Thesis

The Export Performance of
European SMEs
Liane Voerman
Published by: LABYRINT PUBLICATION
P.O. Box 334
2950 AH Alblasserdam
The Netherlands
Tel. 0180-463962
Printed by: Offsetdrukkerij Ridderprint B.V., Ridderkerk
© 2003, J.A. Voerman
Alle rechten voorbehouden. Niets uit deze uitgave mag worden verveelvoudigd, opgeslagen
in een geautomatiseerd gegevensbestand, of openbaar gemaakt, in enige vorm of op eniger
wijze, hetzij elektronisch, mechanisch, door fotokopieën, opnamen, of enig andere manier,
zonder voorafgaande schriftelijke toestemming van de copyrighthouder.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system of any nature, or transmitted in any form or by any means, electronic, mechanical,
now known or hereafter invented, including photocopying and recording, without prior
written permission of the copyright owner.
ISBN 90-5335-022-5
Rijksuniversiteit Groningen
The Export Performance of European SMEs
Proefschrift
ter verkrijging van het doctoraat in de

Economische Wetenschappen
aan de Rijksuniversiteit Groningen
op gezag van de
Rector Magnificus, dr. F. Zwarts,
in het openbaar te verdedigen op
maandag 5 januari 2004
om 16.00 uur
door
Jeannet Alina Voerman
geboren op 11 juli 1972

te Emmeloord
Promotores:
Prof. dr. P.S. Zwart

Prof. dr. M. Wedel
Beoordelingscommissie:
Prof. dr. T.H.A. Bijmolt

Prof. dr. C. Holmquist
Prof. dr. P.S.H. Leeflang
Preface
In the summer of 1995, I said ‘yes’ to a question that changed my life. As a
student-assistant, I was working on a project of Karel Jan Alsem and Janny
Hoekstra. While we were sitting in the sun, enjoying an ice cream, Karel Jan asked
me whether I was interested in applying for a PhD position. Rather naïve, my first
question was “What is a PhD?” The rest is history. I applied for the job, was
accepted and started working on a research project on the export performance of
European small- and medium-sized enterprises. Suddenly, I was on a first-name
basis with the professors and teachers that shaped my passion for marketing. Did I
know what I was getting into? Definitely not. If I could go back in time, would I
take the same decision? I certainly would. Not only did I learn how to conduct
scientific research, but I also became a marketing lecturer myself, teaching students
how to conduct research. It was exactly this combination of doing research and
coaching students that made me fall in love with scientific life, resulting in this
dissertation. The realization of this work would not have been possible without the
help and support of many people.
First of all, I would like to thank my advisors, Peter Zwart and Michel Wedel.
Thank you, Peter, for being my mentor and friend. I could always turn to you when
I got stuck in the research process, or, for that matter, with anything that was on my
mind. I always enjoyed visiting the RENT and ICSB conferences together. I have
particularly warm memories of the conference in Lyon. Michel, thank you for your
support, honesty and advice throughout the years. You are one of the most inspired
researchers I know.
Special thanks for my reading committee, the professors Carin Holmquist, Tammo
Bijmolt, and Peter Leeflang. They are greatly acknowledged for their willingness
to read the manuscript, and their useful suggestions for improvements. Besides,
Peter, I value your confidence and encouragement.
I am thankful to all my colleagues, for their support and interest in both my

research and my personal life. Those 10.30 coffee breaks have become extremely
essential! I remember the soccer talks with Rinus Haaijer, Harald van Heerde, and
Edward Rosbergen, my first PhD-colleagues. I treasure the talks (and chat
sessions) with Marije Teerling, Marieke Woltil, Ewald Hoppen, and Peter Ebbes,
(and the times we played ‘Kolonisten van Catan’). Thanks to Hanneke Tamling
and Jeannet Wiersema, for their endless help and for being there every time I
needed them. I consider Janny Hoekstra as a role model; you are one of the
strongest and truest women that I know. You are very close to my heart. Karel Jan,
thank you for persuading me to apply for the job, and for being such a warm (and
chaotic…) person. You always see the bright side of life. Lastly, I have very fond
memories of Willem Nijkamp; who is no longer with us; he was a true mentor! I
miss his genuine interest and laughter.
I am also indebted to my student-assistants Sietske Vlutters, and Karla Visser for

their work on the references, layout, and their help on whatever task I came up
with, enabling me to free time for my research.
I consider myself very fortunate with my two paranimphs, Brita Janmaat and
Sandra Tillema. Brita, I cherish our life-long friendship and your never-ending
support. This last year brought us even closer together, having our first child three
weeks apart. Sandra, what would I be without our morning chat and your down-to-
earth personality to temper me down? We might seem very different, but we are
very alike. Thank you both for being my friends.
The last words in this preface belong to the most important persons in my life. My
parents, Jaap and Siny, for being the most loving and caring parents possible. My
brother, Antoine, and his partner Carmelita, for being my safety net, especially the
last months. Finally, Michael, thank you for loving me, and Kyran, thanks for
choosing us as your parents. You are my ray of light!
Contents
Chapter 1 Introduction 1
1.1 European Export 1
1.2. Barriers 2
1.3 Success Factors 3
1.4 European Exporting SMEs: The Present 4
1.5 Export Performance of European SMEs 5
Chapter 2 Conceptualizing Export Performance and its Determinants 9

2.1 Conceptual Work 9
2.1.1 Miesenbock (1987) 10
2.1.2 Madsen (1987) 11
2.1.3 Aaby & Slater (1989) 13
2.1.3.1 Chetty & Hamilton (1993) 15
2.1.3.2 Styles & Amber (1994) 15
2.1.4 Gemünden (1991) 16
2.1.5 Zou & Stan (1998) 17
2.1.6 Leonidou, Katsikeas & Piercy (1998) 20
2.1.7 Leonidou, Katsikeas & Samiee (2002) 21
2.1.8 Extending the Existing Reviews 22
2.2 Literature Synthesis: Method & Descriptives 25
2.2.1 Method 25
2.2.1.1 Literature Search 25
2.2.1.2 Data Evolution 27
2.2.1.3 Data Analysis 27
2.2.1.4 Interpretation: Descriptives 28
2.2.1.5 Interpretation: Variables, Method, and Export Performance 31
2.2.1.6 Interpretation: The Actual Review 37
2.3 Literature Synthesis: The Actual Review 38
2.3.1 Environment 38
2.3.2 Firm Characteristics 45
2.3.2.1 Objective Firm Characteristics 46
2.3.2.2 Firm Competencies 49
2.3.3 Managerial Characteristics 54
2.3.3.1 Objective Managerial Characteristics 54
2.3.3.2 Subjective Managerial Characteristics 56
2.3.4 Export Activities 60
2.3.4.1 Firm’s Behavioral Orientation on Exporting 61
2.3.4.2 The Competitive and Expansion Strategy 62
2.3.4.3 The (Behavioral) Commitment of Resources 64
2.3.4.4 Marketing Mix Variables 69
2.4 Towards an Integral Model 78
Chapter 3 The Interstratos Data Set 83
3.1 Introduction 83
3.1.1 Research Model 84
3.1.2 Limitations 84
3.2 Methodology 85
3.2.1 Population and Sampling 85
3.2.2 Survey 86
3.3 Response 87
3.4 Choice for Data Set 89
4 Operationalizing the Integral Export Performance Model 93

4.1 Operationalizing the Theoretical Model 93
4.1.1 Methodology 93
4.1.1.1 Stand-Alone Items and Indices 94
4.1.1.2 Multiple Item Constructs 95
4.1.2 Export Performance 97
4.1.3 Environment 101
4.1.3.1 Developments in the Market 101
4.1.3.2 Stimuli 102
4.1.3.3 The Concept of ‘Environment’ and Operationalization 104
4.1.4 Firm Characteristics 105
4.1.4.1 Firm Size 105
4.1.4.2 International Experience 106
4.1.4.3 Linguistic Skills 108
4.1.4.4 Competencies 109
4.1.4.5 Internal Firm Stimuli 111
4.1.4.6 The Concept of ‘Firm Characteristics’ and Operationalization 112
4.1.5 Managerial Characteristics 112
4.1.5.1 Objective Managerial Characteristics 112
4.1.5.2 Subjective Managerial Characteristics 113
4.1.5.3 Internal Managerial Stimuli 115
4.1.5.4 The Concept ‘Managerial Characteristics’ and Operationalization 115
4.1.6 Export Activities 116
4.1.6.1 Behavioral Orientation on Exporting 116
4.1.6.2 Expansion Strategy 117
4.1.6.3 Behavioral Commitment of Resources: Information Acquisition 117
4.1.6.4 Restructuring the Organization 119
4.1.6.5 Cooperation Abroad 120
4.1.6.6 The Concept of ‘Export Activities’ and Operationalization 122
4.1.7 Variables in Operationalized Model 122
4.2 Multiple Imputation: NORM 123
4.2.1 Multiple Imputation – Theory 124
4.2.2 The Method – NORM 125
4.2.2.1 Assessing Non-Ignorable Missingness 127
4.2.2.2 Variables to be Included in the Imputation Model 131
4.2.2.3 Original and Imputed Original Variables 133
Chapter 5 The Integral Export Performance Model – Estimated 137
5.1 The Operationalized Model 137
5.2 Method of Analysis 139
5.2.1 Two Step Procedure 139
5.2.2 Reflective versus Formative Indicators 140
5.3 Measurement Models for the Latent Variables 141
5.3.1 The Reflective Models 141
5.3.2 Procedure for the Measurement Models 142
5.3.2.1 Confirmatory Factor Analysis 142
5.3.2.2 Outliers 143
5.3.2.3 Which Specific Measures to Include? 144
5.3.3 Results of the Measurement Models 146
5.3.3.1 Goodness-of-Fit of the Measures 146
5.3.3.2 Reliability of the Constructs 149
5.3.3.3 Convergent Validity 149
5.3.3.4 Discriminant Validity and Multicollinearity 150
5.3.3.5 Latent Variables to Include 150
5.4 Index Construction of the Composite Variables 150
5.4.1 The Formative Models 151
5.4.2 Procedure for the Formative Models 151
5.4.3 Results of the Formative Models 152
5.4.4 Issues with Formative Models 153
5.5 The Structural Model 154
5.5.1 The Conceptual Model 154
5.5.2 Procedure for the Conceptual Model 156
5.5.3 Results Structural Model 157
5.5.4 Conclusions 165
Chapter 6 The Information Behavior of Exporting SMEs 169

6.1 Role of Information in Internationalization 171
6.1.1 International Market Orientation 171
6.1.2 International Information Collection 172
6.1.2.1 International Information Collection in Industrial SMEs 172
6.1.3 Information and Export Performance 174
6.1.4 Personality of the Manager 177
6.2 Information Behavior Interstratos 180
6.2.1 The Information Sources 181
6.2.2 The Number of Information Sources 183
6.3 Managerial Personality, Information Behavior, and Export Performance 184
6.3.1 Methodology 184
6.3.2 Measurement of Reflective Models 185
6.3.3 Measurement of Formative Models 187
6.3.4 The Static Model 188
6.3.5 The Longitudinal Model 193
6.3.5.1 Longitudinal Hypotheses 193
6.3.5.2 Results Longitudinal Model 195
6.3.6 Conclusions 202
Chapter 7 Conclusions and Discussion 203
7.1 Introduction 203
7.2 Summary 204
7.3 Important Findings 207
7.4 Limitations and Future Research 210
Appendix 1 The INTERSTRATOS Survey 215

Appendix 2 Results Integral Export Performance Model on Imputed Data Set 2 221
Appendix 3 Results Integral Export Performance Model on Imputed Data Set 3 227
Appendix 4 Results Information Model on Imputed Data Set 2 233
Appendix 5 Results Information Model on Imputed Data Set 3 241
References 249
Samenvatting (Summary in Dutch) 267

Chapter 1
Introduction
Exporting, a centuries-old business concept, has long been at the heart of western
European trade. Yet, it lasted until the second half of the previous millennium for a
true world economy to evolve. The fifteenth century was an age of large fleets of
full-rigged vessels, of the penetration of the Baltic Sea with French and Spanish
bulk goods (e.g. wine and salt), and of overseas discoveries, enabling a world
economy. This world economy was shaped as a network of markets that acted upon
each other closely and on a regular basis, covering most of the globe (Israel 1989).
Antwerp, Amsterdam, and London were the centers of this world economy from
the sixteenth century up to the end of the nineteenth century. This new market
system was inherently different from the world in the antiquity and Middle Ages,
which was characterized by a collective of regional and local economies
(Wallerstein 1974; 1980). Only after 1600, the north European countries entered
into regular relations with the world overseas, and a world economy, as we know
nowadays, was set in motion (Steengaard 1990). Export in European countries
started to bloom due to several success factors, as I will explain, although barriers
did exist that hampered this free trade.
1.1 European Export

In the rise of European export, the Dutch played a central part. Starting at the end
of the sixteenth century, the Netherlands acquired an important role being both the
most highly developed market economy in Europe (Steengaard 1990), and the first
sophisticated staple market involved in high-grade goods as well (Israel 1989).
Already in the sixteenth century, much bulk trade took place with the Baltic States
(salt and wine), for which Holland was the staple market. Besides, the Dutch had a
leading position (near to monopoly) in the trade of herring to the Baltic States. This
position was mostly due to their guaranteed quality of the fish, their use of a new
1
form of vessel, and the high quality of their packaging material: the barrels were of
such high quality that it gave the Dutch herring the competitive advantage over the
cheaper herring from other countries (Boelmans Kranenburg 1977). Not only
herring, but all sorts of commodities were handled on the Dutch staple market.
Later on, Mediterranean commodities and colonial products accounted for a large
part of the turnover on the Amsterdam exchange (Steengaard 1990).
Important factors for the success of the Netherlands were the new and better
shipping vessels, cheap transportation methods, a powerful merchant elite, political
support, guaranteed quality, technical specialization, and innovation. But,
according to Israel (1989), the Dutch state was the primary factor for the long-
lasting pre-eminence of such a small country, giving both support and protection to
the trade, whilst guarding the quality of and setting standards for the economic
activities, such as banking, insurances, and working processes.
The Dutch primacy in world trade lasted from 1585 up to 1740 (Israel 1989), after
which the English took over. Both the Dutch and the English merchants lie at the
heart of modern European trade. For instance, the most successful examples of
merchant organization in the early modern era were the East India companies of
the Dutch and English (Neal 1990). Besides, the discovery of the colonies, and the
subsequent trade in exotic products from Asia and America were stimulating
developments for the seventeenth century international trade, specifically the
import of tobacco, sugar, tea, coffee, and spices (Steengaard 1990)1. This European
dominance of the shipping lanes was a prelude to the European imperial power in
the nineteenth and twentieth centuries (Tracy 1990). Slowly, the world was
becoming smaller, and the possibilities for free trade were becoming larger.
1.2 Barriers
Although European export was successful, differing conditions and rights often
hindered the local and inter-local trade (Kastelein 1984). Examples of such are
tolls, the exclusive right of staple that some cities held for certain goods, guild
1
Unfortunately, these developments had a very negative effect as well, i.e. the trade in
slaves.
2
Introduction
regulations, the preferential treatment of own merchants over foreign merchants,

and the British Navigation Laws (initiated in 1651 and withdrawn not earlier than
halfway the nineteenth century)2.
Besides, the industrial revolution, which gave a head start to England, also had a
downward impact on international free trade. This head start of England rose the
question in other countries what the role of government should be. Given the
technical and economic backlog of England’s surrounding countries, the
governments were given active roles in the protection of certain sectors and the
establishment of the industrial way of production (Kastelein 1984).
Thus, with the possibilities of trade expanding, countries still built up barriers to
protect their own merchants, and, with that, downsizing international free trade.
1.3 Success Factors

In this short historic overview of European export in the previous millennium,
elements can be found that are still current in this millennium. For instance, it is
quite remarkable that such a small country as the Netherlands held a supremacy
over much larger countries as Spain, France or England for such a long time.
Success factors, such as product quality, and governmental influence come
forward; factors that current researchers still mention as crucial for export success
(e.g., Aaby & Slater 1989). With the start of the Industrial Revolution, trade
barriers and protectionism grew abundantly. Today, the destructive effect of trade
barriers is still a hot topic in the free trade discussion and the GATT-rounds
(Cateora & Graham 2002).
2
Stearns & Langer (2001) define these laws as follows: “The British Navigation Laws.
These applied mercantilist doctrine to colonial trade. The Act of 1651, designed to strike a
blow at Dutch shipping, required that colonial products be shipped to England in ships of
Great Britain or the plantations. This law was re-enacted in 1660, with the additional
provision that certain enumerated articles of colonial production could be shipped only to
England. The Staple Act of 1663 required that articles of European production destined for
the colonies must be shipped first to England. The Act of 1673 imposed inter-colonial
duties on sugar, tobacco, and other products.”
3
All these factors, and many more, affect the exporting success of not only
exporting countries, but, on a lower aggregation level, of exporting firms as well.
These last decennia, the emphasis is more on firm-level than on country-level, in
developed countries, such as west European countries. Governments and
companies want to know how to expand international business, diminish barriers,
and boost the international performance of firms.
1.4 European Exporting SMEs: The Present

