The relation between international ambidexterity and firm performance, the degree of internationalization as a moderator

(1)

The relation between international ambidexterity and

firm performance, the degree of internationalization as

a moderator.

Author: Dina Shakirovna Valeeva

Student number: 11406224 Date of submission: 23 June 2017

Version: Final

Supervisor: Dr. Carsten Gelhard

University of Amsterdam, Amsterdam Business School

(2)

Statement of originality

This document is written by student Dina Shakirovna Valeeva 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.

(3)

List of tables and figures

Figure 1: Theoretical framework... 21

Figure 2: Moderating effect of high and low pace of internationalization on the relation between international ambidexterity and firm performance ... 42

Table 1: Descriptive Statistics - Industry ... 29

Table 2: Descriptive Statistics – Number of Subsidiaries ... 29

Table 3: Descriptive Statistics – Number of Employees ... 30

Table 4: Descriptive Statistics – Age of a Company ... 30

Table 5: Descriptive Statistics - Exploration (the number of patents a firm owns) ... 31

Table 6: Descriptive Statistics - Exploitation (the number of trademarks a firm owns) ... 31

Table 7: Means, Standard Deviations, Correlations ... 33

Table 8: Hierarchical Regression Model of Firm Performance ... 34

Table 9: Moderator Analysis for Hypothesis 2 ... 36

Table 10: Moderator Analysis for Hypothesis 2 (exploitation) ... 36

Table 11: Moderator Analysis for Hypothesis 2 (exploration) ... 37

(5)

Abstract

This study examines how the degree of internationalization influences the relation between

international ambidexterity – the equilibrium or a trade-off between exploration and

exploitation – and firm performance. The degree of internationalization is measured in terms

of the pace, scope, and rhythm of internationalization of a company. According to the

business literature, organizations that manage their activities with the focus on international

ambidexterity tend to show a better performance. However, several contextual factors may

influence the abovementioned relation. Also, the theoretical framework includes a number of

control variables, such as age and size of the company, as well as the industry sector it

operates in. The research question of this paper is, therefore: What is the moderating role of

pace, scope, and rhythm of internationalization on the relation of international ambidexterity

on firm performance?

To test the hypotheses, I use data of 261 companies registered in the EU, in 17

industry sectors with a total number of 4120 subsidiaries. All the data is extracted from the

Orbis database. The quantitative analysis is performed in the SPSS (Statistical Package for the

Social Sciences) program using correlation and regression analyses. Results of the analysis

show no support of relation between international ambidexterity and firm performance. The

influence of the scope of internationalization is shown to be insignificant. However, high

rhythm of internationalization shows higher effect on the relation than low rhythm. Also,

results show that pace may positively influence the relation when an organization has a large

(6)

1. Introduction

Prior research in international business literature defines organizational ambidexterity as an organization’s ability to efficiently manage the demand of the business, while simultaneously

adapt to the changes in the environment (Duncan, 1976). Organizational ambidexterity,

therefore, holds in the ability of the organization to find equilibrium between exploration and

exploitation (March, 1991). Nowadays, competitive environment underlines the importance of any multinational enterprise (MNE) to be “ambidextrous” and thus innovate, experiment,

search for new knowledge (exploration), and refine existing knowledge, enhance operations,

invest in efficiency (exploitation) at the same time (March, 1991; Raisch & Birkinshaw,

2008). This focus has resulted in a shift towards “international ambidexterity” where

enterprises incorporate exploitative and explorative strategic decisions while

internationalizing (Hsu et al., 2013).

Previous empirical studies examine managing exploitation/exploration within the

organization using methods, such as surveys and/or interviews of Korean firms, multinational

corporations (MNCs), small and medium-sized enterprises (SMEs) in the United States, and

managers in Spain (Han et al., 2001; Gibson & Birkinshaw, 2004; Bierly & Daly, 2007;

Prieto et al., 2007), and more recently large samples with longitudinal data (Geerts et al.,

2010;Goosen et al., 2012; Caspin-Wagner et al., 2012). Prior research examines the effect of

ambidexterity on various levels: firm, business unit, and individual. Overall, the findings of

previous empirical research are consistent and result in positive effect of ambidexterity on

sales growth, subjective ratings of performance, innovation, market valuation, and firm survival (O’Reilly & Tushman, 2013). Furthermore, previous empirical studies research the

relation between international ambidexterity and firm performance (Hsu et al., 2013; Zhan &

(7)

and look at the Taiwanese multinational enterprises and use exploitative and explorative

Foreign Direct Investment (FDI) forms based on the parent firm’s perspective. Zhan & Chen

(2013) and Luo (2002) both use data of FDIs in China. Other researchers like Barkema &

Drogendijk (2007) address the question of whether internationalization process models still

matter by looking at the experiences of Dutch companies entering Central and Eastern

Europe.

This paper addresses the abovementioned question by means of empirical research

using quantitative analysis. In this study, I focus on the effect of international ambidexterity

on firm performance. I examine if this relation is stronger when there is more pace, scope, and

rhythm of internationalization. I expect a positive effect of international ambidexterity on firm

performance; with moderating effects of the three abovementioned contextual factors. I use

data derived from the Orbis database for all the available years, namely 2007-16. Following

prior research I measure exploration as the number of patents a firm owns and exploitation as

the number of trademarks a firm owns (Youn et al., 2015). Firm performance is measured as

return on assets (ROA) of the firm (Katila & Ahuja, 2002). The data also consists of the

number of domestic subsidiaries, as well as the ones abroad. I use a sample of 261 large

international firms with headquarters in Europe, in 17 industry sectors with a total number of

4120 subsidiaries.

Recently, a lot of research is focused on the issue from the perspective of the emerging

economies, while it seems like the relation between the international ambidexterity and firm

performance from the perspective of developed economies has been seen as something

self-evident. Furthermore, prior research on international ambidexterity tends to focus more on

specific emerging or developed countries rather than on regions. Also, the size of the sample

(8)

exploration/exploitation with various degree of internationalization in Europe. To explore this

issue further, I focus on large international firms with headquarters in Europe.

This paper is organized as follows: in section two I discuss the key literature of the

study and propose several hypotheses. Then, the methods and data are described in the

following section three. Thereafter, section four presents and analyzes the results obtained

from testing the hypotheses. Section five concludes the study and discusses the key findings,

describes the limitations, scientific relevance and managerial implications, and presents

(9)

2. Literature review

In this section, I provide the theoretical background of the empirical study. I begin with a

description of international ambidexterity. The theoretical background leads to the

development of the research question, followed by relevant hypotheses.

