MNE’s Stakeholders ... 14
2.5. MNEs’ strategies for serving internal stakeholders ... 15
2.5.1. Worldwide learning ... 15
2.5.2. Global Integration ... 16
2.5.3. National Responsiveness ... 17
2.6. Organizational structure as measure to achieve MNEs strategies ... 18
3. Research Gap and Research Question ... 20
4. Defining Hypothesis ... 22
5. Research Design ... 22
5.1. Data Sample ... 22
5.2. Sample Selection and Method ... 23
6. Theoretical Framework ... 25
7. Analysis ... 27
7.1. Multiple Linear Regression Analysis ... 28
7.1.1. Reliability and Validity ... 33
7.2. PROCESS Analysis ... 33
8. Discussion ... 37
8.1. Literature Implications ... 39
8.2. Managerial Implications ... 40
8.3. Strengths and Limitations ... 41
9. Conclusion ... 42
References ... 44
3 Abstract:
In this work I test whether organizational structures of Multinational Enterprises (MNEs) have an impact on their financial performance and if so which of the four organizational structures (functional, divisional, matrix or hybrid structure) leads to the best financial output and therefore should be chosen in a preferred manner. First, the different organizational structures and their underling theories are going to be explained, followed by MNE’s essential theory. Then I run a regression analysis as well as a mediator analysis to answer the research question if MNEs perform better financially when their management chooses a certain organizational structure in order to serve their internal stakeholders. I find that the hybrid structure delivers the highest financial performance. The explanation could be found in the fact that with a hybrid structure, MNEs are either able adapt to local markets or focus on correct usage of information flows in those enterprise areas where they see the necessity.
Key Words:
Multinational Enterprise (MNE); organizational structure: functional, divisional, matrix and hybrid; financial performance; transnational solution: worldwide learning, national responsiveness, global integration
4 1. Introduction
In our current time of continuously growing and changing markets with high competition, a
variety of different enterprises with different organizational structures exists. Every enterprise
pursues internal goals to achieve cost benefits and profit optimization. Due to these
challenges enterprises must choose the most beneficial organizational structure to be able to
compete on global markets. This correct choice is a challenging task since it depends on
many internal and external factors. More precisely, enterprises have to consider incentive
problems and information asymmetry when choosing the optimal structure.
Many scholars have previously examined the determinants for organizational structure
and factors that lead to better economic performance of Multinational Enterprises (MNEs)
extensively (e.g. Rugman & Verbeke, 1992; Bartlett & Ghoshal, 1989). However, few
scholars have yet investigated if MNEs’ success, in terms of cost benefits and profit
optimization, is also dependant on the choice of a specific organizational structure. Generally,
organizational factors which have an impact on the financial performance have not been
examined thoroughly (Capon, Farely & Hoenig, 1990). This study, therefore, focuses on
MNEs and will examine how they align their organizational structures while doing business
abroad, which means pursuing international growth strategies. Further, I will test whether
there is a preferred organizational structure and, if so, which of these are leading to the
highest economic success.
2. Literature Review
2.1. Different types of organizational structures
In many firms, hierarchies in which managers are reporting to managers of the next/higher
level are present. This structure varies in its number of levels and in the set of activities
grouped together (Harris & Raviv, 2002). Chandler (1962) describes in his book how
5 communication lines between the different administrative departments and managers and
second, the flow of information and data which runs through these authority and
communication lines. These information flows are crucial to guarantee the effective
coordination, appraisal and planning in order to achieve the goals and policies of an
enterprise. Enterprises’ resources such as physical and financial capital, as well as its personnel’s marketing, technical and administrative skills, play an important part in contributing to the goals and policies. The complexity of the different structural designs also
leads to incentive problems (e.g. in case of setting the right manager incentives to achieve
best outcomes) or information asymmetry (e.g. due to the variety of different communication
lines: information can get lost or misunderstood). These issues will be discussed in section
2.2. in detail.
The existing literature describes a variety of important organizational structures. As a
result of this I have chosen the following four structures which I will describe in the next
section, as follows: 1. Functional Structure, 2. Divisional Structure, 3. Matrix Structure and 4.
Hybrid Structure. Organization charts are used to represent the different structures visually.
2.1.1. Functional Structure
In a functional hierarchy, activities which are related to particular functions are organized into
different departments (Harris and Raviv, 2002). The hierarchy of the functional structure is
organized along functional lines, which illustrates how the functional structure is a simple
structure in which every single manager is in charge of a separate function (Harris and Raviv,
2002).
Several authors, such as Maskin, Qian and Xu (2000) and Sloan (1963), use the Ford
Motor Company, before the Second World War, as a classical example of the functional
6 departments, for example, sales, manufacturing, research, distribution and personnel. The
distribution department in a functional structure would, for example, coordinate all activities
for the produced products. Maskin, Qian and Xu (2000) point out that the number of
departments were responsible for different tasks which were complementary to each other.
These separations lead to the independence of single departments. The structure of the
functional form therefore enacts clear, separated and easy controllable functional areas.
Potential benefits of the functional structure are mainly exploitation of scale
economies (Maskin, Qian and Xu, 2000). The decision of geographical dispersion, meaning a
decision into new types of functions, requires a central office and multi-departmental
structure. This is better described as the emergence of the divisional organizational structure,
as follows (Harris and Raviv, 2002).
Figure 1: Functional structure (Galbraith, 1971).
2.1.2. Divisional Structure
The divisional form structures the activities of the organization on the basis of divisions and
delegates control over the resources needed to create economic value to the stakeholders.
These divisions are either defined by product or geography rather than in a functionally sense
7 Further, the divisional structure provides new systems for capturing information,
allocating recourses and controlling behavior (Roberts, 2007). Harris and Raviv (2002)
summarize several authors’ statements, explaining that the choice for this structure is driven by the relative importance of coordination of functional activities (within a product line) and
economies of scale (combining similar functions across product lines).
According to Roberts (2007), an advantage of the divisional structure is its ability to
coordinate and motivate a large number of staff performing complex interrelated activities
among the various functions in different locations. A disadvantage of the divisional structure
is the separation of the different functions by divisions, resulting in diseconomies of scale, for
example if all products are distributed by a central distribution department (Harris & Raviv,
2002).