In this thesis, I focus on exporting firms from Europe. Although there are many
other ways to internationalize, such as by starting a joint venture or a wholly
owned production subsidiary, or by licensing, I choose to focus on the export
market-entry mode3. After all, exporting firms are at the heart of most European
countries. In 2000, the share of exports in turnover was 17% in non-primary private
enterprises for Europe-194 (Observatory for European SMEs 2002). This
importance of export for European firms can be explained partly from historic
patterns as described above, partly from the small size of most European countries
(the smaller the country, the smaller the domestic market, and the sooner firms
look abroad), and partly from general trends, such as globalization (although this
could also be a consequence of firms exporting).
Moreover, I concentrate on small- and medium-sized enterprises (SMEs). SMEs

encompass the far majority of European companies: in 2000, 99.8% of all
companies in the Europe-19 countries have less than 250 employees. Together,
these firms provide jobs for almost two-third of the total employment in these
3
Albaum, Strandskov & Duerr (2002) characterise ‘exporting’ by the fact that production
takes place in the home country, as opposed to production in a free area or overseas. This
export can either be indirect, using independent marketing organizations in the exporter’s
home country, or direct, through selling directly to foreign buyers using its own dependent
unit or by using a foreign-based marketing organization.
4
Europe-19 exists of: Austria, Belgium, Denmark, Finland, France, Germany, Greece,
Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, United Kingdom,
Iceland, Norway, Liechtenstein, Switzerland.
4
Introduction
countries (Observatory of European SMEs 2002). For SMEs, ‘exporting’ is an

especially interesting entry mode, because of its attractive risk and cost profile,
even if the level of control is less than in other internationalization modes (Kuratko
& Hodgetts 2001; Cateora & Graham 2002). On average, 13% of turnover stems
from export, compared with 21% for large companies5 (Observatory of European
SMEs 2002). The size of the figures mentioned in this section give a clear
indication of the significance of exporting SMEs in European countries. In
addition, literature often describes SMEs to be different from large companies
(e.g., Julien 1998; Kuratko & Hodgetts 2001; Risseeuw & Thurik 2003).
Distinguishing characteristics are for instance the already mentioned lack of
resources, the flexibility, and the central role of the owner-manager. Viz.,
Miesenbock (1987) already states that the characteristics and perceptions of the
small firm’s decision-maker is the single most important variable in the exporting
process.
Considering the above, the export performance of European exporting SMEs and
the factors determining this performance are at the heart of this thesis.
1.5 Export Performance of European SMEs

Therefore, the research question of this dissertation is: which factors determine the
export performance of European SMEs? To answer this question, first a conceptual
framework is being build to find possible determinants, and possible relationships
between these determinants and performance using previous literature.
Furthermore, the decision has to be made what export performance encompasses.
In chapter 2, this theoretical base is developed by examining both existing reviews

on export performance literature, and by reviewing the current empirical studies on
export performance. As will be shown, much of what is know about the
determinants of export performance is in the form of individual relationships in
models that link various hypothesized causal variables to various performance
5
Export share for micro firms (0-9 employees) is 7%, for small firms (10-49 employees)
14%, and for medium-sized firms (50-249 employees) 17%.
5
measures. By integrating the results of both the existing reviews and the empirical
studies an overall export performance framework results.
To test this theoretical framework empirically, I use data collected by the

INTERSTRATOS research group (INTERnationalization of STRATegic Orientation of
6
Small- and medium-sized enterprises) . The purpose of INTERSTRATOS is to
explore and describe the strategies for adjustment to changes in environments for
small- and medium-sized firms in Europe. Chapter 3 portrays the details of this
data set.
The next step is to combine the INTERSTRATOS data set with the theoretical
framework developed in Chapter 2. Therefore, the operationalization of the
conceptual determinants using the empirical data is the topic of Chapter 4.
Although the data set has many merits (such as the extensive and longitudinal
character of the research project), it suffers from quite extensive missing values,
especially in the dependent performance variables. Chapter 4 also introduces a
multiple imputation technique: NORM (Schafer 1999).
Subsequently, I validate the constructs operationalized in Chapter 4, and

empirically estimate the integral export performance framework, operationalized
with the INTERSTRATOS data, in Chapter 5. For this, I use a structural equations
modeling technique (i.e. LISREL).
Given the complex nature of the complete integral framework, the mediocre fit of
the integral models, and the exploratory nature of the data set, I decide to zoom in
on a specific part of the framework. In Chapter 6, I focus on the risk attitude of the
manager, his or her information behavior, and the relationships of these with export
performance. After all, Bilkey (1978) already stated: “the quality of management
6
The international research group known as INTERSTRATOS is formed by J. Hanns Pichler,
Christian Lettmayer, Erwin Fröhlich & Peter Voithofer (Austria); Rik Donckels, Ria Aerts
& Jan Degadt (Belgium); Antti Haahti & Petri Ahokangas (Finland); Yvonne Prince, Peter
Zwart & Liane J.A. Voerman (the Netherlands); Per-Anders Havnes & Arild Saether
(Norway); Håkan Boter & Carin Holmquist (Sweden); Margrit Habersaat & Hans Jobst
Pleitner (Switzerland).
6
Introduction
probably is the greatest single determinant of a firm’s export success”. In addition,

the owner-manager has a pivotal place, in SMEs. This merits the special attention
to the owner-manager as a determinant of export performance. Besides, the
upcoming interest in marketing for market orientation, and especially international
market orientation in which information is essential (cf. Kohli & Jaworski 1990;
Cadogan & Diamantopoulos 1995) points towards the significance of information
in being successful. The results on the integral export performance model (Chapter
5) support this importance of the information behavior. Therefore, in Chapter 6
both the static and the longitudinal relationships between the owner-manager’s risk
attitude, the information collection (i.e. part of the international market
orientation), and export performance is examined. More specifically, I test whether
the collection of information has a lagged effect on export performance and vice
versa.
Finally, in chapter 7 the conclusions and managerial implications are stated.

Besides, recommendations are being made concerning (longitudinal) export
performance research and international research project, such as the INTERSTRATOS
project. An overview is given of possible directions for future research
7
8
Chapter 2
Conceptualizing Export
Performance and its
Determinants
This chapter deals with a review of the most recent export performance studies. For
this, first an overview of the existing conceptual articles on export performance is
given, followed by an extensive review of 43 empirical studies trying to explain
export performance from the internal and external environment. In this review, I
follow the five general steps that Cooper (1998) sets for any literature review. The
goal of the literature synthesis is twofold. Firstly, to represent the state of the art in
export performance research at this moment, by describing the existing conceptual
articles. Secondly, to come up with a comprehensive list of factors that influence a
firms’ export performance, by synthesizing 43 empirical studies. Therefore, the
result is a framework that identifies both the principal influentials of export
performance and the firm-product-market contingencies that influence these
relationships. This integral model is the input for the remainder of this dissertation.
2.1 Conceptual Work
Starting in the sixties, ever more researchers have engaged themselves with the
exporting business, although this line of research really took off in the eighties. In
the early stages, most attention went to the behavior of exporting firms without
focusing on performance. For example, in the seventies, Bilkey (1978) performed a
literature review on forty-three exporting studies, which included only four studies
aimed at explaining export performance. Besides, these four studies identified
successful exporters purely on their being active or not in exporting at the time of
9
research. Consequently, any firm not active on a foreign market was considered a
non-successful exporter. Obviously, this rather naive interpretation of the export
performance concept does not take into consideration the magnitude of the foreign
activities of the exporter, or the effectiveness of these operations.
In the eighties and especially the nineties, internationalization increasingly became

a topic of interest in politics, the business world, and the scientific world. An
increasing number of researchers devoted themselves to the explanation of
companies’ export performance. Over the years, this growing stream of export
research resulted in several (attempts to) theoretical frameworks of export
performance, i.e. Miesenbock (1987), Madsen (1987), Aaby & Slater (1989)
(including the follow-up studies by Chetty & Hamilton 1993, and by Styles &
Amber 1994), Gemünden (1991), Zou & Stan (1998), Leonidou, Katsikeas &
Piercy (1998), and Leonidou, Katsikeas & Samiee (2002). These studies all
contributed to the status quo on export performance measurement and its
determinants. The following sub-sections discuss the most important outcomes of
these studies for this dissertation. In this, the goal is to come up with a broad
conceptual framework of categories of variables that can be used as headings in the
synthesis of the 43 empirical studies. Some of these reviews also include
conclusions on the direction and the strength of the relationships between the
various concepts and variables. If this is the case, I discuss these as well.
2.1.1 Miesenbock (1987)

A special and very interesting feature of the review by Miesenbock (1987) in the
realm of this thesis is his focus on small businesses and their exporting behavior.
An important conclusion is that the decision-maker is the single most important
determinant in small business internationalization. Besides, “the empirical studies
in this review show that exporting is a sequential process, during which the firm
increases gradually its export commitment.” This supports the Stages Theory of
Internationalization (e.g. Johanson & Vahlne 1977)7. Lastly, he concludes that it is
7
This is also labeled the Uppsala School. Although outside the scope of this thesis, it is
worth mentioning that, in time, this theory received more and more criticism (e.g., P-A
Havnes 1998; Gankema, Snuif & Zwart 2000), showing that firms do not necessarily
follow a predestined path in internationalising, but behave much more haphazardly.
10
Conceptualizing Export Performance and its Determinants
still impossible to state which variables influence export performance based on the
publications reviewed. “The research needed is sophisticated in investigation
contents as well as statistical methods. Simply listing of reasons for exporting,
export stimuli, etc. is not likely to cause any progress.”
2.1.2 Madsen (1987)

Madsen (1987) synthesizes 17 export performance studies (conducted at the firm
level), published between 1964 and 1985. He classifies the operationalized
variables in these studies as indicators of twenty-three concepts (unobservable or
latent variables) that pertain to four general categories digging into the Strategy-
Structure-Performance paradigm (see Table 2-1): “The basic idea is that the
performance of an organization (‘O-performance’) is a result of a continuous
interaction with other groups of variables, namely its own organizational structure
(‘O-structure’), the structure and performance of its environment (‘E-structure’),
and its own strategies (‘strategy’).” Table 2-1 shows the associations between the
respective variable and export performance reported in this review. As can be seen,
few determinants have been related to all three performance concepts, and the
results are far from conclusive. This can be partly explained from the fact that the
empirical studies reviewed operationalized the various concepts in different ways.
Besides, the effect of some concepts “depends on the situation”.
Madsen (1987) sums up several limitations. These touch upon the lack of
interaction effects, the question of causality, and the inclusion of only a limited
number of operationalized variables and concepts in each study. Most studies
merely investigate the univariate direct effects of the three categories on
performance, and only some studies include the indirect effect of organizational
and environmental structure on performance through strategy. No study
incorporates interaction effects between the various categories. Related to this is
the question of causality Madsen raises: does ‘commitment to exporting’ cause
good export performance or is it the other way around?
11
Table 2-1 Adapted from Madsen (1987): Concepts and Associations found in
the Review.
E-structure O-performance (export profitability, export sales, export
growth)
Export market attractiveness P+ (2), S+ (2), S- (1), G+ (1), O (3)
Trade barriers P- (1), S- (2), G+ (1)
Physical distance to market P- (1), S+ (1), S- (4)
Psychological/cultural distance P- (1), S- (1), G+ (1)
Domestic market attractiveness P- (1), S (1), S- (1)
Type of market O (4)
O-structure
General company resources P- (2), S+ (4), G- (1), O (5)
Knowledge export marketing E+ (1), S+ (4), S- (1), G- (1), O (3)
Management support E+ (2), P+ (1), S+ (5), O (2)
Status export organization E (1), P (2), S (2), O (3)
Technological intensity E- (1), S+ (2), G+ (1), O (2)
Strategy
Market research intensity E+ (1), S+ (2), O (4)
Price competitiveness E+ (1), P- (2), S+ (2), O (2)
Planning & control intensity E+ (1), O (2)
Product strength E+ (1), P+ (1), S+ (6), O (2)
Internalization marketing function E (1), P (2), S (2), S+ (1), O (4)
Channel support P+ (3), S+ (1), G+ (1)
Communication intensity E+ (2), P+ (1), S+ (4)
Adaptation of marketing policy E+ (1), S+ (3), S- (1), O (1)
Marketing concentration P- (1), S (1), S+ (1), G- (1), O (2)
E//P/S/G = association reported with export success in general/profitability/sales/growth
O = no (or very weak and unstable) association reported
+/- = positive/negative association reported; no + or – = direction of found association is not clear
In parentheses, the number of times the respective association was reported.
Lastly, none of the studies is exhaustive with respect to the number of variables.
For instance, a concept often foregone by is the environment. According to the
author, ”[T]his neglect of environmental variables might be attributed to the fact
that most of the studies are concerned wit the investigated firms’ total exports. In
other words, the environment is not clearly defined, since it consists of many
markets with differing characteristics …”. On the other hand, this neglect does
reduce the value of the studies, since environmental variables are likely to
influence export performance. The consequence of leaving out determinants of
export performance can lead to specification errors and biased estimates. After all,
every one of the twenty-three concepts has at least a univariate impact on a concept
in the export performance category, which means that the specification error is
12
inherent in all studies reviewed, as they focus on just a few of the concepts. To
avoid the problems listed above, Madsen (1987) acknowledges: “Clearly, only
(very) extensive studies (if any) could cope with all these requirements at once.
Taking just some of them into consideration, however, would still have the
potential of pushing empirical export performance research further ahead.”
2.1.3 Aaby & Slater (1989)

In 1989, Aaby & Slater publish their much-cited article on the managerial
influences on export performance. Inspired by Bilkey (1978), they review fifty-five
empirical studies on export performance published between 1978 and 1988. This
synthesizes in a framework that distinguishes four independent variable categories,
i.e. ‘Environment’, ‘Competencies’, ‘Firm Characteristics’ (i.e. firm characteristics,
firm capabilities, and management characteristics), and ‘Strategy’, versus one
dependent variable, i.e. ‘Performance’ (see Figure 2-1). Although ‘Environment’ is
incorporated, their review of the empirical studies only focuses “on aspects closely
related to managerially controllable variables”.
The authors draw several conclusions concerning the effect of these managerially
controllable variables. With regard to ‘Firm Characteristics’, the review shows that
the firm size is only important when linked to other aspects that indicate financial
strength or economies of scale. ‘Management commitment’, ‘management systems
and planning’, and ‘export experience’ all impact export performance positively.
Besides, management should have an international vision, be risk-taking and have
a favorable attitude towards exporting. In addition, non-exporters have many
misperceptions about the risk and costs of exporting. As for ‘Competencies’:
“export success through technology depends on good management and what
markets the firm decides to enter.” Successful exporters have specific export
policies, they plan, and they gather market knowledge, using management systems.
The conclusions concerning the additional value of product quality and
communication capability are unclear. As for ‘Strategy’, it is not surprising that
exporters that are more involved focus more on industrial markets and have
broader world market coverage. The results for the marketing mix are only
conclusively positive for the importance of distribution, delivery, and service as
important determinants of export success. For the other elements (product, price,
13
and promotion) the results do not conclusively point towards either adaptation or
standardization.
Environment
External
influences
Performance
Competencies
• Propensity to export
• Export sales
• Technology
• Export problems
• Export/market knowledge
• Exporters vs. non-exporters
• Planning
• Level of export
• Export policy
• Perceptions towards export
• Management control
• Export growth intensity
• Quality
Strategy • Barriers to export
• Communication
Firm characteristics • Market selection

• Use of intermediaries
• Product mix
• Firm size • Product development
• Management commitment • Promotion
• Management perceptions towards • Pricing
• financial incentives • Staffing
• competition
• market potential
• distribution, delivery, and service
Internal
• government incentives influences
• risk
Figure 2-1 A General Model for Assessing Export Performance and Variables (Aaby & Slater 1989, p.9.)
In addition, some research recommendations are made. Concerning the research

design, the authors conclude that most studies are too simplistic and exploratory
due to their focus on simple bivariate relationships (see also Madsen 1987). They
state two major areas for improvement, i.e. the measurement of export performance
14
and the use of longitudinal designs. The performance measure focuses still too
much on the exporter/non-exporter dichotomy and objective export sales measures.
The reviewers favor to include sustainable profitability, and objectives of the firm.
Furthermore, to make statements about the causation of export performance,
longitudinal research is needed, for most studies are cross-sectional. Their final
conclusion runs parallel to the goal of this research: “[I]t is time to take what is
known, develop new research propositions based on current knowledge and
existing theory, and establish a focused research agenda.”
2.1.3.1 Chetty & Hamilton (1993)

Chetty & Hamilton (1993) accept Aaby & Slater’s challenge, and bring together
111 studies to test the proposed framework. They provide “considerable support for
their conceptual model of export performance by confirming, through meta-
analysis, both the validity and relative importance of a number of key variables in
each part of the general model.” Still, they do not find proof for the inclusion of
some variables, such as ‘management control system’, ’perception of competition’,
and ‘use of intermediaries’. They confirm that firm size indeed is a causal factor in
export success.
2.1.3.2 Styles & Amber (1994)

Rooted in the Aaby & Slater framework (1989) as well, Styles & Amber (1994)
propose a ‘revised hybrid model for future testing’. The novelty or hybrid part in
this model versus the Aaby & Slater model lays in the inclusion of a concept
‘Relationships’, next to the familiar categories ‘Environment’, ‘Firm’, ‘Strategy’,
and ‘Export Performance’. This inclusion is based on the relational paradigm as an
alternative to the traditional marketing mix paradigm. “Under the relationship
paradigm (RP), export marketing is driven by the sequential development of
relationships.” Besides, “[T]he RP asserts that relationships and experience are
primary and supplemented by objective data and analysis, rather than the reverse.”
Aspects of the category ‘Relationships’ are the key actors in the network, the
relationship intensity, the reciprocity, and the long-term commitment. Although
they do not test the model, some recent empirical studies adhere to this relational
based approach and do provide tests of this (see 2.3).
15
2.1.4 Gemünden (1991)