2.1 International ambidexterity

Prior studies describe exploration and exploitation as two different aspects on which

organizations focus. Exploration means searching for new knowledge, creating innovations,

and discovering new opportunities, while also staying flexible and being ready to take risk

(March, 1991). Exploitation means using the existing knowledge, choosing for efficiency,

enhancing operations (March, 1991). Although, the two aspects are different, both are

important in creating value and sustaining competitive advantage (O’Reilly & Tushman, 2011). Other authors define ambidexterity in the organization as “manufacturing efficiency

and flexibility (Adler et al., 1999; Carlsson, 1989), differentiation and low-cost strategic

positioning (Porter, 1980, 1996), or global integration and local responsiveness (Bartlett & Ghoshal, 1989)” (Gibson & Birkinshaw, 2004).

Each company has two kinds of aspects to focus on: exploration and exploitation, and

every company has a part of the organization that is responsible for the present operations and

a part that is responsible for the future operations, which means gaining new knowledge to

evolve the company. It is necessary for a company to invest in, and develop, the operations

related to exploration, as well as exploitation. This means that every evolving company has to

incorporate two contradicting strategies, which is also called – ambidexterity (Levinthal &

March, 1993). In case a company fails to do so, or pays more attention towards one of the two

aspects, the consequences could be either getting into the “suboptimal stable equilibria” or

(10)

managing of these two aspects is called ambidexterity. The problem is to implement these two

conflicting strategies in a consistent way. In other words – finding a balance between the two

(Ireland & Webb, 2007). This is what is defined as ambidexterity within an organization.

Ambidexterity, or the idea of combining different exploration and exploitation

strategies, even contradicting ones, can be traced back in prior studies to the beginning of the 1990’s, when the technological progress took its start, and when companies were forced to

find a balance between innovation and effectiveness (Reeves et al., 2015). Around the same

time, March (1991) started the research of how the companies could balance exploration and

exploitation.

March (1991) is not only one of the first to define the terms exploitation and

exploration, he also underlines the importance of the trade-off between the two operations and

stresses that the trade-off is inevitable. He mainly does so by predicting two possible

outcomes or results within the organization in case if the trade-off is not sustained. One of the

possibilities is getting into the “suboptimal stable equilibria”, or competency trap, if more

emphasis is put on the exploitation (March, 1991). On the other side there are the rising

expenses from the investments in the development of new ideas without the expected output,

which takes place if more emphasis is put on the exploration. In other words, those organizations “exhibit too many underdeveloped new ideas and too little distinctive

competence” (March, 1991). This shows how difficult it can be for companies to maintain the

trade-off. In his conceptual paper, March (1991) looks into the relation between exploration

and exploitation and the crucial trade-off between the two by modeling “two general

situations involving the development and use of knowledge in organizations”. March (1991) also discusses the connections of organizational learning, performance, and organization’s

position in a competitive environment, by proposing certain modes where the competitive

(11)

Following March, Tushman and O'Reilly (1996) described the term ambidexterity and

how companies can put the concept in practice. Gibson and Birkinshaw (2004) then continued the research and contributed by adding the concept of “contextual ambidexterity”, meaning

that employees as individuals should constantly choose between alignment (learning) and

adaptation (using), which helps to avoid certain traps common when dividing the operations.

Gibson and Birkinshaw (2004) look into the relation between ambidexterity and performance,

use interviews and surveys of a sample of 41 business units of multinational firms, with a total

of 4195 respondents. Authors find support for their hypothesis that ambidexterity has a positive relation with performance and point out that “successful business units were able to

simultaneously develop these capacities (alignment and adaptability) by aligning themselves around adaptability” (Gibson & Birkinshaw, 2004).

As mentioned earlier, ambidextrous organizations are the ones that are able to combine

exploitative and explorative operations. However, it does not mean that those operations are

created only inside the organization. Companies use various strategies and create operations

inside the company, as well as outside the company. Therefore, Simsek et al. (2009)

developed taxonomy of organizational ambidexterity with two parameters, which resulted in

four types of ambidexterity. First parameter answers the question of how ambidexterity is

created and distinguishes between simultaneous and sequential ambidexterity. Second

parameter looks into where ambidexterity is realized and distinguishes between independent

and interdependent ambidexterity. Based on these two parameters four types of organizational

ambidexterity are identified: harmonic, cyclical, partitional, and reciprocal (Simsek et al.,

2009). These typologies extend the view on the organizational ambidexterity by providing the

(12)

2.2 Degree of internationalization

The degree of internationalization is described with various definitions. Ietto-Gilles (2009),

describes the degree of internationalization as the extent of operations abroad. In other words,

it is the growth of the operations located outside of the company’s home country. Vermeulen

and Barkema (2002) refer to the pace, the rhythm, and the scope of internationalization as a

process of international expansion. It is important to take these processes into consideration

because companies face various restrictions and limitations when expanding on a global scale.

Firms need to take into account cultural and institutional differences, as well as find a

balanced growth to fully exploit the opportunities (Vermeulen & Barkema, 2002).

The degree of internationalization is a topic of interest and research since 1976, and

this was when several researches presented data which verified a positive relation between

possible revenue and expansion on an international level (Hitt et al., 2006; Hymer, 1976).

This early research provides findings about the possible growth of returns if the scope of the company’s operations increases on an international level. Furthermore, firms have to find

equilibrium between benefits and costs. Hymer (1976) argues that control is achieved through

establishing monopoly position by means of acquisition and usage of production advantages

abroad.

Degree of internationalization may be measured as a single item, however, doing so

does not account for the error that might occur (Sullivan, 1994). Therefore, it is more reliable

to measure the degree of internationalization as multiple items. International expansion, as

argued by Vermeulen and Barkema (2002), should be balanced on a number of elements, such

as pace (the velocity), scope (the span relative to products scope and geographies scope), and

rhythm (the regularity). In case if a company does not maintain the balance between the three components, “diseconomies of time compression” may predominate (Vermeulen & Barkema,

(13)

2002). The idea of time compression diseconomies was introduced by Dierickx and Cool (1989), defined as “the fundamental mechanism of diminishing returns when – everything

else equal – the pace of processes increases” (Vermeulen & Barkema, 2002). In other words,

time is considered to restrict how much a company can process and assimilate during its

expansion process.

In the business literature, several studies look into the relation between

internationalization and the performance of the firm (Lu & Beamish, 2004; Hennart, 2007;

Vermeulen & Barkema, 2002; Chang & Rhee, 2011). Depending on the research, the level of

internationalization of a firm is characterized as scale and scope of the business operations.

The scale of diversification is determined by looking at the level of activity of the company in

the operations abroad, for example revenue of foreign divisions of the total revenue, as well as

similar indicators of operational revenue (Qian & Li, 2002). Because this measurement does

not characterize the geographical differentiation of the presence of a firm, it is being

combined with the scope of diversification. The scope of diversification is determined as the

geographical dispersion of foreign operations of a company. The most convenient

measurement of the scope of diversification is the number of countries in which a company is

operating (Qian & Li, 2002).