Managers of organizations must trade off the advantages and disadvantages of the functional and the divisional structure for their decision making (Harris & Raviv, 2002). The
divisional structure provides a better system of managerial incentives due to the fact that
middle-level managers’ performance can be compared internally and all managers of the one
level face similar industry conditions (Maskin, Qian & Xu, 2000).
8 2.1.3. Matrix Structure
The matrix structure is a mixture of the functional and the divisional structure which
optimally combines the advantages of both (Harris & Raviv, 2002). Matrix structures imply
dual or multiple authorities, also known as a “two manager” situation (Knight, 1976). As a
result, managers are influenced by two factors and coordination through lateral relationships
which enhances cross-departmental boundaries (Knight, 1976).
In the early 1980’s, when enterprises were redefining their strategies and
reconfiguring their operations due to changing external factors such as the growing
globalization of markets, increasing competition or increasing life cycles, many enterprises
were faced difficulties of strategy and structure (Bartlett & Ghoshal, 1990).
The only solution was to create a new structure - the matrix structure - which enabled
multiple, simultaneous management capabilities. In this structure authority is being shared
between project and functional managers as well as individuals having roles which involve
dual reporting (Galbraith, 1971). Its parallel reporting relationships lead to functional,
product and geographic management groups. The various information channels allow the
organization to capture and analyze complexity that it is facing in the economy (Bartlett &
Ghoshal, 1990).
Bartlett and Ghoshal (1990) state that the matrix structure proved manageable
especially in the international context. However, the authors also critique this structure due to
the fact that dual reporting leads to conflicts and confusion. The complex system of reporting
channels are also influenced negatively by factors of distance as for example language,
culture or time (Bartlett & Ghoshal, 1990). The authors further describe a phenomenon which
can be observed when enterprises are getting more complex is that managers focus more on
managing people and processes instead of concentration on organizational strategy and
9 Figure 3: Matrix structure (Galbraith, 1971).
2.1.4. Hybrid Structure
In the 1980s the hybrid structure evolved as a new form of organizational structure. From
then on enterprises were able to react better to customer needs through greater flexibility and
speed (Lentz, 1996). The main characteristics of a hybrid structure are that it decentralizes
decision-making to the operating units and enables the centralization of administrative
functions to the corporate staffs (Lentz, 1996). Hybrid structures focus on a balanced
approach between customer focus and economies of scale. On the one hand, the individual
operating units act like small companies, for example when dealing with customers (similar
to divisional structures). On the other hand, they come under control of the enterprise, such as
when dealing with cost issues and strategic direction (similar to functional structures) (Lentz,
1996). A crucial characteristic of a hybrid organization is that operating units become totally
10 of economies of scale and the integration of operating units to one corporate identity. Hybrid
structures learn how to achieve both goals simultaneously (Lentz, 1996).
Dessein, Garciano and Gertner (2006) describe the existence of a typical design issue
in the decision about which activities are going to be integrated and which are going to be
held on the specific divisions. Hybrid structures require a management which is able to
control activities on functional level as well as on business unit level. Managers of business
units try to create value through the standardization of activities whereas managers of
business unit level try to create value through alignment of activities to maximize their
profits. Because managers pursue different interests, concrete decision making of the
standardization of numerous activities is not easy. Therefore, it is important to create
incentive systems and authority structures to allow the development of synergies out of
standardizations (Dessein, Garicano & Gertner, 2006).
11 2.2. Underlying Theory
Enterprises pursue serving customers’ demands while keeping the costs as low as possible.
These two factors stand in contradiction to each other and challenge enterprises in their
decision making as the required information for optimal decisions is distributed to many
different individuals and expensive to transfer (Brickley, Smith & Zimmerman, 2007). The
following sections (2.2.1. and 2.2.2.) deal with the two most important theories in this
context.
2.2.1. Information Asymmetry
Information asymmetry describes the situation in which two or more parties make use of
different information during the fulfillment (precontractual) or the completion of a task or
contract (postcontractual). Due to these differences, issues of misunderstanding and
misinforming arise (Brickley, Smith & Zimmerman, 2007). Most frequently, asymmetric
information is studied in the context of principal-agent theory which shall be described in the
following section 2.2.2.
Harris, Kriebel and Raviv (1982) see the phenomenon of asymmetric information in
firms from another perspective: different individuals perform or manage various tasks for
which they own special information or expertise. Typically, other individuals as well as the
top management do not have access to that specific information. Furthermore, individuals can
make use of exclusively available information to their private benefits, which might not
converge with the objectives of the company, and therefore find it disadvantageous to
disclose this information freely. (Harris, Kriebel & Raviv, 1982).
2.2.2 Agency Theory
Manager incentives and information flows play a crucial role for the choice of the right
12 enterprise’s organizational structure. They also argue that the design of incentive schemes and the allocation of decision rights become interlinked. This phenomenon can be observed
often in the principal-agent-theory: This attempt formulates organizational problems as a
problem of unequal distributed information, between the principal and the agent, in which the
principle acts on behalf of the agent. In this relationship, achieving individual goals are
foremost and make it difficult for the principal to monitor the agent and to set up the
incentives (Schreyögg, 2008; Shapiro, 2005).
In order to gain overall economic success, enterprises depend on their employees,
especially on their managers, who are leading their division to the highest possible success.
Success and efficient organizational design create economic rents, which tend to be achieved
by their managers. Therefore, the right incentives must be set to achieve those economic
rents. Maskin, Qian and Xu (2000) show that managers’ performance and their information differs among various organizational structures. The right phrasing of incentives is therefore a
vital prerequisite.
McAfee and McMillan (1995) describe organizational diseconomies of scale as a
particular form of an influencing negative factor: managers in enterprises try to turn decisions
to their interests, whereby private information about certain abilities provides them with
increasing power. They use these benefits to gain personal rents. McAfee and McMillan
(1995) further show, how those information asymmetries between the single managers can
lead to inefficiencies, although they act in terms of the enterprise. In order to counteract those
inefficiencies or to keep them as low as possible the principle can use a number of measures:
for example to create controls, threat sanctions, extend the information system or create
incentives for the agent in such a way that deviation of his goals become unlikely, which is
13 Brickely, Smith and Zimmerman (2007) conclude that designing firms (also in terms
of organizational structure choice) has to be performed in a way that all decision makers have
the necessary information available which enables them to make good decisions as well as
the incentives to use that information efficiently.