As Bilkey (1978) inspired Aaby & Slater (1989), two years later Madsen (1987)
inspired Gemünden (1991) to “… perform a quantitative meta-analysis of these
studies in order to identify the key success factors of export marketing, and to
assess their influence by means of objective statistical procedures.” His sample
consists of fifty studies, published between 1964 and 1987. These studies used
more than 700 indicators assumed to influence the performance of more than 9,000
exporting firms in eighteen different countries.
Characteristics of the Firm

Size Ownership
Industry Export restraints
Home Country Managers Activities Export Success

Market size Goals Information activity Share
Market Growth Export expectations R&D Growth
Competition Education Product policy Profit
Import restraints Foreign orientation Communication
Personality Pricing
Distribution
Export Market Characteristics

Market potential Distance, tariffs
Competition Other barriers
Figure 2-2 Theoretical Framework (Gemünden 1991, p.38).
Gemünden (1991) also develops a conceptual framework, synthesizing all variables

in the articles (see figure 2-2). The categories used to classify the studies are
‘Export Market Characteristics’, ‘Home Country’, ‘Managers’, ‘Characteristics of
the Firm’, ‘Activities’, and ‘Export Success’. Note that Gemünden specifically
distinguishes between characteristics of the manager and of the company, as
opposed to Aaby & Slater (1989), who group these together. Yet, only the direct
16
relationships of the various factors with ‘Export Success’ are meta-analyzed. Most
interesting is the comment on the measurement of export performance. After
distinguishing three indicators, i.e. export share, export growth, and export profit,
Gemünden finds that “[T]here is neither a positive relationship between intensity
and growth, nor between intensity and profit.” This leads to the conclusion that it
does not make any sense to develop only one model that explains all three variables
in one ‘export performance’ concept, if these measures of export success are
unrelated. “[T]his means that export sales intensity is no good proxy for growth or
profitability of exporting.”
Concerning the meta-analysis itself, several difficulties appear, i.e. the extreme
diversity of these studies (in unit of analysis, performance aspects, success factors,
operationalizations, and statistical procedures), the low quality of the data gathered,
the exploratory nature of the data analysis, the lack of theoretical arguments, and
an insufficient disclosure of measurement and data-analytic procedures.
Notwithstanding, the review establishes four prominent export success factors that
have been researched rather frequently, namely ‘firm size’, ‘information activities’,
‘R&D intensity’, and ‘export-oriented product adaptations and services’. “All four
factors show a positive influence on export share of total sales, but only export-
oriented information activity also shows a stronger positive influence on growth
and profitability of export.” For all other variables, the results are either very sparse
or inconclusive. Commenting on these results, the author writes: “It is surprising
that information activity is positively related to all three measures of export
success. It appears to be a variable which has been neglected in the export-
marketing field as a critical success factor.” He recommends in-depth studying of
information search measures, information use measures, and communication
measures.
2.1.5 Zou & Stan (1998)

Zou & Stan’s (1998) motivation is to improve the work by Aaby & Slater (1989),
and by Chetty & Hamilton (1993), by first adding the external environment, and,
secondly, by updating this review. They find 50 articles (published between 1987
17
and 1997) to match their criteria8. On a conceptual level, they group export
performance measures into seven categories, i.e. the financial scales ‘profit’, and
‘growth’, and the non-financial scales ‘success’, ‘satisfaction’, and ‘goal
achievement’, and composite scales. The determinants are classified into internal
(“justified by the resource-based theory”) versus external (“supported by the
industrial organization theory”), and into controllable versus non-controllable
determinants. Table 2-2 shows these determinants, including the number of
positive, negative, and non-significant findings reported on the direct relationship
between the respective determinant and export performance.
Table 2-2 Determinants of Export Performance (see also Zou & Stan 1998).
Internal External
Export Marketing Strategy
General export strategy (12/1/0)
Export planning (19/3/12)
Export organization (13/0/13)
Market research utilization (6/1/6)
Product adaptation (12/2/13)
Product strengths (13/2/27)
Price adaptation (7/1/6)
Price competitiveness (3/0/7)
Controllable
Price determination (0/0/11)

Promotion adaptation (3/3/2)
Promotion intensity (15/2/11)
Distribution channel adaptation (2/1/6)
Distribution channel relationships (9/0/17)
Distribution channel type (5/4/8)
Management Attitudes and Perceptions

Export commitment and support (15/0/2)
International orientation (10/0/6)
Proactive export motivation (1/0/4)
Perceived export advantages (11/0/8)
Perceived export barriers (1/6/9)
8
The studies should be empirical, should report data analysis and statistical tests, should
use some kind of export performance measures as dependent variables, and should be
cross-sectional.
18
Management Characteristics Industry Characteristics

Manag. international experience (15/1/10) Industry’s technological intensity (4/0/1)
Uncontrollable Manag. education / experience (11/3/20) Industry’s level of stability (2/0/1)
Firm’s Characteristics and Competencies Foreign Market Characteristics

Firm’s size (9/5/23) Export market attractiveness (6/3/12)
Firm’s international competence (12/3/7) Export market competitiveness (1/0/5)
Firm’s age (0/3/3) Export market barriers (1/3/8)
Firm’s technology (7/1/11)
Firm’s characteristics (4/1/1) Domestic Market Characteristics
Firm’s capabilities / competencies (20/3/21) Domestic market (2/2/6)
In parentheses the number of positive / negative / non-significant direct associations with an export
performance measure found.
For many variables, the results are either very sparse (e.g., for industry
characteristics) or inconclusive, as can be concluded from Table 2-2. In quite some
instances, positive associations are found, although almost as many non-significant
associations are reported as well. This is due to the fact that published studies
hardly report any significant negative results, favoring positive or non-significant
results. Only in some instances, the positive associations clearly outweigh the non-
significant and negative (‘general export strategy’, ‘export planning’, ‘promotion
intensity’, ‘export commitment and support’, ‘international orientation’, ‘perceived
export advantages’, ‘management’s international experience’, and ‘firm’s
international competence’9). I should note that the results of Zou & Stan (1998)
only mention the direct effects.
Another general conclusion is that the various variables influence the various
performance indicators in a non-consistent manner. The authors indicate several
major problems. “First, several conceptual frameworks developed so far are
competing explanations for export performance.” For example, should environment
be a direct (cf. industrial organization theory), or an indirect determinant, or both?
Other problems that come forward from the studies reviewed are the far from
consistent conceptualization and measurement of export performance, the lack of
agreement on the relevant factors and their measurement, on the unit of analysis,
on the controlling for size of the firm, next to the need for multivariate data
9
Maybe ‘industry’s technological level’ could be included as well, but these results are
based on only five empirical studies.
19
analysis, and for cross-cultural studies. “Building on the significant progress made
in the last decade in the export performance literature, research on the determinants
of export performance should and could achieve a greater advancement toward
mature theory in the next few decades.”
2.1.6 Leonidou, Katsikeas & Piercy (1998)

Leonidou, Katsikeas & Piercy (1998) bring together 46 empirical studies
(published between 1960 and 1995), focusing on “the effect of managerial factors
in facilitating or inhibiting various dimensions of exporting”, although only 13
studies concentrate on export performance (measured in financial terms). These
managerial characteristics are arranged according to two dimensions: (1) objective
versus subjective characteristics, and (2) general versus specific characteristics (see
Table 2-3). The review shows that the majority of these 13 export success studies
stem from the 1980s, are US-based, focus on industrial goods, and on SMEs,
collect data through mail surveys, and use only one to three independent variables.
The empirical results show especially positive results for ‘educational level’ and
‘command of languages’, but for the other determinants, the number of studies is
very small to draw specific conclusions (see Table 2-3).
Table 2-3 Managerial Variables and their Effect on Export Performance
(Leonidou, Katsikeas & Piercy 1998).
General Specific
Objective Age group (+/0, 2) Ethnic origin (not found)
Educational attainment (+/0, 5) Language proficiency (+, 5)
Professional experience (+, 3) Time spent abroad (+/0, 4)
Foreign travel (not found)
Subjective Risk tolerance (+, 1) Risk perception (-, 1)
Innovativeness (not found) Cost perception (not found)
Flexibility (not found) Profit perception (+, 1)
Commitment (+, 2) Growth perception (not found)
Quality and dynamism (0, 1) Complexity perception (-, 1)
In parentheses the overall sign of the relationships reported (+ = positive, 0 = neither positive, nor
negative, - = negative), and the number of studies in the review researching this variable.
The authors signal the one-off nature of most studies, the ethnocentric orientation,
the insufficient construct operationalization, the diffused focus of the studies, and
the emphasis on objective variables. Lastly, they state that future empirical inquiry
20
should include the organizational parameters, the external environment, and

behavioral aspects.
2.1.7 Leonidou, Katsikeas & Samiee (2002)

Elaborating on the previous review, Leonidou, Katsikeas & Samiee (2002) meta-
analyze 36 studies on marketing strategy determinants of export performance
published since 1960. Their model “implies a unidirectional causal relationship:
managerial, organizational, and environmental factors influence the firm’s export
targeting and marketing mix, that in turn affect export performance” (see figure 2-
3).
In this review, only the link between export marketing strategy and export
performance is analyzed, leaving out the antecedents related to the firm, the
manager, or the environment. As for the operationalization of export performance,
the authors identify twelve individual dimensions, although more than 20% of the
studies used a composite10. Overall, the empirical results show the effectiveness of
market segmentation, product quality, pricing strategy, dealer support and
advertising on various performance indicators in export markets. Remarkably, the
authors conclude that “[D]espite the affirmative results observed at the overall
export performance level, marketing strategy variables correlated significantly with
only certain individual performance measures.” Positive associations were found
for export intensity, export sales growth, and export profit level, but “marketing
strategy variables were poorly connected with export market share, profit
contribution, and sales volume.”
10
The dimensions are: export intensity, export sales growth, export profit level, export
sales, export market share, export profit contribution, ROI, export satisfaction, perceived
success, perceived export growth, perceived profitability, and perceived market share. Only
the first six dimensions appeared more than once in the studies (16, 8, 7, 5, 4, and 3 times
respectively).
21
Managerial Characteristics
• General-Objective
• Specific-Objective
• General-Subjective Export Targeting
• Specific-Subjective • Market selection
• Market
segmentation
Export
Organizational Factors Performance
• Company characteristics • Economic
• Operating elements • Non-economic
• Enterprise resources
• Corporate objectives Elements of Export
Marketing Strategy
• Product
• Pricing
Environmental forces • Distribution
• Task environment • Promotion
• Macroenvironment
Figure 2-3 A Synthesis of Export Performance Models (Leonidou, Katsikeas & Samiee 2002, p.52).
2.1.8 Extending the Existing Reviews

The reviews discussed in the previous sub-sections greatly enhanced the world’s
knowledge of export performance, the measurement of this concept and the
determinants leading to export success. The stream of research on export
performance is still growing, with a large upswing in the eighties and nineties. Yet,
both the focus and actuality of these reviews call for a new, updated review. Even
the newer reviews suffer from these issues. For example, the most recent empirical
study reviewed in Zou & Stan (1998) stems from 1997. The same goes for
Leonidou, Katsikeas & Piercy (1998); moreover, they only focus on the managerial
characteristics, and incorporate only 13 export performance studies. Another
shortcoming in Zou & Stan (1998) is the focus on direct effects, and the mere
summation of associations without trying to bring them together into a framework
with relationships between the variables. Furthermore, it is unclear why
‘management attitude and perceptions’ are categorized as controllable
22
determinants, while ‘management characteristics’ are not categorized as such11.

The review done by Leonidou, Katsikeas & Samiee (2002) is more recent, but also
suffers from a myopic view, analyzing only marketing strategy variables. Besides,
the 36 studies reviewed stem from the period 1960 – 1996, leaving out the more
recent contributions. Therefore, there is a need for a new review, incorporating
studies that appeared up to 2002, including all possible factors and trying to build a
framework incorporating all indirect relationships as well.
Building upon the reviews discussed, we use four categories to distinguish between
the various determinants of export performance, i.e. ‘Firm Characteristics’,
‘Managerial Characteristics’, ‘Export Activities’, and ‘Environment’. Table 2-4
shows the correspondence between this classification, and the categories used by
Madsen (1987), Aaby & Slater (1989), Gemünden (1991), and Zou & Stan
(1998)12.
Instead of the general description ‘Strategy’ (Madsen 1987, Aaby & Slater 1989)
or ‘Activities’ (Gemünden 1991), we categorize all activities that specifically
follow from the decision to export under the heading ‘Export Activities’. Zou &
Stan (1998) name this category ‘Export Marketing Strategy’, adhering to the
marketing mix paradigm. I think export behavior encompasses all activities
brought forward by the fact that the firm goes international. This can go further
than the marketing strategy of the firm.
All descriptives belonging to the internal environment of the exporting firm, which
are the background of the exporting process, are listed under ‘Firm Characteristics’
11
A reason could be that Zou & Stan (1998) only bring together export related attitudes and
perceptions, which might be changed by new information or experiences. Yet, as I will
show later on, the general attitude of the manager plays an important role as well, and this
is far from uncontrollable, unless one replaces the owner-manager. Replacing managers is
indeed easier when talking of the management team of larger firms versus the owner-
manager(s) in an SME.
12
The other conceptual studies mentioned in this section, either build upon the frameworks
proposed in these four reviews, or are too specific to be useful in building the complete set
of concepts in an integral export performance model.
23
and ‘managerial characteristics’. In this, the competencies are divided according to

the source where they stem from, i.e. firm or manager. Especially in SMEs, it is
important to distinguish between the effect that the manager has on the exporting
performance, apart from the impact of the firm (cf. Gemünden 1991). This
contrasts with Aaby & Slater (1989), who distinguish between ‘firm
characteristics’ and ‘competencies’, but do not specifically mention the manager.
In Madsen (1987), this all falls under ‘O-structure’. Zou & Stan (1998) do
distinguish between firm and management variables, but as I focus on SMEs I
prefer to look into the owner-manager instead of management.
Table 2-4 Comparison of the Concepts Used in Madsen (1987), Aaby & Slater
(1989), Gemünden (1991), and This Review.
Madsen Aaby & Slater Gemünden (1991) Zou & Stan (1998) This review (2003)
(1987) (1989)
Strategy strategy activities export marketing strategy export behavior
O-structure firm charact. firm character. firm’s character. & firm character.
competencies competencies
managers management charact. manager character.
management attitudes & (objective &
perceptions subjective)
E-structure environment home country domestic market charact. environment
export market character. foreign market charact.
industry charact.
As the concept of environment is still researched very fragmentarily, I use an

overall heading ‘environment’ (cf. Madsen 1987; Aaby & Slater 1989), but will use
the subcategories of Zou & Stan (1998), if applicable.
Concluding, the concepts used to review the empirical studies resemble the Zou &
Stan (1998) review closely, and improve the older reviews by being more specific.
24
2.2 Literature Synthesis: Method & Descriptives
In this section, I review 43 empirical studies published between 1988 and 2003.
First, the methodology used in this review is explained. Second, the descriptives of
the empirical studies are discussed, including some remarks concerning the
methods used in the studies, and a short discussion on the measurement of export
performance. Finally, the actual review is carried out, focusing on the variables
used as determinants of export performance and their direct and indirect
relationship with export performance.
2.2.1 Method
The methodology used in the review is that of Cooper (1998), who advises to
follow five general steps in literature research. This systematic reviewing process
begins with the problem formulation stage, followed by the literature search stage,
the data evaluation stage, the data analysis stage, and ends with the interpretation
and presentation stage.
2.2.1.1 Literature Search

After searching the library and electronic databases (www.picarta.nl and
winspirs13), numerous studies were found using the key words ‘export
performance’, ‘export success’, ‘export sales’, and ‘export ratio’. In addition to this
database-search, the references in the articles found led to new articles and,
consequently, new references.
To be included in the review, the studies have to comply with several conditions. In
particular, only those studies are included that:
1) Focus on the performance of an exporting firm. For example, studies that only
take into account export behavior without linking this to any measure of export
performance are left out (e.g., De Mortanges 1990).
2) Focus on export performance, rather than on the effects of exporting on the
overall performance of a firm. For example, Bernard & Jensen (1997), Bernard
13
These databases provide an excellent overview of published articles and books, where the
former is a Dutch site providing information on all articles and books available in the Dutch
(university) libraries.
25
& Wagner (1997), and Gómez-Mejia & Palich (1997) are excluded, as they all
look at the overall performance of the exporter firm, instead of the export
performance.
3) Focus on the export performance of a company instead of a country, a sector,
or an export venture14. This implies that the famous Cavusgil & Zou study
(1994), but Madsen (1989), Stewart & McAuley (2000), and Ling-Yee &
Ogunmokun (2001) are excluded as well.
4) Focus on manufacturing companies, as our intent is to study these lines of
industry. Therefore, studies including services firms, the agro industry (e.g.,
Douglas 1996), or wholesalers (e.g., Louter, Ouwerkerk & Bakker 1991) are
left out as well.
5) Use primary empirical data in their analysis. This excludes the studies of
Cavusgil & Kirpalani (1993), and Ito & Pucik (1993) who only use secondary
data, and that of Leeflang & De Mortanges (1993), Yeoh & Jeong (1995),
Lages (2000), and Toften & Olsen (2003), as they (unfortunately) do not test
their hypotheses.
6) Use a quantitative primary research method. So, the studies by Chryssochoidis
(1996), and by Davis & Keys (1996) are omitted for their use of case studies.
7) Try to find some kind of causality15 from determinants to export success.
Accordingly, studies in which the authors only describe the most successful
exporters (e.g. Styles & Amber 1994), or place exporters versus non-exporters
(e.g. Holzmüller & Kasper 1990, and Moini 1997) are excluded. In the first
case, only ‘successful exporters’ are examined, leaving the researchers ignorant
if the characteristics of these exporters differ significantly (or not) from less
successful exporters. Besides, we can raise questions on when an exporter is
(most) successful. The latter study treats all exporters as ‘successful’ in
contrast with non-exporters (i.e. a ‘non-successful exporter’), with this
equating success with merely being active internationally.
After applying these preconditions, 43 studies enter the literature review.
14
After all, the total export performance of an exporting company can exist of many export
ventures.
15
With this, we do not mean the true statistical meaning of causality, as this can almost
never be determined by cross-sectional use of surveys. In this case, I want to explain
changes in export performance, and not describe pre-determined groups.
26
2.2.1.2 Data Evaluation