Above I discuss what previous studies have researched on the topic of international

ambidexterity and the degree of internationalization. Now I predict a number of hypotheses,

which were not yet done. I contribute to the findings of previous studies by testing a number

of hypotheses. The research question of the study is: “What is the moderating role of pace,

scope, and rhythm of internationalization on the relation of international ambidexterity on firm performance?”

(14)

2.3 Development of hypotheses

The idea of managing conflicting operations within the organization has been discussed for

over 25 years, however, there is still not much literature studying the effect of international

ambidexterity on firm performance (Luo & Rui, 2009; Raisch & Birkinshaw, 2008). In

previous empirical studies, international ambidexterity is measured through various methods

and measures. Han et al. (2001) examine the relation by analyzing the results of a survey of

127 Korean consumer product firms and find a positive effect of ambidextrous strategy on

organizational performance, which is measured in terms of Return on Investment (ROI). He

and Wong (2004) conduct a survey of 206 manufacturing firms in Singapore and Malaysia

and also find a positive relation between international ambidexterity and firm performance,

where the latter is measured in terms of sales growth rate. Gibson and Birkinshaw (2004) use

interviews and surveys as a method and look into 41 business units of 10 multinational

companies and find a positive relation of ambidexterity on the performance of business units.

Furthermore, Lubatkin et al. (2006) also use survey as a method of 139 small to medium-sized

enterprises and find a positive relation between ambidextrous firm orientation (exploratory

and exploitative) relative (to competitors) firm performance, where firm performance is

measured in terms of growth and profitability. Auh and Menguc (2005) use a survey of 260

Australian manufacturing firms to examine the relation between exploration/exploitation and

firm performance and also find a positive effect. Venkatraman et al. (2007) show that

simultaneous ambidexterity does not have a positive effect on the growth of the firm,

however, sequential ambidexterity does. Furthermore, Cao et al. (2009) present their findings and support that “integrative construct of exploration and exploitation” has a synergistic effect

on firm performance. Geerts et al. (2010) also supports the positive effect of organizational

(15)

As can be seen from the discussion above, most of the existing previous empirical

studies that examine the relation between ambidexterity within the organization, meaning

finding the balance between exploitation and exploration (Levinthal & March, 1993), and

firm performance use surveys and/or interviews as their method. Overall, the findings of these

empirical studies are consistent, where the results show that managing exploitation and

exploration has a positive effect on firm performance.

Prior research examines the relation of international ambidexterity on firm

performance focusing more often on specific emerging or developed countries, such as Korea,

Taiwan, Spain, China, the Netherlands, Singapore, Malaysia, Australia etc., rather than on

regions (Han et al., 2001; Hsu et al., 2013; Prieto et al., 2007; Zhan & Chen, 2013; Luo, 2002;

Barkema & Drogendijk, 2007; He & Wong, 2004; Auh & Menguc, 2005). No previous

empirical study has yet examined the abovementioned relation using large international firms

with headquarters in Europe. I expect a positive relation between firm performance and

international ambidexterity of firms located in Europe. I expect this effect to be stronger when

a number of factors are present. The factors are pace, scope, and rhythm of

internationalization. Based on the research discussed above the following hypothesis is

suggested:

Hypothesis 1: There is a positive relation between international ambidexterity and firm performance.

Scope of internationalization may influence the performance of a company in various ways.

According to Delios & Beamish (1999), exploitation of available knowledge and advantages

can stimulate firms to enlarge the scope of internationalization and expand into foreign

markets. The scope of internationalization of a firm also serves as a motive to gain

competitive advantage (Porter, 1990). Companies pursuing internationalization and expanding

(16)

al., 1989; Tallman & Li, 1996). Thereby, the advantages resulting from exploitation and

exploration are linked to geographic expansions due to increase in knowledge and economies

of scale and scope (Kogut, 1985; Lu & Beamish, 2004). Another advantage of

internationalization is the minimization of expenses by making use of local competences

(Ghoshal, 1987; Kogut, 1985). Also, organizations may profit from flexibility by means of

increased scope of internationalization, this being a result of ability to adjust to changes in the

environment (Kogut, 1985). Furthermore, Loth & Parks (2002) and Pangarkar (2008) claim

that higher level of scope of internationalization is probable to result in enhanced firm

performance. Companies might also benefit from the effects of operations abroad and create

synergies. For example, a company may collect the knowledge and expertise it acquired in

Taiwan and implement it in China, or a company may combine the productions of both

countries in one plant to create economies of scale.

In contrast to the abovementioned benefits of internationalization, several researches

describe other possible outcomes. Lu & Beamish (2004), for example, look into how benefits

and costs of geographical diversification in internationalization balance themselves. Thereby,

they mention that pace of internationalization, country choice, organizational structure, and especially firm’s intangible assets are key moderators (Lu & Beamish, 2004). Authors put

forward a model depicting an S-shaped curve with three phases of internationalization. Other

researches, such as Thomas & Eden (2004) suggest that the relation depends on the time,

where it shows to be linear in short-term and a U-shape in long-term. This means that in the

beginning of internationalization the costs outweigh the benefits, however, later on the

benefits do exceed the costs. Hennart (2007) argues that the relation between multinationality

and performance is not so obvious and puts forward two main predictions from the literature.

First prediction is that the higher the multinationality, the lower the risk due to less exposure

(17)

the multinationality, the higher the profitability with three main reasons being: 1) exploitation

of scale economies 2) resources are more easily accessible and 3) higher scope of learning

(Hennart, 2007).

Literature shows that the scope of internationalization is an important aspect which

influences the relation between organizational ambidexterity and firm performance. It also

shows that the effect of internationalization can have different effects on the relation between

international ambidexterity and firm performance. Thus, based on the research discussed

above the following hypothesis is suggested:

Hypothesis 2: Scope of internationalization positively moderates the relation between international ambidexterity and firm performance.

In international business literature pace of internationalization is defined as the speed at which

a company is developing or expanding internationally (Chetty et al., 2014). Other researches define the pace, or speed, of internationalization as “the time elapsed until a firm begins to

export or becomes a multinational” (García-García et al., 2017). Lin (2012) defines pace as “a

time-based measure, which is indicative of how much time passes before achieving a specific target or a specific level of performance”.