2.3. The Multinational Enterprise
In the current literature and the scholarly context a number of definitions and perceptions
exist, which describe the complex terminology of organizations. Mintzberg (1979) describes
the structure of an organization as the simple sum of ways through which the organization
splits its labor into different tasks and where coordination is captured among those tasks.
Living in a continuously changing world forces firms to constantly modify their
growth strategies to keep up with the emerging competition. Through this phenomenon the
so-called Multinational Enterprise (MNE) arises. Buckley and Casson (2009) define a MNE
as a firm which owns and controls its activities in more than one country. This diversity of
activities is a more complex and challenging task in terms of managing these enterprises and
their vast resources, compared to firms which are only located in a single environment
(Geringer, Beamish & daCosta, 1989). Rosenzweig and Singh (1991) state that a MNE can be
described as: (a) a single firm which operates in a global environment and controls it’s widely
spread subsidiaries, or (b) a set of organizations which operate in separate national
surroundings. Mudambi and Swift (2011) show that MNEs must accomplish two things to be
able to succeed globally: first, local adaption to each context in which they operate and
second, knowledge and competencies leveraging across the contexts as well as integrating
14 In the following section I will present the underlying theory and will link these
theories to the attributes of an MNE and its choice of choosing the optimal organizational
structure.
2.4. MNE’s Stakeholders
Taking into consideration that MNEs are operating in many different countries around the
world, enterprises have to deal with complex structures of different stakeholders to achieve
their goals. A stakeholder is defined as an individual or a group of individuals that has any
interest in a business. Businesses can be understood as systems in which value is created for
its different stakeholders (Freeman, 2010). Stakeholder theory states that enterprises have to
be responsible for their actions, objectives or policies also regarding to their stakeholders
because enterprises are able to affect its stakeholders, and vice versa. Fassin (2012) describes
this phenomenon as stakeholder reciprocity which also must be taken into account. Usually
the stakeholder structure is quite heterogeneous in terms of different interests. For example,
customers expect different pricing strategies then employees who are more interested in
attractive working conditions.
R. Edward Freeman was the first to address the concept of stakeholder theory by
identifying and modeling different groups of stakeholders within corporations. Freeman
(2010) states that compared to the traditional view of a corporation (shareholder view), in
which only the owners or shareholders play an important role for its value creation, the
stakeholder view, which incorporates the internal (e.g. owners, customers, suppliers and
employees) as well as the external environment (e.g. government, competitors), also must be
considered.
In particular, multinational enterprises and their different subsidiaries are forced to
15 power and the measures to primarily serve internal stakeholders by choosing several
strategies. Under these circumstances Rugman and Verbeke (1992) define several strategies
based on the work of Bartlett and Ghoshal (1989) to overcome MNEs challenges.
2.5. MNEs’ strategies for serving internal stakeholders
Due to new worldwide pressures, companies must rethink their traditional strategies and
processes in which the choice of organizational structures plays an important role (Rugman &
Verbeke, 1992). The authors take a deeper look at the fit between environment, strategy and
structure which is solved with worldwide learning, national responsiveness and global
integration. They further state that the traditional use of firm specific advantages (FSAs) and
country specific advantages (CSAs) needs to be modified:
MNEs cannot only rely on FSAs developed in their home country due to national responsiveness as a key factor for competitive success. Location
bound-FSAs need to be developed in each country where specific need exists
for national responsiveness; and
CSAs in a specific host country can contribute to the development of new FSAs; and
Internalization advantages depend on a company’s transactional FSAs to operate foreign subsidiaries (Rugman & Verbeke, 1992).
2.5.1. Worldwide learning
MNEs can be described as knowledge-sharing networks in which knowledge and learning
can be understood as the root of gaining and sustaining competitive advantage (Foss &
Pedersen, 2004). This competitive advantage can be further explained as a firm’s greater
ability to transfer, create, integrate and deploy certain kinds of knowledge more efficiently
16 organizational structure issues to knowledge processes because existing literature considers
them only separately, never combined. In addition, the authors describe that MNEs create
FSAs mostly through intangible assets, in other words, their core resources do not physically
exist (e.g. intellectual property, know-how or reputation). To understand MNEs and their
operations, it is crucial to understand the subject of organizing the enterprise and its
knowledge processes (Foss & Pedersen, 2004).
Several authors like Mudambi and Navarra (2004) highlight the importance of
knowledge transfers due to increasing subsidiaries responsibilities and therefore more
bargaining power which in reverse leads to overall increasing performance.
Headquarter-subsidiary relations can be characterized as a principal-agent relationship in which knowledge
flows crucially depend on the motivation of the subsidiary to acquire knowledge and its
sharing. Subsidiary managers can abuse this power due to the fact that they pursue their own
goals and can increase their share of resources provided by the headquarters and in turn
increases performance. It is therefore essential to create incentives and control mechanisms to
avoid this exploitation (Mudambi & Navarra, 2004).
2.5.2. Global Integration
Bartlett and Ghoshal (1989) describe this phenomenon also with the lack of global efficiency.
This arsis due to the fact, that companies have changed their way of doing business from a
classical local model to a more global one. Firms always pursue possibilities to catch
economies which are available abroad by aligning their products on their manufacturing
operations. The forces that lead such alignments are present in the whole bandwidth between
industries, where structural changes through discontinuity are necessary to industries where
new possibilities of creating profitable business arise. If an enterprise is leading its
17 (global) entity. Global integration forces enterprises to setting up their business in a global
integrated way in order to be able to earn efficiency benefits.
Acting more globally is not as easy as it seems due to several factors which influence
the way of operating as well as the performance of an enterprise. To gain further economic
success and competitive advantage it is therefore important to gain benefits from global
integration as well as from national responsiveness (Bartlett & Ghoshal, 1989).