When evaluating the data, the issue of the dependency of estimates is an important
issue to tackle (Cooper 1998). That is, some studies use more than one measure of
the same construct ‘export performance’, and analyze each measure separately. For
example, analyzing export sales and export ratio separately versus one export
performance composite that includes both, leads to two sets of relationships and
results. In addition, some studies use different company samples. Instead of
treating these samples on an aggregated level, the data can be analyzed
disjointedly. In both examples, the elements share the same historical and
situational influences, while in the former case the results even share the same
sample. Therefore, the outcomes of these estimates are dependent on each other. In
the present synthesis, the unit of analysis will be the comparison (in contrast to the
study, or the sample). If any study encompasses several comparisons (e.g. a
comparison of the relationship between export performance and the determinants in
two or more countries or different export performance measures separately
compared with the independent variables), these are handled as separate cases. For
instance, Lee & Yang (1990) carry out seven separate analyzes, on each of their
export performance measures. These are treated as seven comparisons. Similarly,
Beamish, Craig & McLellan (1993) carry out separate analyzes for their United
Kingdom sample and their Canadian sample, which will be treated separately. Yet,
in combining the results of these studies, we have to take into consideration the
more than average impact that a publication can have on the generalizations if it
analyzes various samples or export performance measures.
2.2.1.3 Data Analysis

The next step is deciding how to derive conclusions from the set of studies. A well-
known technique to combine individual studies is meta-analysis16. Yet, a pure
meta-analysis is difficult to implement in this case (see also Glass, McGraw &
Smith 1981, cited in Wolf 1986). Most importantly, the fact that many studies
include different measuring techniques, definitions of variables, and subjects
16
Meta-analysis is defined as a statistical analysis of results from individual studies
(Cooper 1998).
27
makes it hard to compare and aggregate the studies17. This issue will have a large
impact in the present review as well, considering the lack of uniformity in export
performance studies already acknowledged in the conceptual articles discussed in
section 2.1. In such a case, simply estimating overall effect sizes leads to findings
that do not cover for this.
A more appropriate method is the vote-counting method, as this implies measuring

the overall sign and significance of the effect studied by just counting significant
negative, non-significant, and significant positive findings. Still, this technique
treats all findings equally important, thereby foregoing the other point of criticism.
A solution is the combination of the vote-counting method and a narrative
approach, in which analysts take a more subjective perspective on the base of their
experience, theory, and discussion, also distinguishing between poorly and well
designed studies and changing the weights given to the respective studies.
Inferences from export success studies must be tempered for yet another reason.
They have a common deficiency in addressing only surviving exporters. By
definition, only businesses, products, and brands that survived in the foreign
market place(s) in the periods studied are available for investigation. As a
substantial proportion of new exporters fail, this creates a “survivor bias” that
raises concerns about the validity of the empirical findings on the export
performance relationships. Therefore, a narrative approach is more appropriate.
2.2.1.4 Interpretation: descriptives

In interpreting and presenting the studies reviewed, the first step is to compare
them on descriptive details (i.e. the source of publication, the geographical focus,
sample size and response rate, and the firm size of the responding firms) (see Table
2-5).
17
Glass et al. (1981) name some other shortcomings as well, such as the effect of “poorly”
designed studies, the publication bias, and the use of multiple results from the same study.
28
Table 2-5 Studies Reviewed and Descriptives.

Authors (year) Source Geographic Firm Sample (response)
focus size
1 Axinn (1988) IMR USA S/m/l 105 (27.4%)
2 Diamantopoulos & Inglis (1988) IMR Scotland NA 48 (41.4%)
3 Koh & Robicheaux (1988) JBR USA S/m/l 233 (24.5%)
4 Culpan (1989) JBR USA S/m 210 (35.0%)
5 Bourantas & Halikias (1990) EMAC Greece S/m/l 101 (32.1%)
6 Dichtl, Köglmayr & Müller (1990) JIBS Germ. S/m 104 (NA)
7 Lee & Yang (1990) IMR USA S/m 52 (40.6%)
8 Samiee & Walters (1991) JBR USA S/m/l 145 (29.5%)
10 Holzmüller & Kasper (1991) MIR Austria S/m 103 (93.6%)
11 Koh (1991) IMR USA NA 253 (24.5%)
12 Beamish, Craig & McLellan (1993) MIR UK, Canada S/m 116 (25.5%), 126 (37.7%)
13 Donthu & Kim (1993) JoGM USA S/m 640 (23.1%)
14 Kaynak & Kuan (1993) JBR Taiwan S/m/l 154 (12,5%)
15 De Luz (1993) JoGM Brazil S/m/l 18 (36%), + 13 (6.5%)
16 Walters (1993) MIR USA S/m/l 141 (28.7%)
17 Bijmolt & Zwart (1994) JSBM Netherlands S/m 691 (31.7%)
(248 exporters)
18 Evangelista (1994) AiIM Australia S/m/l 193 (9.7%)
19 Naidu & Prasad (1994) JBR USA S/m/l 1156 (42.7%)
20 Johnson & Arunthanes (1995) IMR USA M 224 (18%)
21 Moini (1995) JSBM USA S/m 102 (10.2%)
22 Wagner (1995) SBE Germany S/m/l 923 (na), 1524 (na), 1012
(na), 675 (na)
23 Das (1994) EJoM India S/m/l 58 (NA)
24 Holzmüller & Stöttinger (1996) JoIM Austria S/m 103 (93.6%), 101 (80.8%)
25 Katsikeas, Piercy & Ioannidis (1996) EJoM Greece S/m/l 87 (92.6%)
26 Shoham (1996) JoGM USA S/m/l 81 (5.0%)
27 Katsikeas, Deng & Wortzel (1997) JIM Canada S/m/l 297 (19.7%)
28 Wood & Robertson (1997) IMR USA S/m/l 137 (53%)
29 Leonidou & Kaleka (1998) IMR Cyprus S/m/l 100 (13.1%)
30 Nakos, Brouthers & Brouthers (1998) JoGM Greece S/m 126 (34%)
31 Stump, Athaide & Axinn (1998) JoGM USA (?) NA 94 (52%)
32 Shoham & Kropp (1998) MI&P USA S/m/l 81 (5%)
33 Thirkell & Dau (1998) EJoM N. Zealand S/m/l 253 (39.4%)
34 Beamish, Karavis, Goerzen & Lane (1999) MIR Australia M/l 185 (37%)
35 Shoham (1999) JOIM Israel NA 98 (21.2%)
36 Baldauf, Cravens & Wagner (2000) JOWB Austria NA 184 (53%)
37 Dean, Mengüç & Myers (2000) IMM N. Zealand S 95 (36.5%)
38 Francis & Collins–Dodd (2000) JoIM Canada S/m 88 (51.8%)
39 Robertson & Chetty (2000) IBR N. Zealand NA 70 (42.4%)
40 Shoham (2000) JoGM Israel NA 126 (54.3%)
41 Wolfff & Pett (2000) JSBM USA 157 (10%)
42 Balabanis & Katsikea (2003) IBR UK NA 82 (18,5%)
43 Julien & Ramangalahy (2003) ET&P Canada S/m 346 (11.6%)
29
With respect to the publication source, most studies (i.e. eight) have been published
in the International Marketing Review (IMR). Other journals that issued several
studies on export performance are the Journal of Global Marketing (JoGM, six
studies), the Journal of Business Research (JBR, five studies), Management
International Review (MIR, five studies), Journal of International Marketing
(JoIM, four studies), Journal of Small Business Management (JSBM, three
studies), and the European Journal of Marketing (EJoM, three studies). Besides,
one study appeared in the proceedings of EMAC (the European Marketing
Association Conference), the Journal of International Business (JIBS), Advances in
International Marketing (AiIM), Small Business Economics (SBE), Marketing
Intelligence and Planning (MIP), the Journal of World Business (JoWB),
International Marketing Management (IMM), International Business Review
(IBR), and Entrepreneurship Theory & Practice (ET&P).
Concerning the geographical focus, the United States is over-represented with 16

studies. Fourteen studies focus on Europe, of which four studies take place in
Austria, three in Greece, two in Germany, and the United Kingdom, and one in
Scotland, the Netherlands, and Cyprus. Oceania has been fairly well represented in
our sample, with two Australian studies and three studies originating from New
Zealand. Five studies take place in Canada. The research of export performance in
the other parts of the world is only sporadic. Two studies focus on Israeli firms,
while Taiwanese, Indian, and Brazilian firms only occur once in the sample.
Clearly, the largest part of the studies is USA or Europe based. Within Europe,
Greek and Austrian firms comprise the main part of the studies. Often, the
nationality or affiliation of the researcher(s) determines the geographical focus of
the study.
The average sample size of all studies reviewed is 257,8 respondents, ranging from
31 (De Luz 1993) up to 1524 (Wagner 1995). In most studies, the sample size
ranges between 100 and 200 respondents. Over the years, the studies show a trend
towards larger sample sizes. The response rates are sometimes exceptionally high,
i.e. 93.6% (Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996), and 92.6%
(Katsikeas, Piercy & Ioannidis 1996), while sometimes exceptionally low, i.e. 5%
(Shoham & Kropp 1998). Mostly the response rates fall between 25% and 45%.
30
2.2.1.5 Interpretation: Variables, Method, and Export Performance

The next step is to analyze the studies on their use of independent variables18, the
method used to explain export performance and the measurement of export
performance. Table 2-6 shows the studies, the method used, whether variables in a
specific concept are examined, and a specification of the measure(s) used in
operationalizing export performance.
Table 2-6 Variables and Method Used in the Studies Reviewed19.

Variables
# Method Dependent variable Independent variables
Export Performance (EPF) EV FC MC EB
1 Regressionm ER N Y Y N
2 Discriminant ER (dichotomous) N Y N Y
3 ANOVAo Perceived EP vs. domestic N N N Y
4 Discriminant Composite (ER, ∆ER, #transactions, #foreign markets) N Y* N N
5 Regressionm ER N N N Y
6 Regressionm ∆ in ES N Y Y N
7 ANOVAo Average ER (a), Average ∆ES vs. domestic (b), N N N Y
Evaluation ∆ES vs. domestic (c), Evaluation ∆ES vs.
industry (d), Average EP vs. domestic (e), Evaluation EP
vs. domestic (f), Evaluation EP vs. industry (g)
8 T-test / Chi2 ER (a), # Markets (b), EP(c) N Yi N Yd
9 (a) Regressionm ER N Y Y Y
9 (b) SEM: PLS Composite (ER & ∆ER) Yi Yd, i Yd, i Yd, i
d, i
11 χ2 & ANOVA Perception EP vs. domestic N Y Y Yd
12 Correlation ER in UK (a), EP in UK (b), ER in Canada (c), EP in Y N N Y
Canada (d)
18
As mentioned before, the independent variables are divided into four categories, i.e.
‘Export Activities’, ‘Firm Characteristics’, ‘Managerial Characteristics’, and
‘Environment’.
19
As explained in 2.2.1.1, the unit of analysis is the comparison. Therefore, some studies or
performance measures receive an index such as (a) and (b). These symbolise the different
comparisons, or sets of relationships, that are estimated in the respective study. For
instance, in the seventh study (i.e. Lee & Yang 1990), seven different performance
measures are related with several strategy variables separately, indicated by (a) through (g).
The ninth study (i.e. Holzmüller & Kasper 1991), entails two different ways of modeling
export performance and the determinants, analyzed with different techniques, which receive
the index (a) and (b).
31
Variables
13 Discriminant ∆ES N Y Y Y
14 Discriminants Annual ES (a), Annual EP (b), ER (c), EP-ratio (d) Y Y Y Y
15 Correlation 5 year ∆ES N N N Y
16 T-test, ANOVAo ES (a), ER (b), After tax EP margins (c) N Y i
N Yd
Rate of ∆ ES (d)
17 SEM: LISREL Composite (ER, relative EP, development 5-year ES & N Yi Yd Yd
satisf. with ES)
18 Discriminants Satisf. with EPF N Y Y Y
19 Logistic regression 3-year trend ES and EP Y Y Y N
20 Regressionb EP (a), EMS (b), ES ∆ (c) Yi N N Yd
21 ANOVAo, Composite (ER & ∆ER) N Y Y Y
Duncan’s test
22 Tobit ER N Y N N
23 Discriminants ER Y Y Y Y
24 SEM: PLS Composite (ER, ∆ER) Yi Yd, i Yd, i Yd, i
25 Regressionm (3 Composite (achievement goals EMS, ES & EP) N Y Y Y
steps)
26 Regressionm Composite ES (ER & satisf., ES & satisf., EMS) (a), N N N Y
Composite EP (ROA & satisf., ROI, profit margin &
satisf.) (b), Composite ∆ES (∆ER, ∆ES & ∆EMS) (c),
Composite ∆EP (∆ROA, ∆ROI & ∆ profit margins) (d)
27 MANCOVA, Internationalization degree (experimental/ active/ Yd Yd, i N Yd
ANOVAo committed)
28 Regressionm ER (a), Expected ER (b), Composite (actual ER, expected N Y Y N
ER) (c)
o
29 ANOVA , internationalization degree (experimental/ transitional/ N N N Y
Scheffe’s test advanced)
30 Regressionb ER (a), Perceived EP (b) Y Y Y Y
31 Regressionm ER (a), EP (b) N Y Y Y
32 Regressionm Composite ES (ER & satisf., ES & satisf., EMS) (a), N N N Y
Composite ∆ES (∆ER & satisf., ∆ES & satisf.,∆EMS) (b),
Composite EP (ROA & satisf., ROI, profit margin &
satisf.) (c), Composite ∆EP (∆ROA & satisf., ∆ROI &
satisf., ∆ profit margins) (d)
33 Regressionm Composite (5 year ER, 5 years ∆ER, strategic importance N Yi Yd Yd
(2stage) & self-assessment of EMS, EP, market diversification,
customer satisf., overall self-assessment 5 years ago &
present)
m
34 Regression ES (a), ER (b), ∆ES (c) N Y N Y
35 SEM: LISREL Composite EPF (satisf. with ER, ES & EP) & composite Yi N N Yd
5 year ∆EPF (satisf. 5 year ∆ER, ∆ES, ∆EP)
36 Regressionm ES (a), ER (b), Composite export effectiveness Y Y N Y
(development ES, ER, EMS, # of export markets) (c)
37 Discriminant ES (dichotomous) (a), ∆ES (dichotomous) (b), ER N Y Y Y
(dichotomous) (c)
38 Regressionm ER (a), ES (b), ∆ER (c), Export gross profit margins vs. N Y N Y
domestic (d)
39 Regressionm Composite (importance & satisf. with EP, ER, market Yd N N Yd, i
32
Variables
diversification, ∆ES & overall evaluation EPF 5 years
ago, present and in 3 years)
40 Regressionm Composite static ES (ER & satisf., ES & satisf., EMS) N N N Y
(a), Composite static EP (profit margin & satisf.) (b),
Composite ∆ES (∆ER & satisf., ∆ES & satisf., ∆EMS)
(c), Composite ∆EP (profit margins & satisf.) (d),
Composite management satisf. (6 overall items) (e),
Composite expectations disconfirmation (3 expectation
items) (f)
41 ANOVAo ER N Y N Y
42 SEM: LISREL Composite (self assessment EP, ROI, overall EPF past 5 Y d,i Yd,i Yd N
yrs compared to competitors)
43 SEM: PLS Composite (reputation, ∆ES, EP, ER) N Yi Yi Yd, i
♦ Columns ‘Independent variables’: EV = Environment, FC = Firm Characteristics, MC =
Manager Characteristics, and EB = Export Behavior;
♦ Column ‘Method’: b = bivariate regression, m = multiple regression, o = one-way ANOVA, s =
stepwise discriminant analysis;
♦ Column ‘Export Performance’: EPF = export performance, ER = export ratio / intensity, ES =
export sales, EP = export profitability, EMS = export market share, # = number, ∆ = change/
growth;
♦ Columns ‘EV’, ‘FC’, ‘MC’, ‘EB’: d = direct effect tested = default, i = indirect effect tested, d, i =
direct and indirect effect tested, * = dependent variable.
Variables
Table 2-6 clearly shows that the studies still include one or just a few of the
categories that might influence the performance level of the exporting firms.
Therefore, the set of independent variables studied is just a subset of all variables
that could possibly influence the dependent variable. Most studies did not
incorporate all possible influences on export performance, stating that their
conclusions are conditional and limited to special situations. After all, excluding
possible relevant variables leads to biased results. Just four studies (Kaynak &
Kuan 1993; Das 1994; Holzmüller & Stöttinger 1996; Nakos, Brouthers &
Brouthers 1998) model concepts from all four categories, and only one of these
studies adopts an approach that specifically accounts for indirect effects (i.e.
Holzmüller & Stöttinger 1996).
33
Method
The lack of variables leads automatically to the lack of interrelationships in most
designs, as most studies assume only direct relations. So, Madsen’s (1987)
criticism on this flaw still holds. No more than a few studies have tried to
incorporate indirect relationships in their design. Moreover, when indirect
relationships are assumed, the researchers mostly incorporate these in the form of
control variables, as the method they use is not strong enough to test indirect
effects easily. The most frequent indirect effect examined is the one leading from
firm characteristics through export behavior to export performance. Next, the
indirect effect of the managerial characteristics and the environment has been
tested in some studies. With all, the attention for indirect effects is very sparse.
Related to (or perhaps leading to) the previous point, the most popular techniques
to test the hypothesized associations are simple bivariate or multivariate statistics,
such as correlation, (multiple) regression analysis, and (multiple) discriminant
analysis. Unfortunately, these techniques are incapable of capturing the (often)
complex (indirect) relationships. The trend towards more complex methods, such
as structural equations modeling, is encouraging (Holzmüller & Kasper 1991;
Bijmolt & Zwart 1994; Holzmüller & Stöttinger 1996; Shoham 1999; Balabanis &
Katsikea 2003; Julien & Ramangalahy 2003).
Lastly, all studies are cross-sectional, although they do acknowledge the need for a
longitudinal approach. The inclusion of a time aspect might improve the
understanding of companies’ export performance, as export is an ongoing business.
Some authors try to capture this time aspect of export performance by using change
or growth variables (e.g., Dichtl, Köglmayr & Müller 1990; Shoham 1996; 2000),
by specifically asking for (satisfaction with) export performance in the last five
years (De Luz 1993; Bijmolt & Zwart 1994; Naidu & Prasad 1994; Thirkell & Dau
1998; Shoham 1999; Robertson & Chetty 2000) or by asking for expectations
‘three years from now’ (Robertson & Chetty 2000). Longitudinal designs
strengthen the proof for causal relationships found and enable the researcher to
track the performance factors over time.
34
Measurement of Export Performance