Pace is thus another factor that can affect profitability when considering international

expansion of the company. Vermeulen & Barkema (2002) focus on the process of

international expansion by looking at the pace (its velocity) of the expansion, scope (its span

relative to product scope and geographic scope), and rhythm (its regularity). They

demonstrate that three points should be balanced to increase the firm performance. If the balance is missing then “diseconomies of time compression” predominate (Vermeulen &

Barkema, 2002). This means that higher pace of international expansion has a negative

moderating effect on the firm performance (if taking into account multinational enterprises).

(18)

a positive effect in the long-term (Vermeulen & Barkema, 2002). Furthermore, they conclude

that the greater the pace, the more negative the relation, to the point that higher pace decreases

the return on assets of the company (Vermeulen & Barkema, 2002). Chang & Rhee (2011) on

the other hand look into the relation between international expansion and firm performance

and ask the question of when the high pace strengthens the competitive advantage of an

internationalizing company. According to Chang & Rhee (2011) advertising intensity has a

positive moderating effect on the relation between expansion and firm performance. High

leverage, on the other hand, has a negative moderating effect, whereas low leverage has a

positive one. And last but not least, industry globalization has a positive moderating effect on

the relation (Chang & Rhee, 2011).

Previous research examines the relation between the pace of internationalization and

firm performance; however, I expect that this has another effect. Pace of internationalization

has a direct effect on firm performance, but it also has a moderating effect on the relation. For

example, a company like BMW operating in China must sell the products, as well as

efficiently respond to the future demands, or in other words maintain the balance between

exploitation and exploration activities. Pace has an important influence on the relation

between international ambidexterity and firm performance because a company has a chance to

be one of the first ones on the market, and reap the advantages of a first mover, or possible

costs of a late mover. A BMW company expands to China, where a number of factors are

different than, for example, in Taiwan, so if BMW responds to those factors they can become

a market-leader in China if they operate quickly. In this case, the pace, or speed, of

internationalization has a positive effect on the competitive advantage of the company.

However, when analyzing previous research discussed above, it can be concluded that the

pace is more like to negatively influence the expansion process and thus also the relation

(19)

argue: “experience that comes too fast can overwhelm managers, leading to an inability to

transform experience into meaningful learning”. The literature discussed above shows that the

relation between the pace of internationalization and performance depends on various factors.

Thus, based on the research discussed above the following hypothesis is suggested:

Hypothesis 3: The pace of internationalization negatively moderates the relation between international ambidexterity and firm performance.

As discussed above, the degree of internationalization, as well as firm performance is influenced by the scope and pace of the firm’s internationalization process. Another important

element of the process is the rhythm. The rhythm of internationalization is defined as the

regularity of the expansion pattern (Vermeulen & Barkema, 2002). The pattern of the

expansion may be regular (rhythmic) or irregular (Vermeulen & Barkema, 2002). For

example, a regular pattern would be a firm establishing two subsidiaries abroad every six

months, where an irregular pattern would be a firm establishing five subsidiaries abroad in

one year, followed by zero established subsidiaries abroad the next two years.

A number of studies include the concept of rhythm in their research on the process of

internationalization (Vermeulen & Barkema, 2002; Lin, 2012; Lin, 2014; Chen et al., 2016).

Vermeulen and Barkema (2002) use a sample of 22 Dutch firms over a period of 25 years

(1967-92) to examine the moderating effect of internationalization process (pace, scope, and

rhythm (irregular pattern)) on the relation between foreign subsidiaries and firm profitability.

Their findings show that all the three elements of the process turn out to negatively moderate

the relation just mentioned. Thus, meaning that a regular pattern during the expansion results

in greater profitability of the firm. Later study by Lin (2012) extends the research by

Vermeulen and Barkema and uses a sample of 772 publicly listed firms in Taiwan over a

period of 8 years (2000-08) to investigate the relation between the level of family ownership

(20)

rapid pace, narrow scope, and an irregular rhythm, which contradicts with the findings by

Vermeulen and Barkema (2002). Lin (2012) argues that the reason for that is that family firms

lack international experience, resources, and thus need more flexibility during their expansion

process.

During the internationalization period firms need to balance their exploration and

exploitation activities. The two expansion patterns, which were introduced by Vermeulen and

Barkema (2002) and explained above, show the possible consequences and differences of

choosing rhythmic or irregular pattern. These two patterns can be linked to exploitation and

exploration activities. It can be argued that when a firm decides to choose for the rhythmic

pattern, it can achieve a more balanced trade-off between exploration and exploitation. Firstly,

firms are able to look back on past experience and utilize it in present similar situations without “overstretching their absorptive capacity” (Ellis, 1965; Cohen & Levinthal, 1989;

Barkema et al., 1997; Vermeulen & Barkema, 2001). Secondly, it leads to companies staying

flexible and effective in executing and implementing the knowledge and experience acquired

(Hitt et al., 1998; Vermeulen & Barkema, 2001, 2002). Furthermore, it is even possible for

firms to achieve a “state of flow” as a result of choosing a rhythmic pattern of

internationalization (Brown & Eisenhardt, 1997; Vermeulen & Barkema, 2002). These

arguments suggest that the relation between international ambidexterity and firm performance

can be negatively moderated by irregular rhythm of internationalization, which leads to the

following hypothesis:

Hypothesis 4: The irregular rhythm of internationalization negatively moderates the relation between international ambidexterity and firm performance.

(21)

Figure 1: Theoretical framework International Ambidexterity - exploration - exploitation Firm Performance Pace of internationalization Scope of internationalization Rhythm of internationalization Control variables - Firm size - Firm age - Industry sector

(22)

3. Data and method

In this section, I discuss the data and method employed in this study. I start with an

explanation of the research design. Then, I describe the data sources and discuss the sample

selection. Last, I provide variable definitions.

3.1 Research design

The study addresses the relation between the international ambidexterity and firm

performance and looks into contextual factors that might influence the relation just

mentioned. As discussed above, there are numerous studies testing the relation between

international ambidexterity and firm performance (Luo & Rui, 2009; Raisch & Birkinshaw,

2008; Han et al., 2001; He & Wong, 2004; Gibson & Birkinshaw, 2004; Lubatkin et al., 2006;

Auh & Menguc, 2005; Venkatraman et al., 2007; Cao et al., 2009; Geerts et al., 2010). These

studies find a positive relation between managing the exploitation and exploration of the

organization and firm performance. However, there is no study that measures the influence of

degree of internationalization, such as pace, scope, and rhythm, on the relation between

international ambidexterity and firm performance. There is a chance that the degree of

internationalization acts as a moderator in the aforementioned relation. To test the role of the

degree of internationalization, I measure three variables and three control variables.