2.5.3. National Responsiveness
Bartlett and Ghoshal (1989) describe in their book the need for responsiveness as the force
for local differentiation where short- and medium-trends play an important role while
operating abroad. Barriers and countertrends have forced managers of MNEs to be more
sensitive to differences of international and host country markets. The development of
products which can be sold worldwide does not necessary lead to economic success due to
the fact that consumers reject homogenized products. This fact leads to openings for
competitors who are willing to meet the local needs. Therefore it is important to think about
local responsiveness as well as to serve the local customers with locally differentiated
products and services (Bartlett & Ghoshal, 1989).
Rugman and Verbeke (1992) suggest that local responsiveness is necessary in order to
generate location bound FSAs (LB-FSAs) which cannot be transferred easily. An enterprise’s
competitive advantage in a host country or restricted region is mainly consisting of LB-FSAs.
Such LB-FSAs can be developed by serving local customer or specific market needs as well
as government requirements.
Additionally it can be said that the strategies serve internal stakeholders by enabling
the MNE to create more financial value by a) cost beneficial knowledge sharing in terms of
18 global integration and c) adapting to local markets via national responsiveness to perfectly
serve local markets needs and to be able to gain economic rents from it.
2.6. Organizational structure as measure to achieve MNEs strategies
The abovementioned measures serve MNEs internal stakeholders by creating a higher
financial value. As a matter of fact, the different organizational structures foster different
strategies as they set different incentives within an MNE. The following strategies can be
pursued by choosing the subsequent organizational structure:
Functional organizational structure for worldwide learning (e.g. due to its easy, simple and controllable structure MNEs are able to exploit economies of scale); and
Divisional organizational structure for national responsiveness (e.g. due to its decentralized decision processes MNEs are able to perfectly align their products and
services to local, regional demands. Furthermore, due to higher flexibility and internal
control, MNEs are able to react faster towards changing markets); and
Matrix structure for handling both factors with the same importance (customers and worldwide learning) and additionally fostering global integration (e.g. due to shared
authority MNEs are able to capture and analyze the complexity of this combination);
and
Hybrid structure for finding the perfect balance between national responsiveness and worldwide learning (e.g. MNEs are able to generate synergies in single divisions and
able to make use of local knowledge in combination with economies of scale of a
large enterprise).
As the functional structure provides global functional units that act in different
countries simultaneously, R&D activities and other knowledge is held in a global unit. This
19 achieve high economies of scale due to the fact that research activities are never performed
parallely.
On the other hand, the divisional structure provides units per country which are
independent from each other. This enables the several divisions that are typically structured
by regions to adapt to the local demands perfectly and therefore to earn higher economic
rents. However, knowledge gaining across different divisions is not fostered due to the fact of
lacking cooperation incentives, which is why in this organizational structure the economies of
scale within which, for example, the R&D section cannot be earned.
The matrix structure encases both factors and should therefore enable a MNE to make
use of both advantages at the same time. They can therefore be considered as global
integrated. However, since in the matrix structure the power is basically split between
functions and divisions, the firm cannot adopt to those factors that lead to higher financial
success.
The hybrid structure is, as mentioned, closely related to the matrix structure, but also
allows choosing a functional structure in some parts of the firm and a divisional structure in
other parts of the firm. Therefore, the MNE can decide for which unit it wants to focus more
on national responsiveness to better serve the markets’ needs or to focus, for example, on worldwide learning in order to achieve positive scale effects in the R&D section. It provides
the possibility to make an MNE globally integrated and constantly adapting to gain
advantages.
Considering this, it is interesting to see whether the choice of an organizational
structure has an impact on the MNEs financial performance and, if so, which choice of
20 3. Research Gap and Research Question
Due to fact that MNEs have dispersed subsidiaries all over the world which need to be aligned to a firm’s global strategy, I have linked the growth strategies to the choice of their organizational design in the previous section. As discussed earlier financial performance is a
strategic goal of every MNE and used to serve its internal stakeholders. Therefore, as a
further step, it would be interesting to discover which of the chosen organizational structures
in connection to the desired strategy (worldwide learning, global integration and national
responsiveness) achieves a better financial performance.
Determinants for organizational design and factors that lead to better financial
performance of MNEs have been examined extensively in present literature: MNEs financial
performance can be influenced trough several factors. Environmental variables (e.g., industry
concentration, size and growth) and strategy variables (e.g., market share and R&D) have
proven to have a significant, positive impact on MNEs financial performance (Capon, Farely
& Hoenig, 1990).
However, few scholars have investigated if MNEs success is also depending on the
choice of a specific organizational structure. The framework by Capon, Farely & Hoenig
(1990) shows that organizational factors also determine MNEs financial performance. Lenz
(1981) describes that financial performance is determined by the adaptation of an
organizational structure to deal with changing competitive circumstances. Further, he states
that organizational structures affect financial performance.
In their article, Lu and Beamish (2004) mention organizational design as an important
moderator of the relationship between multinationality and performance. Bartlett and Ghoshal
(1990) describe how the characteristics of an organization’s strategic decision process is
21 associated with different types of structures. Ghoshal and Nohria (1993) describe the idea of
organization theory which focuses on the “fit” between organization’s structure and management process with its environment. Furthermore, they show that the complexity of a firm’s structure must match the complexity of the environment in which it is acting. Goerzen and Beamish (2003) consider the relationship between geographic scope and performance of
MNEs and describe the differences in this relationship found by several scholars. In doing so,
they look at the elements of international asset dispersion and country environment diversity.
Scholars as Kim, Hwang and Burgers (1989) found a positive, linear relationship between
geographic scope and the performance which is consistent with the findings of other research.
Geringer, Beamish and daCosta (1989) show that a firm’s diversification as well as its internationalization strategy are important determinants which explain corporate
performance. Grogaard’s work (2012) builds on the integration-responsiveness framework by Bartlett and Ghoshal (1989) which helps to understand and analyze international strategies
and highlights the importance of aligning strategy and organizational structure to influence
performance positively. It is clear that choosing the “right” organizational structure and making the right decision of “where” and “when” to invest has an important impact on the firm’s financial performance.