The measurement of the dependent variable export performance has evolved
significantly over time in two directions: firstly, by using multiple measures instead
of single measures, and, secondly, by incorporating subjective measures next to
objective ones (see Table 2-6). Ever more researchers acknowledge that (export)
performance is not a one-dimensional concept. Undoubtedly, a multiple measure is
much more in accordance with the complex nature of export performance.
Literature lends us a number of ways how to dimensionalize (export) performance.
Zou, Taylor & Osland (1998) make a distinction between (1) financial (or
economic) outcomes, (2) strategic (or non-economic) outcomes, and (3) perceptual
or attitudinal measures (plus composites of these three). After reviewing firm-level
export performance studies, Katsikeas, Leonidou & Morgan (2000) use a three-way
categorization, i.e. economic (sales-related, profit-related, market share-related),
non-economic (market-related, product-related, and miscellaneous), and generic
measures (degree of satisfaction, perceived export success, and degree to which
export objectives have been fulfilled). Other options are to divide (export)
performance into the three dimensions of effectiveness, efficiency and adaptability
(Styles 1998), or into sales-related measures, profit-related measures, and growth-
related measures (Shoham 1996; 1998; 2000). All these measures can be measured
either objectively or subjectively, although preferably both. This
multidimensionality of the concept and the choice for objective and/or subjective
proxies clearly recurs in the studies reviewed here.
Returning to the empirical studies in this review, ever more studies use several
measures of export performance, rather than using just one proxy. Some research a
one dimensional export performance measure, while others use multidimensional
measures. These can be labeled either as one-dimension/single-proxy measures (i.e.
covering one dimension with one proxy), as one-dimension/multiple-proxy
measures (i.e. covering one dimension with more than one proxy), or as
multidimensional/multiple-proxy measures (i.e. covering more than one dimension
with more than one proxy). Table 2-7 clarifies this categorization.
35
Table 2-7 Categorization Measures of Export Performance.

Dimensions / proxies Single Multiple
One-dimensional One-dimensional single proxy (a) One-dimensional multiple proxy (b)
Multidimensional Not available Multidimensional multiple proxy (c)
The decision to use one-dimensional/single-proxy measures (option a, see Table

2-7) results in testing every relationship hypothesized separately with each export
performance measure in the design (e.g.,. Samiee & Walters 1990; Dean, Mengüç
& Myers 2000). In some cases, this one by one testing is due to restrictions in the
analytical technique (e.g., when using discriminant analysis). In other cases, the
reason lies in the belief that the diverse dimensions of export performance relate in
different manners to the determinants, and, thus, should not be forced into one
single composite measure for all dimensions, and must be tested separately. Some
measures are also one-dimensional in nature, but are composed of several proxies
(option b, see Table 2-7). An example is the Shoham (1996) study that uses four
one-dimensional composites that each consist of several proxies, i.e. multiple
proxies for the one-dimensional measures ‘export sales’, ‘export profitability’,
‘change in export sales’, and ‘change in export profitability’. Conversely, Bijmolt
& Zwart (1994), Thirkell & Dau (1998), and Julien & Ramangalahy (2003) are
examples of studies that use a multidimensional/multiple-proxy composite (option
c, see Table 2-7) to operationalise export performance, aiming to measure the
impact of determinants on overall export performance without considering effects
that can be outbalanced by the different dimensions20.
In addition, researchers rely more and more on subjective measures of

performance, using perceived values of variables, such as the ‘perceived relative
20
Although not included in this review, it is interesting to mention the measurement of
export performance in Louter, Ouwerkerk & Bakker (1991). Here, four export performance
measures are distinguished, i.e. ‘export profitability’, ‘relative export profitability’, ‘export
contribution to total profit’, and ‘export intensity’. Louter, Ouwerkerk & Bakker (1991)
builds one model, but uses these four as four different dependent variables, with even
linkages between the performance measures. For instance, both ‘export profitability’ and
‘export contribution to total exports’ are hypothesized to affect ‘relative export
profitability’. None of the studies reviewed in this chapter acknowledge such relationships
between performance measures.
36
export profitability’ (Koh & Robicheaux 1988; Koh 1991), or satisfaction variables
as used by Bijmolt & Zwart (1994), Shoham (1996; 1999; 2000), and Shoham &
Kropp (1998). Shoham (2000) also introduces another interesting measure: the
disconfirmation of expectations. This interest into expectations connects closely to
the essence of SMEs: the prominent role of the owner-manager in the firm. If
researchers want to investigate the success of export in SMEs, an important issue
that cannot be ignored is how pleased the owner-manager is with the
internationalization project(s). After all, if he is not satisfied, or if his expectations
are not met, he can easily decide not to proceed with the internationalization and
blow the whole operation off.
Some proxies in the list give rise to questions regarding their value as a
performance indicator, such as the number of export markets (e.g., Samiee &
Walters 1990)21, the degree of internationalization (Katsikeas, Piercy & Ioannidis
1996; Leonidou & Kaleka 1998)22, or reputation (Julien & Ramangalahy (2003)23.
When looking at the frequency with which the proxies have been investigated,
Table 2-6 clearly shows the favorite role of export ratio as a proxy for export
performance, either as a single measure or as part of a composite. A good runner
up is the export sales proxy, followed by profit considerations. Clearly, multiple
proxy (composites) and subjective measurements become increasingly popular.
2.2.1.6 Interpretation: The Actual Review

The last part of the interpretation phase is the actual literature synthesis. This
examination of export success determinants aims at finding general patterns among
the empirical studies. In this synthesis, the results are ordered in sections according
to the concepts of the external environment, firm characteristics, managerial
21
The variable ‘number of export markets’ is discussed in section 2.3.4.2 as well, while
some studies see this variable as a proxy for export strategy, or for export experience.
22
The degree of internationalization says more on the path that the firm follows in
internationalising (e.g., the Uppsala school), although most authors simply operationalise
‘degree of internationalization’ as export intensity or export ratio.
23
The reputation can also be considered as a determinant of export performance, and is
used as such by some researchers (see 2.3.4.4).
37
characteristics, and the export behavior. As this is a quite extensive piece, this part
will be treated in a separate section.
2.3 Literature Synthesis: The Actual Review
First, the variables researched in the empirical studies are content analyzed and put
into the right category. Next, the results on the various variables in the categories
are structured, described, compared and analyzed. Every subsection ends with an
overview of the variables found and the synthesized results.
2.3.1 Environment
No company operates in a vacuum, but deals with an external environment. Often,
this constitutes a complex reality for a manager; even more when a firm is
operating outside the domestic market as well. Yet, the literature on export
performance has largely ignored the environment as a determinant of export
performance. Most studies only focus on the managerially controllable or internal
aspects. Only 11 studies incorporate external aspects in their research design, with
an even lower number considering characteristics of both the domestic market and
the export market. Furthermore, an exhaustive description of the external setting of
the firm is almost impossible, as this means including everything outside the firm
which could impact business. The empirical results reflect this complexity. Only a
few studies incorporate uncontrollable aspects outside the firm, the variables
researched vary greatly amongst the studies. After content analyzing the variables
pertaining to the environment in the studies reviewed, the following general
external aspects can be distinguished:
• the attractiveness of the export market,
• the attractiveness of the home market, and the
• the influence of the domestic government.
Although this distinction is often used, some studiescombine these separate aspects
in composites, which often come forward from treating these aspects as ‘barriers’,
or ‘stimuli’ (‘motives’) to which a firm reacts (or not). These are often categorized
as either internal versus external stimuli, or as reactive versus proactive stimuli (cf.
Albaum, Strandskov & Duerr 2002; Leonidou 1995; 1998). Internal factors stem
38
from motives internal to the organization, while external factors pertain to stimuli
coming from outside the company. Proactive factors are unique competencies or
market opportunities to which the company reacts, while reactive motives refer to
environmental pressures, which force companies to respond. In this section, we are
only concerned with the external stimuli or barriers. The sections on managerial
and firm characteristics discuss the internal barriers or stimuli 24.
Export Market Attractiveness

Variables that come forward when investigating the attractiveness of the export
market relate to the economic, socio-cultural and the political situation in a
country. Besides, differences between the export and home market on some of
these aspects are under examination.
Including 19 variables on issues such as regulations, potential demand, and the

economical and political situation, the Kaynak & Kuan (1993) study is the most
extensive article on environmental determinants. They perform a discriminant
analysis to identify the external variables differentiating significantly between high
and low export performers on export sales (9 discriminating variables), export ratio
(no discriminating variables), export profit (1 discriminating variable) and export
profit ratio (3 discriminating variables). The results are far from uniform for the
various performance measures. To the authors’ surprise, export sales flourishes in a
market, which is economically less developed, less industrialized, has a less
positive attitude towards the foreign product, a stricter import (licensing) policy,
and more standards. Yet, unemployment rates are lower, and foreign exchange is
more convenient. Similarly, in relation with the contribution of exporting to total
profit: “the foreign market environment in which the high performers operate is
characterized by unstable public policy, very low untapped and unexpected
demand, but very convenient foreign exchange.” For export profitability, “the
suitable foreign target market environment… are more industrialized places, where
24
As said before, some authors try to compose a comprehensive list of both stimuli and
barriers, which is often factor-analyzed to reduce the number of variables. As such, the
studies often combine internal and external aspects into one overall construct, which makes
it difficult to distinguish between the two. In this review, I try to follow a clean-cut
categorization as good as possible.
39
the quality control standards are more inter-acceptable and price competition
pressure and market price fluctuations are lower”. Therefore, the impact of the
export market varies with the performance measures, with export ratio being
unaffected in this study. In addition, different variables bring about higher sales
versus higher profit, which also applies to Baldauf, Cravens, & Wagner (2000).
They find that the respondents perceive foreign political environment (i.e. inflation
rates, exchange rates, and import restrictions) to affect only the export sales in a
negative manner, while export ratio and export effectiveness go unchanged. In
addition, the socio-cultural environment (customs, culture, and religion) has no
influence at all on performance. Yet, Balabanis & Katsikea (2003) find that a
dynamic, i.e. unstable environment, induces managers to adopt an entrepreneurial
posture, thereby boosting export performance.
Another important aspect of the export market attractiveness is the competition in

the host country. Beamish, Craig & McLellan (1993) find that the presence of
direct competition in foreign markets negatively affects export ratio (although only
for Canadian and not for UK firms), whereas competition does not influence export
profitability25. Thirkell & Dau (1998) do not hypothesize the foreign competition to
change the performance directly, but indirectly by increasing the need for a
marketing orientation. The results show that with more intense competition abroad,
the firm heightens its marketing orientation, which improves (the self-assessed)
export performance. Robertson & Chetty (2000) incorporate competition in the
hostility of the domestic and foreign environment (i.e. market risk, investment and
marketing opportunities, and the level of competition plus hindrances) in their
study on New Zealand exporters, and link this to the management orientation. They
hypothesize that the more hostile (benign) the environment, the more
entrepreneurial (conservative) the strategic orientation should be to perform
successfully. The firms are advised that “the entrepreneurial approach is desirable
as it is able to perform successfully in all contextual situations”. On the other hand,
Balabanis & Katsikea (2003) fail to establish this hypothesized effect of
environmental hostility on the entrepreneurial posture of managers, although there
25
Beamish, Craig, and McLellan (1993) categorize foreign competition as a characteristic
of the product. In my view, the presence of competition says inherently more about the
target market than the product, and should be treated in this section.
40
is a direct negative effect on economic export performance. Overall, high

competition seems to be a threat, but might be overcome by getting to know the
market26.
Shoham (1999) incorporates economic, socio-cultural, and political externals,

partly as the differences between the home and export market on some aspects (i.e.
competitive structure, physical climate, local laws, marketing infrastructure), next
to the image of the exporting country, and the influence of the local government.
An indirect effect is assumed, hypothesizing that the distinctiveness of the foreign
environment encourages firms to adapt (or standardize) the marketing mix.
Eventually, only the similarity of the physical climate (more adaptation of
distribution), and the foreign governmental influence (product standardized, and
promotion marginally adapted) are significant in this. In turn, product and
promotion adaptation enhance performance (see also 2.3.4.4). Similarly, Johnson &
Arunthanes (1995) hypothesize that ‘government regulation’, ‘infra structural
differences’, ‘market lag’ (i.e. difference in product life cycle), ‘competitive
intensity’, ‘cultural differences’, and ‘end-user differences in preferences and
tastes’ heightens ideal and actual product adaptation, improving export market
share, sales growth, and profit. The data show that when governmental regulations
and infrastructure differ more, the actual adaptation level rises. Besides, the
governmental regulations and the market lag affect the extent of ideal adaptation
upwards. So, differences between the home and export market ask for more (actual
or ideal) adaptation, whether these stem from governmental rules or market
peculiarities. Wagner (1995) also finds market differences to increase the planning
propensity (the more ‘distant’ the export market is geographically, the higher the
planning propensity of the US exporter). Balabanis & Katsikea (2003) include
environment diversity as well, but find a non-significant effect on both
entrepreneurial posture and export performance. Maybe the diversity of the
markets served only impacts the strategic activities displayed. Examining
interaction effects, they do establish that firms exporting to culturally dissimilar
26
Although not suitable for this review, as they focus on the export venture level (instead of
the exporting firm level), Cavusgil and Zou (1994) find that the ‘export market
competitiveness’ affects the adaptation of the product and promotion, and the support to the
foreign distributor positively.
41
countries may benefit slightly more from the adoption of an entrepreneurial

approach. Looking at the level of the economy and infrastructure in the host
country, Das (1994) finds that their successful LDC exporters export to developed
countries, but incorporate no indirect relationships.
Home Market
Concerning the home market, Nakos, Brouthers & Brouthers (1998) find only
limited proof for the impact of domestic competition: firms in highly competitive
home markets appear to have higher profitability, but no higher export sales than
firms in less competitive home markets. Yet, Das (1994) does find that successful
exporters (using export intensity) operate in turbulent (highly competitive and
unstable) environments. Naidu & Prasad (1994) deduce that “companies in export-
intensive industries learn to become more regular exporters.” They explain this by
stating that “[W]hen competitor firms are engaged in export activity, this serves as
a great incentive for firms to pursue exporting on a regular basis.” Therefore, some
copying behavior takes place in the industry. Besides, the 3-year trend in export
sales and profitability are influenced positively as well.
Government
Until now, the impact of the authorities is primarily focused on the restrictive role
of governments (regulations, standards, exchange policy, etc.). On the other hand,
governments often provide support for exporters27. Frequently, this is in the form
of information or assistance provided to exporting companies in special programs
or institutions. Holzmüller & Kasper (1991) and Holzmüller & Stöttinger (1996)
both find the indirect effect of the perceived quality of export consultancy provided
by the Federal Chamber of Trade and Industry, by financial institutions, and by
government agencies on the export ratio and the change herein to be positive
through the heightened (competitive) position held in foreign markets and the
foreign orientation of the manager.
27
Although not included in this review, for the lack of empirical testing, Leeflang & De
Mortanges (1993) discuss the effect of the single European market on advertising. They
hypothesize that EU-92 will induce firms to standardize their advertising, but they also
warn that the EU-consumers will only blend to a certain extent.
42
Some authors ask managers for a list of stimuli or problems that enhanced or
hindered their export operations. In this section, only those problems or stimuli
related to the external environment are covered28. Evangelista (1994) finds that
poor performers (measured in manager’s satisfaction with export performance)
perceive problems with obtaining capital as relatively more limiting than high
performers. Katsikeas, Piercy & Ioannidis (1996) investigate both barriers (also
named problems), and stimuli for their affect on export performance. Respondents
were asked how frequently eight possible problems were experienced during the
export operations, and to what extent each item negatively affected the firm’s
export operations. Of these, two are externally related (i.e. the problem of national
policy, and of domestic currency devaluation), but they do not have any influence
on the achievement of export goals. Concerning the stimuli, Katsikeas, Piercy &
Ioannidis (1996) do obtain significant results for a positive perceived impact of
national policy, although domestic market pressures, fortuitous conditions, and
exogenous market conditions are not significant. Dean, Mengüç & Myers (2000)
find similar results with ‘foreign restrictions and standards’, and ‘financial
impediments’ being non-significant for export sales, export ratio, and export
growth, and ‘exchange concerns’ positively affecting export sales, but not the other
two measures. Baldauf, Cravens & Wagner (2000) find that the proactive external
motives (i.e. physical closeness to customer abroad, obtaining tax advantages,
development of new sales territories, and taking advantage of promising foreign
business) appear to be positively related to export effectiveness and export
intensity (no significant impact on export sales), while reactive external motives
(competitive pressures in domestic market, overproduction in domestic market,
disadvantageous legal changes in domestic market, and increased fixed costs)
deteriorate export intensity (export effectiveness and export sales go unchanged).
Hence, reacting to negative pressures seems not to improve export performance,
while opportunities in foreign markets as a stimulus is advantageous.
28
De Luz (1993) also touches upon the subject of barriers (‘motives’), and adds these to his
survey. Unfortunately, the study does not give any results on this part of the survey.
43
Summary Environment
Table 2-8 gives a short overview of the results found in the studies discussed,
categorized into export market attractiveness29 (encompassing the macro-economic
situation, the political environment, the level of foreign competition, and the
(similarity of the) foreign market), home country attractiveness (domestic
competition), and home government (export assistance and national policy). As the
results of the studies are fragmented in the sense that different studies use different
performance measures, and different variables, operationalized in various ways, it
is impossible to univocally state a direction of the relationship between the external
factors and performance, or other variables.
Table 2-8 Summary Environment.
direct effect indirect effect
tested tested (on)
export market attractiveness (composite) * * (on orientation)
macro-economical situation * not tested
political situation (composite) * not tested
tariffs * not tested
regulations / standards * * (on adaptation)
foreign competition * * (on orientation & adaptation)
(similarity) market (composite) * not tested
demand * * (on adaptation)
infrastructure * * (on adaptation)
socio-cultural environment * * (on adaptation)
home market attractiveness
domestic competition * not tested
home government
export assistance not tested * (on attitude & export activities)
national policy * not tested
Some cautious conclusions are drawn. Only a few significant direct effects have
been found, which can be opposite for various studies, or various performance
measures. The strongest results can be found for indirect effects from market
differences on the adaptation level and orientation of management. The level of
competition in the foreign market also affects this orientation. National policy can
29
Although Madsen (1989) only investigates one export case study, thereby not qualifying
for this review, his findings are worth mentioning here. In this study, only one out of five
‘market’ variables, i.e. the ‘attractiveness of the export market’, has a positive impact on
one of the three performance measures, i.e. export sales.
44
have a positive effect, directly, but also indirectly, as export assistance improves a
companies’ position and management orientation. Lastly, the farther away the
markets, the more companies plan.
2.3.2 Firm Characteristics