I take international ambidexterity as an independent variable and firm performance as

a dependent variable. The pace of internationalization, the scope of internationalization, and

the rhythm of internationalization are moderating variables. The control variables are: firm

size, age of firm, and industry sector. I use a quantitative research method to test my

(23)

3.2 Data and sample

To test the theoretical framework depicted in Figure 1, I gathered data from the Orbis

database for the period 2007-16, as this is the time span for which data is available. In the

Orbis database various financial and accounting data and information is available from

companies worldwide. I use this data to measure international ambidexterity, performance of

the companies and the degree of internationalization. The sample population consists of large

international firms with headquarters in Europe.

The first data set consisted of 1038 companies, with headquarters in EU, publicly

listed, having patents and trademarks, and foreign subsidiaries. The filter set in Orbis database

was region, publicly listed, with available data on number of employees (together with recent

financial results). All companies should have patents and trademarks, as well as ROA using

Net income. The last, but not the least criterion is holding min 50,01% of shares of foreign

subsidiaries, i.e. firm located in another country other than the country of headquarter

company.

Unfortunately, this search did not provide all the information, namely year of

establishment of subsidiaries was missing. So the next search was all about finding the

missing pieces. The second data set consisted of all the companies, including needed

subsidiaries and their year of establishment, as well as their name, and ID number. The data

was merged, using ID number as criterion for matching, and checked with the company name

if matching was done properly. That is, each company owner had info about the company (performance, size…) and their subsidiaries (that are entered in the Orbis database) with the

year of establishment of each. The year of establishment of each subsidiary was needed for

calculations of rhythm of internationalization per company. The calculations were made in the

(24)

further analysis. Firms with missing values are excluded from the sample. After excluding

firms with missing values, the final sample consisted of 261 companies, with 4120

subsidiaries (with available data). The Orbis database provides data in different format than

needed, so the data for subsidiaries were transformed in order to be used together with the

data for the companies (owners of those subsidiaries). I use both correlation and regression

analyses. In the next chapter I discuss the results of the analyses.

3.3 Measures

This section describes how the variables are measured and the data collection is gathered for

every variable. First, I describe the dependent variable, then the independent variable,

afterwards the moderators and the control variables.

3.3.1 Dependent variable

The dependent variable is firm performance, which is measured by return on assets (ROA) of

the company for the years 2007-16 (Katila & Ahuja, 2002; Lin, 2012). Several studies discuss

that ROA shows to be a reliable measure of firm performance (Contractor et al., 2007; Katila

& Ahuja, 2002). Return on equity (ROE) may also be used as a measure of firm performance;

however, ROE focuses on return to the shareholders, which does not automatically indicate

the (possibly declining) operational profitability of the firm (Hagel et al., 2010). Furthermore,

ROA provides a better perspective on the foundations of the company, as well as asset

utilization (Hagel et al., 2010). Return on sales (ROS) is another possible measurement but it

is considered to measure the income statement profitability instead of firm performance,

whereas ROA considers the assets utilized to maintain business activities (Hagel et al., 2010).

For that reason, I use ROA as a measurement of firm performance. The ROA is computed by

(25)

3.3.2 Independent variable

The independent variable is international ambidexterity, which includes exploration and

exploitation. Exploration is measured by taking the number of patents a firm owns and

exploitation is measured by the number of trademarks a firm owns. As mentioned earlier,

March (1991) defines exploitation as the refinement of the existing knowledge and

exploration as the search for new knowledge. Youn et al., (2015) have used these definitions

in their research on inventive search processes. To measure the “invention as a combinatorial

process" which they see as exploration, authors use US patent records. Trademarks, therefore,

can be seen as an exploitative operation (refinement of existing knowledge). International

ambidexterity is measured by the formula presented below (Hsu et al., 2013).

𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝐴𝑚𝑏𝑖𝑑𝑒𝑥𝑡𝑒𝑟𝑖𝑡𝑦 = 1 − |𝑝𝑎𝑡𝑒𝑛𝑡𝑠𝑖 − 𝑡𝑟𝑎𝑑𝑒𝑚𝑎𝑟𝑘𝑠𝑖| (𝑝𝑎𝑡𝑒𝑛𝑡𝑠𝑖 + 𝑡𝑟𝑎𝑑𝑒𝑚𝑎𝑟𝑘𝑠𝑖)

Here, patentsi is the number of patents a firm owns and trademarksi is the number of

trademarks a firm owns. The outcome is between 0 and 1, which indicates high ambidexterity

when the number is closer to 1 and low ambidexterity when the number is closer to 0. These

data are directly used from the Orbis database.

3.3.3 Moderating variables

This study takes the degree of internationalization as a moderator on the relation between

international ambidexterity and firm performance. The moderating variables are the pace of

internationalization, scope of internationalization, and rhythm of internationalization. All these data were available on the subsidiary level. In order to fit the theoretical model, below

calculations were made, and data per each owner-company were ready for testing the

(26)

Pace of internationalization is chosen as a moderator as according to March (1991),

there is a difference in returns from exploitation and exploration depending on the pace. “What is good in the long run is not always good in the short run” (March, 1991). Pace is

measured as the average number of foreign subsidiaries per year (Lin, 2012).

Scope of internationalization is a second moderator and is measured by taking the

number of foreign subsidiaries to the total number of subsidiaries of the company (Sullivan,

1994).

Rhythm of internationalization is measured through the kurtosis of the first derivative

of the number of foreign subsidiaries of the company over time (Vermeulen & Barkema,

2002; Lin, 2014): 𝑘𝑢𝑟𝑡𝑜𝑠𝑖𝑠 = { 𝑛(𝑛 + 𝟏) (𝑛 − 𝟏)(𝑛 − 𝟐)(𝑛 − 𝟑) ∑ ( 𝑥_i− 𝑥̅ 𝑠 ) 𝟒 } − 𝟑(𝑛 − 𝟏) 𝟐 (𝑛 − 𝟐)(𝑛 − 𝟑)

Here, 𝑛 = number of observations, 𝑥i = number of expansions in year i, and s = standard deviation of the number of expansions (Vermeulen & Barkema, 2002). High kurtosis, or large

peaks with concentrated distribution, indicates an irregular pattern. Low kurtosis, on the other

hand, can be seen from flat distribution, which thus is linked to rhythmic pattern of

internationalization (Vermeulen & Barkema, 2002). This calculation was made for each

company owner in separate Excel sheet, and afterwards reverted to the SPSS file for further

analysis.

3.3.4 Control variables

I use a number of control variables in this study because of the chance that they might

(27)

Firm size is chosen as a control variable because the size of the firm can be one of the

incentives of the firm to internationalize, thereby, it is presumed that large firms operate more

on the global scale than small firms (Dunning, 2000). Firm size is measured by taking the

number of employees of the company (Auh & Menguc, 2005).