A vast majority of scholars focus on the strategy that worldwide learning, global
integration and national responsiveness lead to better financial performance. Only a few
academics, however, address the choice of organizational structure as a measure to improve
financial performance. As in this work the link between strategy and organizational structure
is established, the following research question arises:
Which organizational structure leads to the best financial performance in terms of profitability and returns and what does that imply for the three strategies of an MNE, meaning worldwide learning, national responsiveness and global integration?
22 4. Defining Hypothesis
Accepting the information about MNEs, their strategies and their organizational structures in
existing literature, I define four hypotheses able to answer the abovementioned research
question. The hypotheses will be examined in an empirical analysis in section 7. This analysis
will provide the answers to the hypothesis’ validity. The hypotheses are defined as followed:
H1: MNEs perform best if they choose a functional structure to incentivize knowledge sharing globally (worldwide learning).
H2: MNEs perform best if they choose a divisional structure to better respond to local market demands (national responsiveness).
H3: MNEs perform best if they choose a matrix structure to focus on global complexity. H4: MNEs perform best if they choose a hybrid structure to balance customer focus and economies of scale resulting from global integration.
To test the hypothesis and answer the research question, the empirical analysis will be
conducted with the S&P Global 100. The procedure and analysis of the quantitative research
is explained and described in the following sections.
5. Research Design 5.1. Data Sample
As an appropriate sample for my study I chose the constitutes of the S&P Global 100 due to
the fact that I am looking at the organizational structures and financial performance of MNEs.
With this sample I want to test whether the organizational structures of enterprises around the
world can be used as a determinant of their financial performance. The following attributes of
the S&P Global 100 are therefore suitably appropriate: The S&P Global 100 is a stock market
23 Global 1200, operating in multiple countries around the world. It is weighted by
market-capitalization, measuring the performance of multinational, large-cap companies and covers
seven distinct regions and 29 countries (S&P Index Methodology, 2014).
5.2. Sample Selection and Method
To create my dataset I used Wharton Research Data Services (WRDS). At WRDS I used
COMPUSTAT from Standard and Poor’s to gather the S&P Global 100 data. Within the Index Constituents I first used the ‘Compustat Global - Index Constituents Code Lookup’ to receive the Ticker for the S&P Global 100 (I6UNK111). Following this, I added the codes to
the list and completed the query for the year 2013 which was finally generated in an Excel
datasheet. The data output provided me with the following information: company name,
company ticker (SEDOL) and number of employees.
To gather financial data I used Thomson ONE database. The use of financial data is
necessary for the performance analysis of my study which will be examined via a multiple
linear regression. With Thomson ONE and the company ticker (SEDOL) I was able to gather
the following data: debt ratio, gross income, EBIT, operating cash flow, net income, ROA,
ROE, sales and total assets.
Unfortunately Thomson ONE was not able to provide me with information about the
abovementioned MNEs organizational structures (functional, divisional, matrix and hybrid).
Due to my knowledge no database exist which provides this information. Therefore, I made
use of the businesses’ homepages and their 2013 annual reports in order to gain the required data. In those cases where I was not able to gain any information through these two sources I
contacted the enterprises personally by telephone or by e-mail.
In my thesis I conduct a quantitative experiment where I look at the impact of firms`
24 (e.g. Net Income/Total Assets). I test which organizational structure leads to a higher
financial performance in order to be able to advise if a MNE should choose a specific
organizational structure to gain a higher financial performance.
To test the correlation between the organizational structures and performance of the
particular firms I conduct a multiple linear regression analysis by using SPSS. A regression
analysis is a common and widely used statistical technique for analyzing the relationship
between a dependent variable Y (key performance indicator) and an independent variable X
(organizational structure) (Edwards, 1976). With this procedure I observe if the choice of a
certain organizational structure leads to a higher financial performance. Furthermore, I
execute a mediation analysis in SPSS using the macro PROCESS. This is used to test whether
the effect of the independent variables on the dependent variables is direct.
To be able to test the relationship between organizational structures of MNEs
(independent variable) and their financial performance (dependent variable), I defined
variables for financial performance by calculating several ratios via excel. These are
represented through six performance indicators for the year 2013: GrossMargin/Sales,
EBIT/Sales, OperatingCashFlow/Sales, NetIncome/TotalAssets, return on assets (ROA) and
return on equity (ROE). Due to the fact that I aim to test the effect not only on a specific year,
I created six more variables representing the 10 year’s average of each performance indicator. I therefore run two sets of multiple linear regressions: the first set is run with the
performance indicators of the year 2013, while the second set is run with the mean of the
performance indicators of the last ten years. Finally, the analysis of the outcome shows if a
significant correlation between the financial performance and the choice of the organizational
25 but do not have anything to do with the organizational structure, I used one mediator variable
(employees).
To test my hypothesis including the different organizational structures I recoded the
organizational structures into three dichotomous (binary) dummy variables (1 = representing
the organizational structure’s variable name, 0 = not represent the organizational structure’s variable name) in order to perform a multiple regression analysis in SPSS. The dummy
variables are: DummyFunctional, DummyDivisional and DummyMatrix, hybrid structure is
used as the baseline organizational structure and indicated when all dummies show “0”. The
regression equation appears as follows:
where PI represents the performance indicator, F the functional structure, D the divisional
structure and M the matrix structure.
6. Theoretical Framework
In order to generate a theoretical framework, I created a correlation matrix (appendix 1) from
the abovementioned variables and additional ones that might also impact the financial
performance, such as size (measured by number of employees) and financial leverage (debt
ratio). This enabled me to determine which variables can be considered as control variables,
which may be mediators, and if there are preliminary signs for rejecting or accepting the
different hypotheses.