Next to the external environment, the internal environment plays a crucial role in
shaping the context in which the exporting activities are conducted and/or the
export performance is achieved. Thus, the organization or SME itself is a non-
ignorable factor in the quest for export performance determinants. I divide the firm
concept into (1) background characteristics (or demographics) and (2) firm
competencies. The background characteristics found in this review are (related to)
firm size, company age, industry specifics (and the technological complexity of the
product30), firm structure, and firm culture. Although not exactly the same by
definition, competencies are also referred to using related terms, such as
capabilities (e.g., Evangelista 1994), competitive advantages (e.g., Moini 1995), or
expertise (e.g., Moini 1995)31. These incorporate some or all aspects of overall
(international) experience, managing-, marketing-, and exporting competencies.
30
Here, the technological complexity of the general product offering of the firm is
examined. I discuss the export product in the section on export behavior, the choice which
product to export being a conscious export decision.
31
Grant (1991) offers a good distinction between the various terms. Resources are inputs
into the production process, which can be financial, physical, human, technological, and
organizational resources or reputation. In addition, Barney (1991) distinguishes between
physical, human, and organizational capital resources (although he does add to the
confusion by labelling ‘capabilities’ under resources as well). Accordingly, firm size,
(international) experience, and expertise should be looked upon as resources. On the other
hand, “the capabilities of a firm are what it can do as a result of teams of resources working
together” (Grant 1991). In some instances, these are also named ‘distinctive competencies’
(Snow & Hrebiniak 1980) or ‘core competencies’ (Prahalad & Hamel 1990). In addition,
Collis (1994) uses the definition for organizational capabilities “as the socially complex
routines that determine the efficiency with which firms physically transform inputs into
outputs.” Lastly, ‘competitive advantages’ are those capabilities in which the firm
dominates over the competitors, including an assessment of the firm’s capabilities related to
the environment (e.g., Barney 1991; Alsem 2001; Leeflang 2003). Notwithstanding these
45
2.3.2.1 Background Characteristics

Firm Size
With regard to SME demographics, the influence of ‘firm size’ on export
performance has been researched extensively, using various operationalizations.
The most popular proxy is ‘number of employees’, followed by ‘sales volume’32.
The effects of size measured as total sales are only estimated directly in the studies
reviewed. The majority of these results are non-significant. The few significant
results do show a positive effect of ‘total sales’, but merely on export sales, and not
on other performance indicators. In contrast, firm size as the number of employees
is often assumed to impact performance directly and indirectly through behavior.
That is, having more employees increases export planning and information
collection (Samiee & Walters 1990; Walters 1993), the competitive market
position (Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996), the
marketing orientation (Thirkell & Dau 1998), the entrepreneurial posture
(Balabanis & Katsikea 2003), and changes in the structure in the (export)
organization (Samiee & Walters 1990; Balabanis & Katsikea 2003). Large firms
attach less importance to ‘foreign market accessibility’, ‘export competence’, and
‘distribution competitiveness’ (Katsikeas, Deng & Wortzel 1997). These authors
accredit their results to “the resource constraints inherent in smaller firms”, giving
lead to the inference that the number of employees affects the way the organization
views exporting. Using the resource based view as well, Wolff & Pett (2000)
definitions, most empirical and conceptual literature on export performance place

‘experience’ and ‘expertise’ under ‘capabilities’ or ‘competencies’. In Barney's (1991)
view, this is correct as he categorizes the capabilities under resources. Therefore,
concerning these terms, confusion still exists. To be consistent with the existing stream of
export performance literature, we label ‘experience’ and ‘expertise’ as competencies,
keeping in mind the definitions mentioned above.
32
Besides, ‘stock of export goods’, ‘receivables’, ‘resources’, ‘number of international
countries and manufacturing facilities’, and ‘revenue’ are proxies that come up (Dichtl,
Köglmayr & Müller 1990; Naidu & Prasad 1994; Beamish et al. 1999; Baldauf, Cravens &
Wagner 2000). As these proxies each come up only once, and give a very diffuse
impression, we will not elaborate on them, but focus on the number of employees and sales
volume.
46
hypothesize that firms with less employees follow narrower based competitive
patterns than larger firms do. To their surprise, they have to dispel this hypothesis:
“It is not the breadth or quantity of resources but the types of resources available to
the firm that determine a firm’s competitive patterns and competitive action”.
Concerning the direct effects of the size of the work force, some proof can be
found that having more employees implies higher export sales (Dichtl, Köglmayr
& Müller 1990; Kaynak & Kuan 1993; Baldauf, Cravens & Wagner 2000),
although most studies find a non-significant relationship (Axinn 1988;
Diamantopoulos & Inglis 1988; Culpan 1989; Donthu & Kim 1993; Evangelista
1994; Naidu & Prasad 1994; Moini 1995; Katsikeas, Piercy & Ioannidis 1996;
Stump, Athaide & Axinn 1998; Wolff & Pett 2000). The proof for the influence of
employee accumulation on export intensity is inconclusive with Wagner (1995) and
Nakos, Brouthers & Brouthers (1998) finding a higher, and Kaynak & Kuan (1993)
a lower export ratio. Wagner (1995) does mention that the positive impact of firm
size decreases with size, consistent with the hypothesis that a positive relationship
should exist, but only up to a point33. This is in line with the findings of Bonaccorsi
(1992). He concludes that the general consensus is that firm size (as measured by
annual sales or number of employees) increases the probability of exporting, but
that no general support can be found for the assumption that firm size positively
impacts export intensity. The “limited resources argument only takes into account
internal resources, while organizations try to stabilize their environment through
relationships with external actors.”34 Similarly, the effect on export effectiveness
(profit) is found to be both positive (Nakos, Brouthers & Brouthers 1998) and
negative (Kaynak & Kuan 1993; Baldauf, Cravens & Wagner 2000). Therefore,
notwithstanding the popularity of firm size as an antecedent of export strategy
and/or performance, the results on firm size are not conclusive and cannot be
generalized, although some evidence exists that a firm with more employees has
33
Although not incorporated in this review, the hypothesis tested by Cavusgil & Kirpalani
(1993) is interesting to mention here. That is, they assume a curve-linear effect from size on
export performance, i.e. small and large firms are expected to be more successful than
medium-sized enterprises. They find mixed results, which they attribute to the interaction
between size and industry technology, and between size and export strategy.
34
Styles & Amber (1994), and Johnson & Arunthanes (1995) also adhere to this relational
paradigm
47
higher export sales. An important criticism is the question whether firm size causes
or is caused by export performance (e.g., Diamantopoulos & Inglis 1988; Wagner
1995). After all, firm size can be both cause and effect of export performance. Yet,
most studies assume a unidirectional path from firm size to export performance.
Longitudinal studies are needed to see whether this relationship is recursive.
Firm History
Concerning the firm’s history, the studies provide some evidence that younger
companies achieve better on the international market, although the majority of the
estimates are non-significant (Diamantopoulos & Inglis 1988; Kaynak & Kuan
1993; Das 1994; Nakos, Brouthers & Brouthers 1998; Baldauf, Cravens & Wagner
2000).
Technology Level of the Product

The technology level of the product is sometimes proxied by measuring the
specific industry in which the company operates, as the type of product differs per
sector. The industry (consumer-product exporters are more successful) does affect
export performance both directly (Das 1994; Beamish et al. 1999), and indirectly
through the factors deemed important for international success (Katsikeas, Deng &
Wortzel 1997), or through product adaptation (Johnson & Arunthanes 1995)35.
Other studies specifically investigate the technical nature of the firm’s offer.
Overall, the technical complexity of the product seems not to have a significant
direct effect on export performance (Stump, Athaide & Axinn 1998; Francis &
Collins-Dodd 2000), although Wagner (1995) finds the innovation level (which
encompasses the technology level as well) to impact export ratio positively for
three out of four industries. Regarding indirect relationships, a technologically
intensive product offering induces firms to plan more (Walters 1993), to be more
marketing orientated (Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996;
35
Some studies incorporate industry specifics as environmental variables. We choose to
include these variables with firm characteristics as they pertain to the internal specifics of
the organization. That is, industry is often used as proxy for the technological intensity of
the product (i.e. a background characteristic), or can be seen as a source of experience, such
as industry export intensity (i.e. a competence).
48
Thirkell & Dau 1998), and to gain a better competitive market position
(Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996)36.
Structure and Culture

Just a few studies incorporate structural and cultural characteristics. Some evidence
is found on firm ownership, the administrative efficiency, and the formality present
in the organization. Bijmolt & Zwart (1994), and Nakos, Brouthers & Brouthers
(1998) both find that a firm benefits if it is part of another firm, especially when
this is an international firm. So, foreign owned firms outperform domestically
owned firms, maybe due to the enhanced level of international experience available
in the company. Conversely, Wagner (1995) finds for two out of four industries
that being part of a multi-established company is not beneficial for export ratio.
Holzmüller & Kasper (1991) find that “firms with a more efficient administration
turn out to be more successful in export business”, and to have a improved foreign
orientation (see also Holzmüller & Stöttinger 1996). In the same studies, the degree
of formality in the organization (‘Y-orientation in values’) affects the competitive
market position negatively, but hardly affects export performance. This coincides
with their findings on the internal culture of a firm; a high ‘dynamic cultural
orientation’ (i.e. more task- and people-oriented than power- and role-oriented)
contributes to export performance (‘export ratio’ and ‘change in export ratio’) both
directly, and indirectly through the position that the firm holds in international
markets. Similarly, Balabanis & Katsikea (2003) establish that firms with a more
organic structure have a more entrepreneurial posture, herewith increasing
economic export performance.
2.3.2.2 Firm Competencies

Firm Experience
The second part of the firm variables reviewed concern the competencies or
capabilities of the firm. The best-examined competence variable is the firm
experience (18 studies), either in general business or in export business. Whereas
36
Cavusgil & Zou (1994) also find that the ‘technology orientation of the industry’ affects
the degree of product and promotion adaptation of the export venture negatively. Besides, if
the industry is more technologically oriented, they find that the foreign distributor receives
more support.
49
the day-to-day business experience is barely modeled and, when taken into the
model, found to be of no significance for exporting performance (Moini 1995;
Dean, Mengüç & Myers 2000), the exporting experience is examined extensively.
Most (i.e. thirteen) studies measure export experience as the number of years a firm
exports. Most of the direct relationships hypothesized turn out to be non-
significant. As for the significant relationships found: For ‘export sales’ only one
positive result can be found (Francis & Collins-Dodd 2000), for ‘export ratio’ two
(Baldauf, Cravens & Wagner 2000; Francis & Collins-Dodd 2000) one for
‘expected export ratio’ (Wood & Robertson 1997)37, while for ‘export profit’ the
significant results show one negative result (Kaynak & Kuan 1993). Using
composite measures for export performance, Thirkell & Dau (1998), and Leonidou
& Kaleka (1998) both find a strong positive effect of ‘export market knowledge’
on ‘overall export performance’, whereas Naidu & Prasad (1994) find
inexperienced exporters to perform better, albeit that experienced exporters are
more regular exporters38. According to the authors, “[O]ver time, experienced
exporters become more realistic about the profit impact of exporting, tempering
their profitability expectations.” Incorporating indirect effects as well, Bijmolt &
Zwart (1994) find that the export policy improves when exporters are more
experienced39. Once more, the evidence of these studies is not conclusive, but
slightly tends towards export experience having a positive effect.
37
In this study, the sign for the objective measure ‘export ratio’ is non-significant, while on
a subjective level the respondents belief in the impact of export experience on export ratio,
as proven by the positive influence on expected export ratio.
38
In our view, this is not a very strong conclusion, as regularity implies being in business
for a somewhat longer time, and experience is measured as years in exporting. Therefore,
there is of course correlation between the operationalization of both measures.
39
A more negative note on the indirect effect of export experience comes from the study by
Cavusgil & Zou (1994), which is not reviewed due to the criteria set earlier. They conclude
that the more experienced firm adapts the promotion campaign more while it should not,
and wishes to adapt the product less while it should. Therefore, experience leads to an
erroneous assessment of the needs of the foreign market. This is in line with the negative
note on firm age and firm experience in Mintzberg (1989).
50
A critical note on the use of years as a proxy is not redundant, as experience can
develop from much more than simply years in exporting, e.g., from the learning
effect from intense contact with the foreign market(s). Furthermore, the implication
“the older, the wiser” does not necessarily hold. One can imagine that people doing
something for a couple of years are more rusted and not as alert and innovative as
‘fresh’ people. On the other hand, experience in years gives the company a set of
historic actions to learn from and to improve their actions upon, whenever a similar
situation comes along. Yet, “years in exporting” correlates strongly with the age of
the firm, and “the older an organization, the more formalized its behavior”
Mintzberg (1989, p. 106). We already established the negative impacts of
bureaucracy and firm age on export performance in the previous section. Nakos,
Brouthers & Brouthers (1998) introduce a somewhat different view by
hypothesizing that a U-shaped relationship might exist that linear regression
techniques are unable to detect. “It might be that both highly experienced and new,
younger firms have higher performance than average experience firms.”
Unfortunately, no evidence can be found for this statement.
In my view, one dimension of experience can be measured by using years in

exporting business, but the measurement improves when complemented with other
proxies, capturing all dimensions. Examples are ‘the number of markets’ and ‘the
frequency of visiting markets’, as both activities essentially tap into the concept of
international experience40. Talking of (export) experience is more real when
looking further than just the number of years people are in (international) business.
Experience also stems from international activities, and the level of these activities.
Some authors explicitly include other proxies for experience. Katsikeas, Piercy &
Ioannidis (1996) explicitly use the number of export countries, next to the number
of years, as they distinguish between the length (years) and scope (number of
countries) of export experience. Another example is Thirkell & Dau (1998) who
use a multi-faceted concept named ‘export market knowledge’, measured by a
combination of the number of years in exporting, the number of export countries
and the number of market visits. Unfortunately, most studies use just one proxy for
40
Of course, counter-arguments can be brought up, by saying that this last variable belongs
to the export activities of the firm, while the years in exporting is more a sunk characteristic
of the firm.
51
export experience. Reasoning even further, we can state that export experience
should be replaced with a measure named ‘international experience’, placing just as
much importance on international experience stemming from importing as from
exporting. After all, importing also implies dealing with companies abroad, which
all leads to experiential knowledge on international business.
Other Capabilities
In addition to a broad concept as ‘export experience’, capabilities related to general
and specific aspects of management, marketing and exporting activities come
forward in the review. These are all tested in direct relationship with export
performance, foregoing any indirect impact, with the exception of Julien &
Ramangalahy (2003). By using a composite ‘export competencies’ construct
(comprised of the competencies with regard to overall export competence, to
marketing, to export ability, and to information) Julien & Ramangalahy (2003) find
a strong positive association with the ‘competitive strategy’ of the exporting SME.
In turn, a good competitive strategy boosts export performance (reputation, export
sales growth, export profitability, and export intensity). Besides, the more
competent an SME is in exporting, the more important it considers export
information41.
Concerning the direct effects, the wide-ranging capabilities of general

management, personnel management, (export) production, and planning have no
significant influence on respondents’ satisfaction with export performance
(Evangelista 1994), nor on export ratio (Moini 1995), nor on export performance as
a composite (Katsikeas, Piercy & Ioannidis 1996), while expertise in finance, and
marketing positively impacts export performance (Evangelista 1994; Moini 1995;
Katsikeas, Piercy & Ioannidis 1996). On the more specific marketing capabilities,
Moini (1995) finds that competitive advantages in product quality, technology,
service, or newness do not affect export ratio, whereas the availability of a patented
product is beneficial. Next, Beamish, Craig & McLellan (1993), and Katsikeas,
Piercy & Ioannidis (1996) find the broader concept of ‘product superiority’ to be
non-significant. Yet, exporters tend to be more satisfied when they possess a higher
41
The direct effect of ‘information perception’ on ‘competitive strategy’ is non-significant,
but it does increase the actual use of information, and boosts competitive strategy as such.
52
capability in product R&D (Evangelista 1994). Concerning price, a competitively

priced product seems not to be significant (Kaynak & Kuan 1993; Katsikeas,
Piercy & Ioannidis 1996). The conclusions related to place are minimal and
inconsistent, with distribution capability being ignorable in Evangelista (1994), and
advantageous in Moini (1995). Similarly, the results for the capabilities of
information collection and utilization capabilities are ambiguous (Kaynak & Kuan
1993), although the availability of a business plan is constructive (Bijmolt & Zwart
1994).
Summary Firm Characteristics