The age of firm is the second control variable and is calculated by subtracting 2017

from the year of incorporation. The age of the firm is also seen as firm experience and thus

shows the ability of the firm to successfully enter foreign markets (Oviatt & McDougall,

1997). Furthermore, the age of the firm has an impact on the company’s performance,

short-term, as well as, long-term (Galbreath & Galvin, 2008).

Last, the industry sector is the third control variable, which is chosen for several

reasons. First, it may be expected that some industries are more likely to create patents and

trademarks than others. The reason may be that characteristics of the industry generate more

unique ownership advantages (Dunning, 2000). Also, some industries may be more

knowledge related and invest more in innovations, which in turn may result in more patents

and trademarks. Thereby, industry sectors significantly differ in the level of profitability (de

(28)

4. Results

In this chapter I describe the analyses that I used to test the hypotheses proposed in the

previous chapter two. This section includes the results of the correlation analysis, as well as

the regression analysis.

4.1 Correlation

The data used is derived from the datasets found in Orbis database – observed are

firms registered in EU, with patents and trademarks, having foreign subsidiaries. Second

subset was database of subsidiaries, with main information of year of their establishment,

which was later matched with the first dataset. While matching two datasets, the missing

values were checked, and the result of the analysis is a decision to use only available data.

Namely, there are more subsidiaries than information regarding them, for example, a firm has

458 subsidiaries, but for only 324 data is available. This is a reasonable approach since the

number of missing values would be high and there is no scientific approach to replace missing

data, to make estimation. The descriptive statistics of the dependent, independent, and control

variables used are presented in the following tables.

Out of 261 companies registered in EU, with patents, trademarks, foreign subsidiaries

and data available, 76 is in machinery, equipment, furniture, recycling business. 19.5% is

working in field of chemicals, rubber, plastic, while 9.2% is in services. Metal industry makes

7.3%, while the same percentage goes to wholesale & retail trade companies. Food and

beverage and tobacco make 6.1%, 12 firms are registered for energy (gas, water, electricity),

while wood and publishing make app 3%. Public administration & defense, education, health

(29)

Table 1: Descriptive Statistics - Industry

Frequency Percent

Machinery, equipment, furniture, recycling 76 29,1

Chemicals, rubber, plastics, non-metallic products 51 19,5

Other services 24 9,2

Metals & metal products 19 7,3

Wholesale & retail trade 19 7,3

Food, beverages, tobacco 16 6,1

Gas, Water, Electricity 12 4,6

Wood, cork, paper 9 3,4

Publishing, printing 8 3,1

Post & telecommunications 6 2,3

Construction 5 1,9

Primary sector 4 1,5

Textiles, wearing apparel, leather 4 1,5

Transport 4 1,5

Hotels & restaurants 2 ,8

Education, Health 1 ,4

Public administration & defense 1 ,4

Regarding the share of companies with certain number of foreign subsidiaries, almost

half of the observed companies have 11-50 subsidiaries in their network (45.2%). 64 (24.5%)

of them owns 4-10 subsidiaries. In this dataset only 2% have more than 200 subsidiaries. The

average number of subsidiaries per company is 43. This is the variable that has normal

distribution (skewness = 0.306, kurtosis = - 0.789).

Table 2: Descriptive Statistics – Number of Subsidiaries

Frequency Percent 1-3 subsidiaries 32 12,3 4-10 subsidiaries 64 24,5 11-50 subsidiaries 118 45,2 51-200 subsidiaries 42 16,1 200+ subsidiaries 5 1,9

(30)

When the size of the company is in question, measured by the number of employees,

the average company size is 5960 employees (Table 3). Companies with 5000 - 20000

employees make almost 20% of the companies included. Companies with 1500 - 3000

employees make 13.8% of all companies with foreign subsidiaries. All other categories of

number of employees are equally distributed (cc 10%), except the smallest (8.4%) and the

biggest one (6.9%). Overall, the distribution is a bit skewed to the right and rather pointy

(skewness = 5.882, kurtosis = 42.149).

Table 3: Descriptive Statistics – Number of Employees

Frequency Percent 2-50 employees 22 8,43 51-200 employees 27 10,34 201-500 employees 30 11,49 501-1000 employees 30 11,49 1001-1500 employees 24 9,20 1501-3000 employees 36 13,79 3001-5000 employees 24 9,20 5000-20000 employees 50 19,16 20001 and more 18 6,90

The average age of a company registered in EU with foreign subsidiaries, with patents

and trademarks, is 60 years, i.e. established app. 10 years after WWII. Median is 45 years.

Distribution of the age of company is close to the normal – it is skewed to the right and a bit

more pointy than Gauss curve (skewness = 1.198, kurtosis = .792). There are 7% of

companies established 151 years ago or even more, 10% older than 100 years.

Table 4: Descriptive Statistics – Age of a Company

Frequency Percent

up to 10 years 9 3,45

11-20 years 41 15,71

21-30 years 55 21,07

(31)

41-60 years 30 11,49

61-80 years 33 12,64

81-100 years 25 9,58

101-150 years 28 10,73

151 and more 19 7,28

The independent variable, international ambidexterity, includes exploration and

exploitation, i.e. number of patents and trademarks a firm owns. In average, these firms have

217 patents and 34 trademarks. More than 40% of firms have up to 10 patents (116 firms),

while 23% have 11 - 50 patents. With trademarks is a bit different – 103 companies belong to

group of companies owning 11 - 50 trademarks. Still, while 37 entities have more than 200

patents, only 7 have more than 200 trademarks (Tables 5 and 6).

Table 5: Descriptive Statistics - Exploration (the number of patents a firm owns)

Frequency Percent 1-3 patents 66 25,3 4-10 patents 50 19,2 11-50 patents 60 23,0 51-200 patents 48 18,4 200+ patents 37 14,2

Table 6: Descriptive Statistics - Exploitation (the number of trademarks a firm owns)

Frequency Percent 1-3 trademarks 55 21,1 4-10 trademarks 63 24,1 11-50 trademarks 103 39,5 51-200 trademarks 33 12,6 200+ trademarks 7 2,7

The variables are tested for multicollinearity by assessing the correlations between all

the predictor variables. Out of the two ways to measure correlation covariance and the

correlation coefficient, the two-tailed correlation is used, also known as the Pearson

(32)

variables are perfectly negatively correlated, the coefficient equals -1, while perfect positive

correlation is indicated by a coefficient of 1. The value of correlation coefficient shows the

size of the effect - a small effect has a value of ±.1, a medium effect a value of ±.3, and a

large effect a value of ±.5.

Interestingly, only one variable shows (medium) correlation with firm performance.