This correlation is interpreted by the so called Pearson’s correlation coefficient (r). The correlation coefficient (r) measures the strength and direction (+/-) of a linear
relationship between two variables and ranges between +1 (perfect positive correlation) and
-1 (perfect negative correlation). Positive correlation means a positive relationship, while
26 I found that the variable ‘debt ratio’ can be considered as a control variable because there is no relationship (correlation) between the independent variables and ‘debt ratio’ (DummyFunctional: r = 0.00854, DummyDivisional: r = -0.0434 & DummyMatrix: r =
0.00625), but that there is a correlation between the control variable and the performance
indicators of the dependent variable (e.g. EBIT/Sales10yrs: r = 0.30212, ROE10yrs: r =
0.36851). Furthermore, it can be seen that there is a ranking of preferred organizational
structures: the functional organizational structure tends to be the best suitable (most positive
correlation values towards performance indicators), the matrix structure tends to be the
second best suitable (less positive correlation values towards performance indicators), and the
divisional structure tends to be the least suitable due to the most negative correlation values
towards performance indicators.
The resulting framework describes my project visually. A dependent factor (financial
performance) and an independent factor (organizational structure) are included. Furthermore,
one mediator (‘employees’) is incorporated. With the framework’s information I tested the
hypothesis, by running multiple linear regressions, in order to answer whether any type of
organizational structure is beneficial to the financial performance.
27 7. Analysis
Before analyzing my data I ran frequencies of the string variables (organizational structure) to
check my dataset on any errors or missing data. The output (Table 1) shows that there are no
errors or missing data and the number of organizations is valid, meaning it consists of 100
enterprises with different organizational structures.
Additionally, the frequency table provides first information about possible preferences
of MNEs organizational structure choice. It shows that 46 MNEs chose a divisional
organizational structure to obtain best results. 24 MNEs chose a functional organizational
structure as the most appropriate. The matrix structure is presented by 16 MNEs and the
hybrid structure by 14 MNEs. These results can be seen in the following table:
Table 1
Frequencies of organizational structures
Frequency Percent Valid percent Cumulative percent Valid Divisional 46 46.0 46.0 46.0 Functional 24 24.0 24.0 70.0 Hybrid 14 14.0 14.0 84.0 Matrix 16 16.0 16.0 100.0 Total 100 100.0 100.0 Missing System 0 Total 100
Furthermore, the descriptive analysis of the dummy variables (Table 2) shows that the
organizational structure dummies were created in a correct manner (see the mean) and that no
data is missing or wrong. As a result, the output from the abovementioned frequency table
28 Table 2
Descriptive analysis of dummy variables
DummyFunctional DummyDivisional DummyMatrix
N Valid 100 100 100 Missing 0 0 0 Mean .24 .46 .16 Standarderror of mean .04 .05 .04 Standarddiviation .43 .50 .37 Variance .18 .25 .14 Skewness 1.236 .163 1.883 Standarderror of Skewness .24 .24 .24 Kurtosis -.482 -2.014 1.578 Standarderror of Kurtosis .48 .48 .48 Minimum 0 0 0 Maximum 1.00 1.00 1.00
Based on the distribution of organizational structures within the sample, it is
interesting to find out whether the preferred organizational structures lead to optimal financial
performance, or whether the majority of those firms’ managers have made wrong decisions
and other, not so common organizational structures lead to better financial performance.
Therefore, I run multiple linear regression analysis using SPSS (upcoming section 7.1.) to see
whether there is any indication to support this first result and to be able to give managerial
advice about organizational structure choice.
7.1. Multiple Linear Regression Analysis
I ran 12 multiple linear regressions (all 12 performance indicators separately including all
three independent variables and the control variable) to find the most significant output to
further test. Due to the fact of using dummy variables for my regressions I have to choose the
29 because I do not want to rank them, the variables are coequal. This method is known to force
all independent variables (predictors) into the model simultaneously (Field, 2009).
Five cases were found in which organizational structures have a significant (p <=
0.05; resulting from a t-value > 1.96) effect on the respective performance indicators. These
were as follows: DummyDivisional on EBIT/Sales (β = -0.374; t = -2.459), DummyMatrix on EBIT/Sales (β = 0.290; t = 2.230), DummyDivisional on NetIncome/TotalAssets (β = -0.390; t = -2.629), DummyFunctional on ROE (β = -0.285; t = -1.981) and DummyDivisional
on NetIncome/TotalAssets10yrs (β = -0.299; t = -1.976).
I further found two cases in which the confidence interval was in the 90% range (p <=
0.10): DummyFunctional on EBIT/sales (β = -0.261; t = -1.892), DummyMatrix on
NetIncome/TotalAssets (β = -0.220; t = -1.673). The significance regression output (p < 0.05
= significant, p < 0.1 = fairly significant) can be found in the tables below:
Table 3
Regression model of EBIT/Sales
R R² R² Change B SE ß t 0.359 0.129 0.091 DummyFunctional -6.429 -.261 -.261** -1.892 DummyDivisional -7.853 -.372 -.374* -2.549 DummyMatrix -8.183 -.290 -.290* -2.230 Debtratio .166 -.235 .235 2.404 Statistical significance: *p<0.05; **p<0,01; **p<0,001
30 Table 4
Regression model of NetIncome/TotalAssets
R R² R² Change B SE ß t 0.321 0.103 0.064 DummyDivisional -4.125 -1.569 .390* -2.629 DummyMatrix -3.135 -1.874 .220** -1.673 Debtratio .056 .035 .158 1.601 Statistical significance: *p<0.05; **p<0,01; **p<0,001 Table 5
Regression model of ROE
R R² R² Change B SE ß t 0.201 0.040 0.000 DummyFunctional -14.706 7.422 .285* -1.981 Debtratio -.008 .150 -.005 -.054 Statistical significance: *p<0.05; **p<0,01; **p<0,001 Table 6
Regression model of 10yrs' NetIncome/TotalAssets
R R² R² Change B SE ß t 0.249 0.062 0.022 DummyDivisional -.030 .015 .299* -1.976 Debtratio -.000 .000 -.124 -1.249 Statistical significance: *p<0.05; **p<0,01; **p<0,001
31 Beta values tell the relative difference between each group on the outcome variable
(Field, 2009). By interpreting the beta values of the organizational structures, I identified
some form of pattern which was already shown in the correlation matrix: in general the betas
of all three organizational structure dummies (functional, divisional and matrix) were
negative and ranked in the order of (1) functional, (2) matrix and (3) divisional. Here,
negative betas mean that the change in the outcome variable decreases as an MNE chooses
either a functional, matrix or divisional organizational structure over the hybrid structure.