To summarize, firm characteristics have an impact on exporting activities and/or
export performance, but not all of them have and not to the same extent. The best-
researched characteristics are firm size and export experience, but it is hard to
make some explicit conclusions. A few cautious inferences are that larger firms
perform better (especially measured in export sales), plan more, and that more
experienced firms do gain from that knowledge. Besides, firm culture should be
people- and task-oriented, the structure not too formal. The firm benefits from a(n)
(international) mother firm, but the evidence is not too strong. The competencies
do not give an equivocal picture either, but overall exporters should try to gain a
marketing competence. The results pertaining to the other competencies are not
clear, mainly because of the small base of empirical evidence. Table 2-9 gives a
summary of the direct and indirect relationships between firm characteristics and
export performance that have been tested in the studies reviewed.
Table 2-9 Summary Firm Characteristics.
direct effects indirect effects
tested tested (on)
Background variables
firm size * * (on manager attitude, export activities)
company age * not tested
structure * * (on manager attitude)
culture * not tested
technology level/ industry * * (on export activities)
Competencies
export experience * * (on export activities)
general capabilities * not tested
marketing capabilities * not tested
product competencies * not tested
53
2.3.3 Managerial Characteristics

Next to ‘firm characteristics’, a second important internal determinant of export
performance are the characteristics of the owner-manager. After all, the features of
the firm as well as the manager are fixed internal conditions, influencing both the
activities implemented by an SME and, subsequently, performance. Especially in
an SME, the manager can be seen as a decision-making nucleus of the firm
(Kuratko & Hodgetts 2001). Almost half of the export performance studies in this
review contain variables concerning the ‘Manager’42. To capture the essence of the
characteristics, we distinguish between (1) objective managerial characteristics,
and (2) subjective or psychosocial managerial characteristics. Objective managerial
characteristics concern both (demographic) descriptives, such as the manager’s
age, and capabilities, such as international experience. On the other hand,
subjective features concern the way the decision-maker views the world, and, thus,
also business and exporting.
2.3.3.1 Objective Managerial Characteristics

The objective managerial variables researched for their impact on export
performance are the manager’s age, education level, nationality, professional and
export experience (here, the manager’s experience is researched and not the firm’s
experience as in the previous section), and language competencies.
Age
Overall, the studies provide no evidence for a significant direct relationship
between age and export ratio (Kaynak & Kuan 1993; Moini 1995; Nakos,
Brouthers & Brouthers 1998). On the other hand, Holzmüller & Kasper (1991), and
Holzmüller & Stöttinger (1996) find a negative indirect impact on export ratio,
with older managers having lower ‘dynamic cultural orientation’, a higher
tendency towards an ‘Y-orientation’, and lower ‘psychic stress tolerance’.
Concerning the impact of age on other performance indicators than export ratio, the
studies reviewed are less univocal and fragmented. That is, the effect on perceived
export satisfaction is non-significant in Evangelista (1994), negative on export
42
Only a part of the studies reviewed focus on SMEs. The rest does not distinguish between
larger and smaller companies. Therefore, the number of studies examining the effect of the
manager is smaller in this sample, compared to an all SME-sample.
54
sales in Kaynak & Kuan (1993), while the results for the impact on profitability are
mixed (i.e. positive in Kaynak & Kuan 1993, versus negative in Nakos, Brouthers,
& Brouthers 1998).
Education
Although some authors do not detect any significant impact of education (Axinn
1988; Kaynak & Kuan 1993; and Evangelista 1994), or find a negative influence
(with export profitability, Kaynak & Kuan 1993), some evidence exists that higher
educated managers perform better on the international market. For example,
Nakos, Brouthers & Brouthers (1998) find a direct positive effect on both export
ratio and export profitability. Holzmüller & Kasper (1991), and Holzmüller &
Stöttinger (1996) also report that a higher educated manager performs better by
having a higher dynamic cultural orientation, while relying more on a Y-orientation
of values. The same picture emerges for the impact of foreign education and
perceived export profitability (Koh 1991): “[T]hese findings imply that exporters
who are…. knowledgeable about exporting and their export markets tend to adopt
marketing strategies that lead to better performance.” Therefore, education might
not have a strong direct effect on performance, but the quality of managers’
exporting decisions varies depending on the manager’s schooling level, with a
higher educated manager taking decisions that are more successful.
Export Experience
Next to the knowledge acquired through education, the manager gains export
experience when internationalizing. For this, several proxies have been used.
Holzmüller & Kasper (1991) find that a foreign nationality and the number of days
spent on business trips abroad43 both impact export ratio positively. Kaynak &
Kuan (1993) confirm this positive impact of traveling abroad on export sales and
export profit margin, but not on export ratio or profitability. They also investigate
the impact of the foreign affiliation of managers, but this only affects the export
ratio. Besides, Axinn (1988), and Nakos, Brouthers & Brouthers (1998) deduce
that firms with managers with more experience abroad achieve higher export ratio
43
Although business trips or international travel can be considered strategic activities of the
manager, these studies categorize the variables as ‘objective managerial characteristics’
(Holzmüller & Kasper 1991) and ‘decision-maker characteristics’ (Kaynak & Kuan 1993).
55
and profitability, whereas both Das (1994) and Evangelista (1994) find that the
number of years lived or worked abroad does not seem to heighten the export
intensity or the level of export satisfaction felt. Das (1994) also concludes that
successful exporters have less past export experience, although the managers are
longer in their position, thus might have more overall business experience.
Similarly, Stump, Athaide & Axinn (1998) find the percentage of managers with
overseas experience to be non-significant for export ratio and profit. According to
the authors, this might be because the impact should not be modeled directly but
indirectly, mediated by export commitment. Lastly, Katsikeas, Piercy & Ioannidis
(1996) find that their respondents do not think that ‘international management
outlook’ (including managerial export experience) is important for achieving the
export goals.
Languages
The fluency with which managers speak foreign languages is another competence
that is closely linked to export experience. Although some record non-significant
results (Moini 1995; Kaynak & Kuan 1993; De Luz 1993), the command of foreign
languages seems to improve export ratio and export profitability both directly
(Kaynak & Kuan 1993; Nakos, Brouthers, & Brouthers 1998), and indirectly by
increasing the firms competitive position, the foreign orientation and the psychic
stress tolerance (Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996).
Overall, the owner-manager of a high performing export company seems young,

well educated on both domestic and international business, with experience abroad
(either by job, courses, or vacations), and speaks his foreign languages fluently.
2.3.3.2 Subjective Managerial Characteristics

The subjective managerial characteristics all evolve around the manager’s general
personality, his attitude towards (operational aspects) of exporting, and his
perceptions and beliefs (as far as they relate to exporting and export performance).
This managerial attitude towards exporting is also labeled attitudinal commitment
or export motivation of management44.
44
Both Evangelista (1994), and Stump, Athaide & Axinn (1998) distinguish between
attitudinal commitment (“an enduring positive disposition of management towards
56
General Personality
First, the findings on the manager’s general personality arebrought together. A
manager who is less rigid, more willing to change, or to accept product-policy risks
generates higher export sales growth according to Dichtl, Köglmayr & Müller
(1990). In addition, the better a manager tolerates psychical stress, the better the
foreign orientation of the manager (‘higher attitude towards stays abroad’ and
‘higher economic world-mindedness’), which in turn increases export performance
(Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996). These variables could
be labeled as risk-taking propensity, or how well equipped the manager is in
handling (extra) uncertainty (for instance from exporting)45. The entrepreneurial
posture tested by Balabanis & Katsikea (2003) goes even further, encompassing
both innovativeness, proactiveness, and risk-taking, and also seems to have a
positive effect on export performance.
Attitude towards Exporting in General

Next to general personality attributes, quite some interest has been given to the
managerial attitude towards the exporting process (also labeled the attitudinal
commitment, or the export motivation). The export motivation of management
relates closely to this attitudinal commitment, which can be either proactive or
reactive. A proactive motivation proves oneself in the manager wanting to pursue
the opportunities in exporting actively, showing commitment, while a reactive
orientation manifests oneself in reacting to (mostly negative) stimuli from outside
the firm. All authors find a positive and significant influence of the attitude towards
exporting on all types of performance measures (Dichtl, Köglmayr & Müller 1990;
Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996; Nakos, Brouthers, &
Brouthers 1998; Stump, Athaide & Axinn 1998). The exception is Evangelista
exporting”), and behavioral commitment (“the manifestations” or “the dedication of

financial and managerial resources”). This behavioral commitment is the translation of the
attitudinal commitment into behavior, and, subsequently, is covered in the section on
‘Export Activities’.
45
Other personality traits often mentioned in small business and entrepreneurship literature,
such as locus-of-control and need-for-achievement (e.g., Kuratko & Hodgetts 2001), have
not been found in the empirical studies reviewed.
57
(1994) who concludes that the attitudinal commitment does not increase the
perceived satisfaction. The literature also shows, that a proactive spirit enlarges
export performance, both directly (Wood & Robertson 1997; Baldauf, Cravens &
Wagner 2000) and indirectly by instigating management to raise prices and to
institute an export department (Koh 1991). Holzmüller & Kasper (1991) proxy
exporting motivation by measuring the respondents’ attitude towards stays abroad,
concluding that a more positive preposition improves export ratio. Lastly, Baldauf,
Cravens & Wagner (2000) find the proactive motive of a manager who wants to
attain international reputation to relate positively to export effectiveness and export
ratio (but not to export sales).Therefore, a proactive attitude of managers towards
exporting, i.e. really wanting to export, turns out to be an important factor in
explaining export performance.
Attitude towards Specific Aspects of Exporting

In addition to the attitude towards exporting in general, managers often have
preformed ideas or attitudes on specific aspects of exporting, such as complexity,
risk, or costs. That is, starting (or continuing) export can be highly complex, risky,
or costly and, therefore, is sometimes unattractive in the eyes of managers. For that
reason, Axinn (1988) regards exporting to foreign markets as an innovation to the
firm. Subsequently, she tests whether the characteristics of the innovation ‘export’
impact export ratio. It appears that the compatibility, trialability, and observability
of exporting have no influence, while a higher perceived relative advantage and a
lower perceived complexity of exporting both improve export performance. Yet,
Katsikeas, Piercy & Ioannidis (1996) find that managers do not think that the
perceived complexity of exporting is of importance for the achievement of export
goals.
Other authors emphasize the risk aspect (either as general risk in exporting or
compared with domestic business risk). The majority concludes that the perceived
risk of exporting has no impact (Kaynak & Kuan 1993; Bijmolt & Zwart 1994;
Moini 1995). Next, Bijmolt & Zwart (1994) do find that higher perceived risk of
geographical distance has a positive impact on export success. Dichtl, Köglmayr &
Müller (1990) focus on psychic distance, and find that managers who experience a
greater-than-average psychic distance to foreign markets or countries have lower
sales growth than their counterparts do.
58
The studies reviewed show a non-significant relationship of perceived costs and

profitability of export business with export performance (Kaynak & Kuan 1993;
Bijmolt & Zwart 1994; Naidu & Prasad 1994; Moini 1995). To put these findings
in perspective, the fact that this review focuses on exporting firms, implies that
firms already set their first step in the exporting process. It is conceivable that the
perceived risk, profitability and complexity of exporting have a stronger impact on
an SME that wants to initiate exporting. This is confirmed by Roux (1987) in his
study on the impact of the manager’s attitudes toward risk on export entry, where
he concludes that “[B]y finding that individual characteristics prevail over firm
factors in predicting the export entry in small and medium-sized firms, this
research reinforces previous behaviorally based studies of exporting.”
Attitude towards Priority of Exporting

Next to these perceptions with regard to the difficulties of exporting, managers
often have a specific opinion on the importance or priority of exporting. Kaynak &
Kuan (1993) observe that when respondents perceive exporting as a major daily
task, both the export ratio and the export profit ratio improve, although the export
sales and export profitability do not alter. The priority given to export policy
related to domestic policy boosts export performance both directly (Bijmolt &
Zwart 1994), and indirectly through the establishment of an export department
(Koh 1991) or by increasing the actual regularity of implementing exporting
activities (Naidu & Prasad 1994)46. By contrast, Diamantopoulos & Inglis (1988)
do not find any significant result on export ratio. Similarly, Katsikeas, Piercy &
Ioannidis (1996) find that the managerial belief that export is important is of no
significance for the achievement of export goals.
Summary Managerial Characteristics

Table 2-10 gives a summary of the relationships found on the manager’s impact on
the export performance. It appears that the objective managerial characteristics
have a strong indirect effect on export performance, both through export behavior,
46
Here, the distinction between the actual regularity with which exporting is performed
(i.e. a strategic decision) and the perceived regularity, which is more attitudinal (e.g.,
Kaynak & Kuan 1993), comes forward.
59
and by shaping the attitude of the manager. It appears that a manager who has more
competencies on the internationalization field, performs better because he better
able to take the right decisions. Next, the subjective attitude of the manager has a
direct and indirect impact as well, mostly through export behavior. The general
picture is that managers who dear to take more risks, and are more proactive or
positive minded towards exporting, take actions that are more suitable and perform
better.
Table 2-10 Summary management characteristics.

direct effects indirect effects
tested tested (on)
Objective characteristics
Age * * (on attitude)
Education * * (on attitude & export activities)
Export experience * not tested
Language fluency * * (on attitude & export activities)
Subjective characteristics
General personality * * (on attitude)
Attitude towards exporting * * (on export activities)
Attitude towards specific aspects of * not tested
exporting
Attitude towards priority of exporting * * (on export activities)
2.3.4 Export Activities

In the previous three sections, I focused on the surroundings and conditions for
exporting, i.e. the environment, firm characteristics and features of the manager.
Together these encompass the external and internal context in which the firm
operates. In this section, the focus is on the firm’s actual behavior in the exporting
process (also labeled ‘export strategy’, ‘export behavior’, or ‘export policy’). This
internal category of export performance determinants is very broad, and has been
researched extensively. Because of the breadth of this category, content analysis
has been performed to find structure in the various strategic variables. After
carefully assessing the various export-related activities as operationalized in the
studies reviewed, the following sub-categorization is proposed and implemented:
1) Firm’s behavioral orientation on exporting;
2) Competitive and expansion strategy;
3) Behavioral commitment of resources:
60
a) (Amount of) planning;

b) Information behavior;
c) Structure;
4) (Standardization / adaptation) marketing mix variables:
a) International channel management;
b) International pricing policy;
c) International product policy, and
d) International promotion policy.
In the following sub-sections, these categories are discussed in depth for their
impact on the different export performance measures.
2.3.4.1 Firm’s Behavioral Orientation on Exporting

This first category relates to the general mode of doing business internationally,
and deals with the way the firm carries out his business abroad, or the firm’s
behavioral orientation on exporting. This can be either proactive, or reactive, and
is, as such, closely linked to the managerial exporting attitude brought forward in
section 2.3.3.2. The main difference is that here the actual activities are being
discussed, instead of the attitude that managers have towards doing business. In
particular, the focus is on the translation of the attitudinal motivation into
behavioral components; the actual manifests of the managerial attitude towards
exporting. For instance, an exporter with a proactive attitudinal motivation
generally searches actively for possibilities and customers abroad, while an
exporter with a reactive motivation by and large starts exporting unintended, e.g.,
after receiving an unsolicited order from abroad.
The mode in which exporting was actually initiated is the first demonstration of
this behavioral translation of a reactive or proactive stance. Beamish, Craig &
McLellan (1993) find minimal proof for the hypothesized negative effect of firms
that began exporting due to an unsolicited order (i.e. reactive). The same can be
said for Diamantopoulos & Inglis (1988) who ask whether export initiation was a
consequence of external factors, or a conscious decision by management. Others
investigate the general outlook on business activities. The overall tenure of these
studies is that proactive, entrepreneurial, or aggressive exporting has a positive
impact on export performance (Leonidou & Kaleka 1998; Thirkell & Dau 1998;
Dean, Mengüç & Myers 2000; Francis & Collins-Dodd 2000; Robertson & Chetty
61
2000). Using contingency theory, Robertson & Chetty (2000) conclude that a
conservative firm in a benign environment, performs equally well as an
entrepreneurial firm operating in a hostile environment. In addition, Francis &
Collins-Dodd (2000) state that “… conservative strategies in turbulent markets
prevent firms from taking advantage of fast-mover opportunities, changing market
entry barriers, or rapidly evolving market life cycles that require a direct presence
and immediate information”. Therefore, an entrepreneurial approach is profitable
in any circumstance, while a conservative outlook only seems suitable in a benign,
more stable environment. On a marketing level, Kotler & Armstrong (2004)
describes a product, production, selling, social marketing, and marketing
orientation. All orientations imply taking different type of decisions on activities.
For instance, the product orientation is mainly concerned with product quality,
while a marketing orientation revolves around finding out more on the needs of the
customer. Shoham (2000) shows that no orientation is optimal under all marketing
conditions. “[W]ith the exception of a selling orientation and, to a point, a societal
marketing orientation, the other orientations appear to be equally viable as drivers
of export performance.” He ends with: “Our results suggest that a marketing
orientation, in and by itself, may be insufficient to enhance all facets of export
performance”.
This all suggests that looking actively for opportunities to market your product
abroad and being close to the market is better than to sit back and wait until you
have to react to outside stimuli. Yet, this is dependent on the market conditions.
2.3.4.2 The Competitive and Expansion Strategy