The year of incorporation moderately positively correlates with firm performance (.206, p <

.01), meaning that the older the company, the better is the performance. Another interesting

result shows that there is high and significant correlation between number of patents, as

measure of existing knowledge, and trademarks, as measure of search for new knowledge

(.555, p < .01). This finding shows that the premise that there is a trade-off created when

selecting between exploration and exploitation is not justified, but denotes the idea that the

two concepts are instead complementary.

Also, pace, scope and rhythm of internationalization take into account the number of

(foreign) subsidiaries, so there is no surprise that analysis shows (negative) correlation

between these items - pace vs scope (-.140, p < .05), while pace vs rhythm (-.337, p < .01).

This means that the more foreign subsidiaries per year a company has, scope and rhythm will

decrease. As seen in Table 7, the number of employees is correlated to both exploration and

exploitation, pace and scope. Still, there is no significant correlation of company size with firm performance ROA, the rhythm of internationalization, nor with the industry. This could be explained that certain mergers/acquisitions were profitable and successful with both high

and low staff number in different industries. However, the age of firm correlates with all the

variables, except scope of internationalization and industry sector. As can be observed in

Table 7, no evidence of multicollinearity is found, where all correlations are below .7.

(33)

Table 7: Means, Standard Deviations, Correlations

Mean Std. Deviation

1 2 3 4 5 6 7 8 9

1 EXPLORATION Number of patents 216,7 920,39 1

2 EXPLOITATION Number of trademarks 33,7 81,44 .555** 1

3 FIRM_PERFORMANCE ROA using Net income 0,4 15,81 ,053 ,106 1 4 PACE_OF_INTERNATIONALIZATION No of recorded subsidiaries 37,0 74,86 ,077 .177** _,054 ₁ 5 SCOPE_OF_INTERNATIONALIZATION the number of foreign subsidiaries to the total number of subsidiaries

43,0 26,78 -,010 ,037 -,093 -.140* 1

6 RYTHM_OF_INTERNATIONALIZATION the kurtosis of the first derivative of the number of foreign subsidiaries of the company over time

20,0 24,59 -,045 -,092 ,009 -.164** -.337** 1

7 FIRM_SIZE Number of employees Last avail. Yr

5959,6 14319,07 .369** .488** ,072 .422** -.211** -,045 1 8 THE_AGE_OF_FIRM the year of

incorporation

60,5 48,14 .130* _.163** _.206** _.167** _-,104 _.441** _.153* ₁

9 THE_INDUSTRY_SECTOR BvD major sector

8,3 4,64 -,065 -.142* ,095 -,037 -,054 ,111 ,031 -,054 1 **. Correlation is significant at the 0.01 level (2-tailed).

(34)

4.2 Regression analysis

Hypothesis 1 is tested using a multiple linear regression, with the last year ROA as

dependent variable. The multiple linear regression, as a predictive analysis, is used to explain

the relation between one continuous dependent variable and two or more independent

variables. The linear relation between the independent variable international ambidexterity

and dependent variable firm performance, by using control variables as part of the model, as

well as other variables in order not to get the omitted variables bias. In a hierarchical

regression analysis the ANOVA tests whether the model is significantly good at predicting

the outcome.

In the first step, the model was statistically significant F (3,257) = 4.985; p < .01 and

explained 5,5% variance in firm performance. So it can be concluded that most variance of

the variable firm performance is explained by other factors. In step 2 international

ambidexterity was entered as a predictor and the total variance explained by the model as a whole was 5.5% F (4,256) = 3.724; p = 0.975. The introduction of international ambidexterity

explained an additional 0.03% in firm performance, after controlling for firm’s size, age, and

industry sector (R2 Change = .0003; F (1,256) = 0.001; p > 0.05).

Table 8: Hierarchical Regression Model of Firm Performance

R R2 R2 Change B SE β T Step 1 0,234 0,055** Firm size 0,00 0,00 0,04 0,61 Age of firm 0,07 0,02 0,21 3,35 Industry sector 0,36 0,21 0,11 1,72 Step 2 0,235 0,055 0,00 Firm size 0,00 0,00 0,037 0,609 Age of firm 0,07 0,02 0,20** 3,27 Industry sector 0,357 0,21 0,105 1,72 International ambidexterity -0,104 3,313 -0,02 -0,031 Step 3 0,279 0,078 0,023

(35)

Firm size 0,00 0,00 0,008 0,121 Age of firm 0,093 0,024 0,283*** 3,884 Industry sector 0,413 0,209 ,121* 1,977 International ambidexterity 0,343 3,318 0,006 0,103 Pace -0,008 0,015 -0,038 -0,536 Scope -0,071 0,04 -0,12 -1,779 Rhythm -0,113 0,049 -0,167* -2,286 Statistical significance: *p <.05; **p <.01; ***p <.001

In step 3, in order to avoid an omitted variable bias, other variables were entered in the

model (moderation variables), and the total variance was 7.8% F (7,253) = 3.042; p = 0.105.

In the final model three out of seven predictor variables were statistically significant, with

age of firm recording a higher Beta value (β = .28, p < .001) than industry sector (β = 0,121, p < .05) and rhythm (β = -.167, p < .05). In other words, if age of firm increases for one, their

performance will increase for 0.27. On the other hand, if Rhythm increases for one, the

performance will decrease for 0.17.

Still, number of patents and trademarks, are not the factors that are explaining firm

performance (defined with the ROA) significantly. Similar is with other variables – scope, pace and rhythm – the explanation of dependent variable is not significant.

In other words, international ambidexterity, defined by number of patents and

trademarks, is not the factor that is explaining firm performance (defined with the ROA)

significantly. This leads to rejecting hypothesis 1, of positive relation of international

ambidexterity on firm performance.

To test the hypotheses 2, 3 and 4, moderation analysis was performed, in order to

determine whether the size of the effect of an independent variable on a dependent variable

interact with a moderator(s). In order to test this moderating effect, the Process macro written

(36)

As stated by the model summary of the moderation analysis, the model of hypothesis 2 is

significant with F (6,254) = 2.59, p = 0.0188. Then, the significance of interaction effect was

looked at. If this effect is significant, the research can confirm the hypothesis 2 regardless of

the significance of the main effect. However, reading the interaction outline, obtained from

the analysis, I observe that the interaction effect is not < 0.05, but p = 0.460. Therefore, since

the interaction effect is not significant, it can be concluded that there is no evidence of

moderation, hypothesis 2 is rejected. In other words, scope of internationalization, the

number of foreign subsidiaries to the total number of subsidiaries, is not moderating the

relation between international ambidexterity and firm performance.