This can be seen for example by comparing the three different beta values of the
organizational structure dummies on EBIT/Sales: The beta value of the functional structure (ß
= -0.261) is lower than the beta value of the matrix structure (ß = -0.290) and again lower
than the beta value of the divisional structure (ß = -0.374). As all betas have a t-value greater
than 1.64 and the betas for divisional and matrix organizational structure generate a t-value
higher than 1.96, all results can be considered fairly significant (at least) and are therefore a
factor to answer my hypotheses. Consequently, I can say that H4 can be accepted, while the
other ones can be rejected.
Further, the example of the organizational structure dummies on
NetIncome/TotalAssets support this result: the beta value of the functional structure (β =
-0.180) is lower than the beta value of the matrix structure (β = -0.220), which in turn is lower
than the beta value of the divisional structure (β = -0.390). However, only the dummy for the
divisional and for the matrix organizational structures are in the range of a 90% confidence
interval and are therefore at least fairly significant (t-value of functional dummy is with 1,288
below the threshold of 1.64, t-value of divisional structure is with 2.629 over the 1.96
threshold for 95% confidence, and the t-value of matrix dummy is with 1.673 over the 1.64
90% confidence interval threshold). This indicates that H2 and H3 can be rejected, while the
32 In addition, three other regressions delivered results with significant betas for the
dummy variables indicating the organizational structure: The regression for the matrix
structure dummy on return on assets (one year basis) generated a significant beta -0.227
(t-value of -2.118 and therefore over the 95% confidence interval threshold of 1.96) and thus
indicates that H3 can be rejected as it leads to significantly lower performance in return on
asset terms compared to the baseline of the regression that is indicating the hybrid structure.
The regression for the functional structure dummy on equity on the other hand generated a
significant beta for the dummy representing the functional structure of 0.285 (tvalue with
-1.981 also higher than the 95% confidence interval threshold) which is a sign to reject H1 as
it is significantly lower than the baseline representing the hybrid structure. Only one
regression of those with the ten-year performance indicators delivered a significant beta: the
regression on the ten-year’s average NetIncome/TotalAssets ratio generated a negative beta
for the dummy of the divisional structure of -0.299 (t-value with -1.976 over the 95%
confidence threshold of 1.96). This indicates that H2 can be rejected, as it is significantly
lower than the baseline that is representing the hybrid structure.
In sum, it can be said that there are several significant indications to reject H1, H2 and
H3. Even, if there is also the trend that the effect of a divisional structure is the most negative,
the effect of a matrix structure less negative and that the effect of a functional structure is
even less negative then the other two. Positive betas for any organizational structure dummy
cannot be found, which is why, the baseline seems to be significant and H4 has to be
accepted.
The abovementioned results explain that hybrid organizational structures tend to be a
better choice in relation to the MNE’s financial performance. The other three organizational structures tend to have less positive impact on the enterprises financial performance and
33 7.1.1. Reliability and Validity
The reliability of the analysis was tested with R² values. Due to the limited number of
independent variables per regression I believe there is no necessity to make use of the
corrected R². The range of the R² for significant outcomes was between 0.040 and 0.129. The
regressions on EBIT/ Sales (1 year) as well as on NetIncome/TotalAssets (1 year) had a R²
greater than 0.1 and can be considered as reliable. The regressions in which the confidence
interval was 90%, the R² values were below 0.1 – this indicates, that the results are not
completely reliable, and must therefore be considered with caution.
When interpreting the significant results of the reliable regressions (in terms of R²), I
found that the dummy variables always have a negative effect (negative beta) on performance
indicators, which is a sign for acceptance of H4. Furthermore, it can be assumed that the
divisional organizational structure has the largest negative effect on the performance
outcome. This leads to the conclusion that a divisional structure should not be chosen by
MNEs (rejection of H2). Since the betas of the dummy variables for functional and matrix
structures are negative it can be assumed that the corresponding hypothesis should also be
rejected.
7.2. PROCESS Analysis
In order to verify whether the organizational structures have a direct impact on the
performance indicators or whether they are only a feature of a specific group of companies
and to further test their significance, I conducted a mediation analysis using SPSS PROCESS.
PROCESS Model 4 proved to be the most suitable for testing the mediator ‘employees’.
A mediator analysis is used to test the relationship between a predictor variable
(independent variable) and an outcome variable (dependent variable) which in turn can be
explained by a third variable (the mediator). The direct effect between the predictor and the
34 This occurs if the correlation between the dependent variable and the independent variable is
completely wiped out by including the mediator into the model. In a case close to perfect
mediation the direct effect decreases while the indirect effect increases. Mediation is shown
by basically significant indirect effect (Field, 2009). By interpreting the bootstrapped
intervals I can determine what the value of the indirect effect is. For a significant mediating
effect the confidence interval should not contain zero, but the lower (Bootstrapped Lower
Level Confidence Interval = BootLLCI) and the upper significance border (Bootstrapped
Upper Level Confidence Interval = BootULCI) should be either positive or negative.
For the five cases in which organizational structures have a significant effect (p <=
0.05) on the respective performance indicators I found that the there was no significant
mediating effect, meaning the number of employees do not have an impact on the MNEs
organizational structure choice. All five confidence intervals included zero. This can be seen
in the following data: DummyDivisional on EBIT/Sales (β = -0.4955; BootLLCI: -1.2274 and BootULCI: 0.1226), DummyMatrix on EBIT/Sales (β = 0.2324; BootLLCI: -0.6183 and BootULCI: 1.0310), DummyDivisional on NetIncome/TotalAssets (β = 0.0789; BootLLCI: 0.2712 and BootULCI: 0.4283), DummyFunctional on ROE (β = 0.1272; BootLLCI: -1.0260 and BootULCI: 1.4484) and DummyDivisional on NetIncome/TotalAssets10yrs (β = 0.0003; BootLLCI: -0.0032 and BootULCI: 0.0035).