Table 9: Moderator Analysis for Hypothesis 2

Coeff SE T p intercept i1 -4,843 4,546 -1,066 0,288 international ambidexterity c1 0,001 0,003 -0,311 0,756 SCOPE c2 -0,046 0,055 -0,848 0,397 INT_AMB X SCOPE c3 0,000 0,000 0,739 0,460 R2=0,63, p<0.05 F(6,254)=2,5883

In order to see if there is a difference in 2 variables that make international ambidexterity,

separate analyses were done, one with exploitation and one with exploration, see Table 10

and 11.

Table 10: Moderator Analysis for Hypothesis 2 (exploitation)

Coeff SE T p intercept i1 -4,843 4,546 -1,066 0,288 Exploitation c1 0,001 0,003 -0,311 0,756 SCOPE c2 -0,038 0,505 -0,753 0,452 EXPLOIT X SCOPE c3 0,001 0,001 0,610 0,543 R2=0,76, p<0.05 F(6,254)=2,894

(37)

Table 11: Moderator Analysis for Hypothesis 2 (exploration) Coeff SE T P intercept i1 -6,5644 3.2844 -1,9987 0,0467 Exploration c1 0,001 0,003 -0,311 0,5754 SCOPE c2 -0,038 0,505 -0,753 0,4879 EXPLOR X SCOPE c3 0.000 0,001 0,6837 0,4948 R2=0,616, p<0.05 F(6,254)=2,5699

However, in neither of cases there is a significant moderation, or the difference in effects

on each of the variables is significant. This result confirms previously said – the hypothesis 2

is not accepted.

As it is shown in Table 12, in the model summary of hypothesis 3, the model is significant

with F (6,254) = 2.55, p = 0.0204. Interaction effect is low, thus not significant (p = 0.946).

Having in mind these results of analysis, hypothesis 3 is not accepted, the relation between

international ambidexterity and firm performance is not moderated by the pace of internationalization. The number of subsidiaries per year is not the variable the relation between international ambidexterity and firm performance is depending on. The better

company performance is not reliant on number of subsidiaries per year – it is explained with

some other factors.

Table 12: Moderator Analysis for Hypothesis 3

coeff SE T p intercept i1 -7,044 3,250 -2,167 0,031 international ambidexterity c1 0,006 0,002 0,354 0,723 PACE c2 0,003 0,025 0,121 0,904 INT_AMB X PACE c3 0,000 0,000 -0,069 0,946 R2=0,558, p<0,05 F(6,254)=2,5525

(38)

As in the case of measuring moderating effect of scope of internationalization, in the case

of pace of internationalization as a moderator, separate analyses are done - one with

exploitation and one with exploration as independent variables, see Table 13 and 14.

Coeff SE T p intercept i1 -6,1872 2.9753 -2,0795 0,0386 Exploitation c1 0,0514 0,0193 2,6287 0,0091 PACE c2 0.0130 0.0150 0.8676 0,3864 EXPLOIT X PACE c3 -0,0004 0,0005 0,610 0,543 R2=0,76, p<0.05 F(6,254)=2,894

Table 14: Moderator Analysis for Hypothesis 3 (exploration)

Coeff SE T P intercept i1 -6,8342 3.1373 -2,1783 0,0303 Exploration c1 0,0004 0,0009 0.4289 0,6684 PACE c2 0.0028 0,0238 0.1189 0,9054 EXPLOR X PACE c3 0.000 0,001 -0.8468 0,3979 R2=0,0748, p<0.05 F(6,254)=2,8047

Similarly, both in the case of exploitation and exploration, and their relation with firm

performance, pace of internationalization is not moderating the relation. These results confirm that hypothesis 3 is to be rejected.

Similar to the results of analysis regarding hypothesis 2 and 3, hypothesis 4 - the

irregular rhythm of internationalization negatively moderates the relation between

international ambidexterity and firm performance - is rejected. There is no interaction (coeff = 0.00), while p = 0.595. It seems that the way foreign subsidiaries were opened is not

influencing the relation of dependent and independent variables, as if each case is specific,

(39)

Table 15: Moderator Analysis for Hypothesis 4 coeff SE t p intercept i1 -6,820 3,135 -2,175 0,031 international ambidexterity c1 0,000 0,001 -0,331 0,741 RHYTHM c2 -0,081 0,030 -2,655 0,008 INT_AMB X RHYTHM c3 0,000 0,000 0,533 0,595 R2=0,0666, p<0,05 F(6,254)=2,76252

Again, the testing of hypothesis 4 is done by analyzing the moderating effect of rhythm of

internationalization on exploitation and exploration and their separate relation with firm performance, see Table 16 and 17.

Coeff SE T p intercept i1 -8.1390 3.4996 -2,3257 0,0208 Exploitation c1 0,0185 0,0152 1.2162 0,2250 RHYTHM c2 -0.0648 0.0341 -1.9015 0.0584 EXPLOIT X RHYTHM c3 0,0003 0,0008 0,3424 0,7323 R2=0,706, p<0.05 F(6,254)=2,7901

Table 17: Moderator Analysis for Hypothesis 4 (exploration)

Coeff SE T P intercept i1 -8.2762 3.5059 -2,3606 0,0190 Exploration c1 0,0007 0,0014 0.5012 0,6167 RHYTHM c2 -0.0682 0,0334 -2.0429 0.0421 EXPLOR X RHYTHM c3 0.0001 0,0001 0.629 0,5299 R2=0,0665, p<0.05 F(6,254)=2,7645

Having in mind that no moderating effect is detected of rhythm of internationalization

on the relation of international ambidexterity on firm performance, it is no surprise that the

(40)

ambidexterity. All three moderation analyses confirm that there is no significant moderation effect, and that hypothesis 4 is to be rejected.

Further on, multi-group analysis is performed by making two subsets per each

moderating variable - with high and low values of pace, scope and rhythm of

internationalization, checking the moderation effect on those subsamples, as well as, comparing the result in pursuit of differences. As seen in Table 18, the only moderation that is

significant is the one in the subset with high level of pace of internationalization.

Comparing moderating effects, which are not significant, within subsets with high

scope and low scope values, the difference is also not significant. All in all, it can be

concluded that scope of internationalization is not influencing the relation between firm

performance and international ambidexterity.

When rhythm of internationalization is in question, moderation is not significant

within subsets of low and high rhythm. Hence, the difference between these results is

significant. Meaning that those firms with high rhythm (irregular pattern of opening

subsidiaries) have higher influence on the relation of international ambidexterity and firm

performance than the ones with regular pattern, i.e. low rhythm, thus the influence is negative.

As for the pace of internationalization, multi-group analysis shows significant

influence amongst companies with high pace. This means that the companies which have

large average number of foreign subsidiaries per year have positive influence on the relation

between international ambidexterity and firm performance. In the subset of low pace, this

The relation between international ambidexterity and firm performance, the degree of internationalization as a moderator