35 Table 7
Y = EBIT/Sales; X = DummyDivisional & DummyMatrix; M = Employees; CONTROL= Debtratio
Model Summary R R² F df1 df2 p
DummyDivisonal (1) .1931 .0373 1.8214 2 94 .1675
DummyMatrix (2) .1155 .0133 .6352 2 94 .5321
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
Employees (1) .3653 .3179 -.3036 .9401
Employees (2) -.4955 -.3817 -1.2274 .1226
Table 8
Y = NetIncome/TotalAssets; X = DummyDivisonal; M = Employees
Model Summary R R² F df1 df2 p
DummyDivisonal .1512 .0229 22.453 1 96 .1373
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
Employees .0789 .1735 -.2712 .4283
Table 9
Y = ROE; X = DummyFunctional; M = Employees
Model Summary R R² F df1 df2 p
DummyFunctional .1090 .0119 1.1535 1 96 .2855
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
36 Table 10
Y = 10yrs'NetIncome/TotalAssets; X = DummyDivisional; M = Employees
Model Summary R R² F df1 df2 p
DummyFunctional .1512 .0229 2.2453 1 96 .1373
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
Employees .0003 .0017 -.0032 .0035
The two cases in which the organizational structures have a moderate effect were also
not mediated by employees. This is represented in the following data: DummyFunctional on
EBIT/sales (β = 0.2324; BootLLCI: -0.6526 and BootULCI: 0.9560), DummyMatrix on NetIncome/TotalAssets (β = -0.0098; BootLLCI: -0.2263 and BootULCI: 0.0035).
Table 11
Y = EBIT/Sales; X = DummyFunctional; M = Employees; CONTROL= Debtratio
Model Summary R R² F df1 df2 p
DummyFunctional .1109 .0123 1.1827 1 95 .2796
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
Employees .2324 .3883 -.6526 .9560
Table 12
Y = NetIncome/TotalAssets; X = DummyMatrix; M = Employees
Model Summary R R² F df1 df2 p
DummyMatrix .0484 .0023 .2252 1 96 .6362
Indirect effect of X on Y effect Boot SE BootLLCI BootULCI
37 The PROCESS analysis did not confirm my assumption that the number of employees
could have an impact on MNEs financial performance and is simultaneously dependent on
the choice of an organizational structure. Nevertheless, I found that the former trend
(preference of different organizational structures) identified in the multiple linear regressions
did not change during the PROCESS analysis with the mediators. For that reason, I can
assume that my methodological approach is correct and that these results could be used for
further interpretation.
8. Discussion
Due to the fact that MNEs operate in a global environment where they own and control their
widely spread subsidiaries, various complex stakeholder structures arise. These different
stakeholder groups influence and determine the basic goals of MNEs. Several authors like
Ghoshal (1989), Rugman and Verbeke (1992) or Foss and Pedersen (2004) define a number
of strategies (worldwide learning, global integration and national responsiveness) which can
be used by MNEs to serve its different stakeholders.
My work focuses on MNEs’ internal stakeholders (e.g. owner, employees or
suppliers) and those who they have to deal with to achieve their goals. Furthermore, the
management, which should be using this work as advice, has the power and measures to serve
these stakeholders by choosing a certain aforementioned strategy. Moreover, organizational
structures serve as a measure to coordinate information flows and incentives within an
organization. These different organizational structures enable the management to direct the
various information flows and incentives into different directions of the company. For this
reason I was able to link these organizational structures to the different strategies (section
38 Hence, I was able to generate my research question to ask what organizational
structure in combination with what strategy leads to the highest financial success. Based on
the findings from the multiple linear regressions, I can assume that the hybrid structure, with
its possibility to adapt to local demands as well as to adapt to correct information flows, is the
most beneficial organizational structure in terms of MNEs’ financial performance. Based on
these findings the question arises why so little MNEs choose hybrid organizational structures
in order to gain better financial performance.
One possible reason for MNEs not to choose a hybrid organizational structure could
be the fact described by the authors Goerzen and Baemish (2003). They state that this
complex structure must match the complexity of the environment in which the enterprise is
operating (Goerzen and Baemish, 2003). Further, it requires a high number of employees in
charge with knowledge about for example foreign cultures, legal requirements or foreign
languages and in turn increases the cost of the enterprise. Furthermore, MNEs with hybrid
organizational structures have to deal with high transaction or communication costs, for
example due to different managerial interests based on information flows and incentive
systems.
Another reason not to choose a hybrid organizational structure could be the maturity
of an MNE and its managers in charge, based on the fact that this structure is a relatively new
and unexplored one which might lead to uncertainty (e.g. due to lack of knowledge).
Therefore, managers rather stick to old, familiar patterns as the well-known functional or
divisional organizational structure.
Additionally the question arises, why H4 in particular is accepted even though the
analyzed enterprises do not seem to prefer it? An important fact could be that the hybrid
39 different areas in which either local responsiveness (followed by a divisional structure) or
worldwide learning and other centralization strategies (followed by a functional structure) are
required. These areas provided with different organizational structures lead to one perfectly
aligned structure with right incentive flows and the capability to always respond optimally to
market’s needs.
The aforementioned factor could at the same time be the reason why H1, H2 and H3
are rejected: None of the connected organizational structures offers the possibility of such a
thorough alignment. A functional or divisional structure is always applied within the whole
enterprise, even in areas where it might not be optimal. The matrix structure requires both
ways of reporting towards middle layer managers, but usually, one dimension (functional or
divisional) is overruling the other, and managers get incentivized to perform in favor of the
dominating dimension. In conclusion it can be said that the concentration on only one
dimension (local responsiveness or worldwide learning representing worldwide optimal
information flows) has negative outcomes compared to an aligned mixture of strategies that
the hybrid structure is offering.
Further, the test for mediation showed that the size of a company, measured by its
number of employees, does not depend on the organizational structure of a company and
consequently does not affect the financial performance at the same time. This could be a sign
that the perfect choice of an organizational structure does not depend on the sheer size of a
company, and more so (as previously stated), depends on the complexity of information flows
and incentives a MNE is operating with.
8.1. Literature Implications
This study enriches and complements the existing body of knowledge in organizational