The general strategic orientation of the firm translates into the competitive and
marketing strategies. When investigating the generic competitive strategies,
Baldauf, Cravens & Wagner (2000) deduce that “…, differentiation strategies are
important for firms’ survival (author: i.e. export profitability), but they are
inversely related to export intensity and export sales.” They explain this finding
assuming that “lower market shares are leading to lower sales”, and that firms deal
with a more demanding customer base. On the other hand, maintaining a low cost
strategy deteriorates export profitability but has no impact on export sales or export
62
ratio47. Besides, the market position strategy selected (i.e. market leader versus
follower) does not affect export sales or export ratio (Ito & Pucik 1993).
Another decision in international marketing is the choice for an expansion strategy,

i.e. whether the firm follows a concentration or diversification strategy. Although
various proxies have been used, the expansion strategy is often measured by the
number of different export countries covered by a firm48. Most studies in the
review find direct and positive relationships between market coverage and export
performance (mostly with export sales and export ratio). Some authors find non-
significant relationships (various performance measures), while only one study
finds a negative relationship (with export ratio) (Diamantopoulos & Inglis 1988;
Lee & Yang 1990; Holzmüller & Kasper 1991; Beamish, Craig & McLellan 1993;
Dominguez & Sequeira 1993; Donthu & Kim 1993; Kaynak & Kuan 1993;
Holzmüller & Stöttinger 1996; Nakos, Brouthers & Brouthers 1998; Stump,
Athaide & Axinn 1998; Dean, Mengüç & Myers 2000). For example, Lee & Yang
(1990) divide their sample into three groups, depending on the expansion strategy
used by the firms. The results show that the market diversification group
outperforms the two other groups (market concentric diversification and market
concentration) in both export intensity and the export sales growth. Other aspects
of diversification researched are the use of market segmentation, the geographical
distance to the foreign market, the number of geographically separated markets, the
number of regions, the diversity of markets, and the export to LDC’s or
industrialized countries, but these are measured only in singular studies
(Diamantopoulos & Inglis 1988; Beamish, Craig & McLellan 1993; De Luz 1993;
Kaynak & Kuan 1993; Shoham & Kropp 1998). For example, Diamantopoulos &
Inglis (1988) have to reject the hypothesis that exporting to many countries
improves performance, but they do find proof that exporting to many regions (as a
bundling of multiple countries) is beneficial for export ratio. Similarly, Kaynak &
47
Surprisingly, Baldauf, Cravens & Wagner (2000) conclude that going international based
on the wish to achieve a competitive cost position positively impacts export effectiveness
and export intensity (but not export sales). This seems contradictory to their results on the
competitive strategies.
48
As mentioned before, 'the number of markets a firm exports to' also serves as a proxy for
export performance and for experience in some studies.
63
Kuan (1993) find the number of geographically separated countries not to be

significant for any export performance measure.
Concluding, the evidence from the studies above does suggest that the more
diversified the exporter is, or the more countries are served, the higher overall
export performance is. In the authors view, the results should be considered with
some caution, as there is tautology in measuring the number of countries exported
to and the export sales or export intensity. Beamish, Craig & McLellan (1993)
acknowledge this: “[I]t cannot be ascertained whether this is a causal relationship
or a representation of an evolutionary cycle.” After all, hardly any significant
results can be found for the impact of the number of countries on export
effectiveness, as most results concern sales or intensity. It might not just be the
absolute number of countries that is important, but the choice for specific countries,
considering aspects such as the industrialization level or the geographical
dispersion of these countries. Exporting to multiple countries might be more
interesting if the countries are somewhat developed and not so widely distributed
(i.e. regions).
2.3.4.3 The (Behavioral) Commitment of Resources

A third level of activities thought to have an impact on the export performance of
firms is the behavioral commitment of resources, or the amount of resources
committed to exporting. For example, Evangelista (1994) finds that translating
attitudinal commitment into actual behavior (measured as a composite of allocated
funds, travel, priority and proactiveness) heightens the managers’ satisfaction with
exporting. Dean, Mengüç & Myers (2000) also find that the lack of focus and
commitment on foreign markets negatively explains the difference in export sales
between high and low performing exporters (but not for export ratio or export
growth). In the exporting process, firms can reallocate these resources either to
export planning, to the collection of export information, to restructuring the work
force especially for the exporting activities, or to all three49. The following
paragraphs discuss these three options of resource commitment.
49
‘Commitment of resources’ is sometimes used as a dependent variable, distinguishing
between experimental, committed, and active exporters based on their degree of
64
Planning
Most strategic management literature stresses that a manager should plan, set
objectives and state corresponding strategies to be able to perform successfully
(e.g., Zwart 1991; Postma & Zwart 1998). The export performance literature
suggests the same, using proxies such as strategic planning, functional planning
and budgeting planning (Samiee & Walters, 1990), the propensity to undertake
export planning (Walters, 1993), the intensity of planning (Madsen 1989; Bijmolt
& Zwart 1994), or export planning as an overall composite (Katsikeas, Piercy &
Ioannidis 1996; Shoham 1996; Shoham & Kropp 1998; Shoham 1999). All but one
show the positive performance impact of allocating resources to general export
planning activities. The exception is Katsikeas, Piercy & Ioannidis (1996), who
find a negative impact on the managers’ perceptions of export performance
objectives of the use of export planning and control (measured dichotomously).
They attribute this result to the lack of specialized personnel in planning and
control in Greek firms, and the need to response immediately to outside stimuli,
which a formal planning structure might prevent. Next, Samiee & Walters (1990)
fail to find any significant impact of planning on export profit, notwithstanding the
improvement in export ratio and the number of export markets. As the authors
write: “Planning activity does not guarantee either good plans or effective
implementation of plans.”
The results show that firms should demonstrate other, more specific, export
planning activities as well: firms benefit from establishing separate export
development budgets, (export) goals, a written export plan, a foreign price policy50,
and a promotion plan for abroad (Beamish, Craig & McLellan 1993; Donthu &
internationalization, which is, among others, measured by export sales, and the
commitment of resources such as structure (e.g., Katsikeas, Deng, & Wortzel 1997)
50
The studies by Koh & Robicheaux (1988), and by Koh (1991) both fail to find a
relationship between determining an export price strategy and the perceived export
profitability relative to that in the home country (as opposed to actual export profitability).
This underlines the often-ignored importance of business plans in the eyes of the managers,
as these authors focus on a subjective measure of export performance instead of an
objective one.
65
Kim 1993; Bijmolt & Zwart 1994; Evangelista 1994)51. Overall, it appears that
SMEs benefit greatly from both general and specific export planning activities.
Use of Information
To plan effectively firms should dedicate resources to the collection and use of
information to reduce uncertainty and to understand the needs of the market (e.g.,
Kotler 2003). As for this information behavior, a useful distinction is that between
export marketing research, export market intelligence, and export market assistance
(cf. Souchon & Diamantopoulos 1996)52.
Several authors conclude that the extent to which firms use (formal) export market
research impacts export performance directly (Dominguez & Sequeira 1993; De
Luz 1993; Bijmolt & Zwart 1994; Moini 1995; Katsikeas, Piercy & Ioannidis
1996; Hart & Tzokas 1999)53. Koh (1991) also examines the indirect effect of
formal information, with frequent and extensive international marketing research
leading to better strategic decisions, which subsequently boosts performance.
Julien & Ramangalahy (2003) establish a positive indirect association from the use
of formal information sources (related to industry, assistance, and written
information) on export performance as well through the competitive strategy.
The results concerning the benefits of export assistance are mixed

(Diamantopoulos & Inglis 1988; Donthu & Kim 1993; Cafferata & Mensi 1995;
Nakos, Brouthers & Brouthers 1998). For example. Diamantopoulos & Inglis
(1988) find that the number of governmental and trade agencies consulted does not
impact export ratio. High involved exporters have a lessened need for outside help,
maybe due to the higher developed internal capacity of these firms compared with
less involved exporters. On the other hand, Donthu & Kim (1993) find that using
more governmental and private agencies companies (summated to one variable)
51
De Mortanges (1990) introduces expert systems to improve the level of (export)
preparedness.
52
Chapter 6 will elaborate further on this threefold categorization of export marketing
information.
53
Although De Luz (1993) finds a non-significant correlation between marketing research
and the 5-year growth of export sales, the sign is positive.
66
improves management’s perceptions of profit. The discrepancy between these

results is probably because the operationalization in both studies is inherently
different. First, private agencies pertain to the category ‘export market research’,
and not ‘export assistance’. This might be the reason for the positive results in
Donthu & Kim (1993). Another reason might be that managers who use many
agencies might perceive this as beneficial (Donthu & Kim 1993). This is not
reflected in export ratio, which is an objective measure (Diamantopoulos & Inglis
1988). Overall, export assistance might be too general to be of use in the course of
internationalization.
Similarly, the results regarding export market intelligence or informal information,

gained through day-to-day business, are mixed, but tend to be positive. Several
authors find that visits to foreign markets are advantageous (Beamish, Craig &
McLellan 1993; Moini 1995; Hart & Tzokas 1999). Moini (1995) states that
“[F]requent visits to foreign markets attending trade shows are absolutely critical in
developing the export markets.” Yet, De Luz (1993) finds that Brazilian managers
deem visits to the export market not to be important, confirming the findings of
Koh & Robicheaux (1988), Koh (1991)54, and Katsikeas, Piercy & Ioannidis
(1996) who all uncover that the frequency (regularity) of face-to-face contacts with
dealers is non-significant.
Next to collecting information, the subsequent use can affect export performance.
Souchon & Diamantopoulos (1997) find that the immediate use of export market
research (export intelligence and export assistance) enhances (weakens) the
international performance. According to the authors, the reason for this lies in the
fact that ‘immediate use’ of the information is measured, while much intelligence
and assistance information is stored for future research and can enhance
performance in the longer run. Reasoning the other way around, Julien &
Ramangalahy (2003) ask respondents how important they consider several
information categories (which is, of course, closely related to the usefulness of the
information type). They find that respondents that perceive a certain type of export
54
Koh (1991) examines only the direct effect of face-to-face contacts, while including both
the direct and indirect effect of formal marketing research. Probably, the author considers
face-to-face contact not as a mean of receiving information, but as channel management.
67
information important, the subsequent use of the respective information sources

increases55.
The overall picture emerging from this literature is that it is beneficial for the
exporting firm to collect data and to use it subsequently in export-related decisions.
Especially important is formal research, with some smaller impact stemming from
export intelligence, while the utility of export assistance is ambiguous.
Restructuring the Organization

Lastly, the behavioral commitment of resources in restructuring the organization is
investigated. This restructuring can be accomplished by committing manpower (i.e.
the number of exporting employees or the amount of time devoted to exporting), or
by establishing a specific structural arrangement for the export activities. Largely,
exporting performance is enhanced by devoting special manpower to exporting
(Diamantopoulos & Inglis 1988; Holzmüller & Kasper 1991; Beamish, Craig &
McLellan 1993; Kaynak & Kuan 1993; Bijmolt & Zwart 1994; Leonidou & Kaleka
1998; Nakos, Brouthers & Brouthers 1998; Stump, Athaide & Axinn 1998).
Spending money on specific export courses for personnel also turns out to benefit
export performance (Bijmolt & Zwart 1994). Nevertheless, Diamantopoulos &
Inglis (1988) rightly question the causality of this relationship found in cross-
sectional studies: “…another plausible explanation is that more staff are allocated
to the export side of business as a result (emphasis by authors) of greater export
activity.” Lages & Montgomery (2001) confirm that previous export performance
is an antecedent of the firm’s commitment to exporting (besides, it impacts the way
managers define their exporting strategies).
The structural arrangement of export activities is usually operationalized as a

statement concerning the presence or absence of a separate export manager or
department (either at home or abroad). Without going into specifics of this
adjustment, both Samiee & Walters (1990), and Bijmolt & Zwart (1994) find that
adjusting the organization structure explicitly for exporting activities boosts export
performance (though a non-significant effect on export profits is found, partly due
55
This perceived importance of information could also be seen as a proxy for the priority
managers attach to exporting (see section 2.3.3.2).
68
to the costs in the adjustment of the organizational structure). More specifically,

this adjustment can be a special manager/executive responsible for export (Donthu
& Kim 1993; De Luz 1993; Bijmolt & Zwart 1994; Evangelista 1994; Beamish et
al. 1999) and/or a separate export department either at home or in the export
country (Diamantopoulos & Inglis 1988; Bijmolt & Zwart 1994; Katsikeas, Piercy
& Ioannidis 1996; Leonidou & Kaleka 1998; Beamish et al. 1999). For example,
Beamish et al. (1999) distinguish between export (1) as part of domestic business
(i.e. no special organizational arrangement), (2) by using a domestic-based export
unit or overseas agents, and (3) by using company-based employees who are based
in export markets, overseas subsidiaries and/or a joint venture. These
internationalization stages (as labeled by the authors) appear not to impact export
ratio, but the latter stages outperform the first on export revenue. In short, the
results show that employing a distinct manager is beneficiary while raising a
separate export unit brings about no significant advantages.
Kaynak & Kuan (1993) also investigate the structure within the export department
(i.e. administrative autonomy, number of hierarchical levels, and the role of
formalization), but find hardly any significant results. Lastly, Katsikeas, Piercy &
Ioannidis (1996) find that managers do not perceive the problem of a poorly
organized export department to be unfavorable in the accomplishment of export
objectives.
When we combine the results mentioned above, devoting manpower especially for
exporting seems rewarding, but this does not necessarily imply that a special
structural arrangement should be employed.
2.3.4.4 Marketing Mix Variables

The last subcategory in export behavior, the (standardization/ adaptation of the)
marketing mix variables, received most attention compared to the other behavioral
categories. The marketing mix includes product, price, place, and promotion. In
international situations, the question arises how to shape these four traditional Ps
for the foreign nations. De Luz (1993) adds the concept of firm reputation, which
can be considered as the summated outcome of all marketing mix activities. De
Luz (1993) finds that managers perceive the overall reputation of the company to
affect the 5-year growth of export sales positively (although the sign is non-
69
significant). Next, Bourantas & Halikias (1990) enter an overall factor ‘export
marketing’. This ‘development of marketing activities’56 affects export ratio
positively. Although named ‘competitive strategy’ by Julien & Ramangalahy
(2003), their strategic construct is rather an overall assessment of marketing
activities (a composite of overall marketing, segmentation, products, distribution,
and promotion), and relates positively to export performance. In contrast to these
overall constructs, most studies focus on specific or individual marketing mix
variables. Within these studies, we can roughly distinguish two main issues,
namely the issue of international channel management and (the standardization or
adaptation of) product, price and promotion.
International Channel Management

The issue of international channel management encompasses two broad aspects,
namely the actual design of the channel, and the relationship with members within
the channel chosen.
‘Channel design’ touches upon issues such as the standardization of the export
channel, the length of the channel, and what specific channel members are used. As
for standardizing the distribution channel design, the results are mostly non-
significant (Beamish, Craig & McLellan 1993; Shoham 1996; Shoham 1999),
implying that adaptation of the channel for the international market is not a
necessity for success. Robertson & Chetty (2000) do find that there should be a fit
between orientation and channel structure (with an entrepreneurial management
channel structure can be either mechanic or organic, while a mechanic structure
outperforms an organic structure in a conservative firm).
When choosing a channel structure, an important aspect is channel length.

Although Leonidou & Kaleka (1998), and Shoham & Kropp (1998) both find very
weak evidence for the need for a short channel, Koh & Robicheaux (1988), and
Koh (1991) find that selling to foreign end-users raises perceived relative
profitability compared to selling through foreign middlemen or a US export agent.
These last two studies also suggest that managers perceive the relative profitability
56
This factor encompasses control of channels, advertising, innovation in marketing
techniques, and inter-state agreements.
70
to be higher when they use their own export unit instead of the domestic marketing
unit or a US export agent. Here, centralization of international channel
management seems to be the case. Likewise, Beamish, Craig & McLellan (1993)
find that it is not important to involve the representatives in coordinating the
marketing plan or to hold them responsible for setting sales goals.
Next to these results on the (functional) structure of the exporting channel, the
choice of the actual channel members is important. Bijmolt & Zwart (1994) show
that listing and evaluating distribution channels and possible trading partners is
favorable. On the other hand, Katsikeas, Piercy & Ioannidis (1996) conclude that
using foreign market entry and customer selection criteria has no impact on the
perceived accomplishment of export goals. This is quite similar to Beamish, Craig
& McClellan’s conclusion (1993) that the investigation of different methods of
distribution only impacts UK export ratio (and not sales or Canadian export
performance). Evangelista (1994) also concludes that less satisfied performers
deem finding and instructing overseas distributors as more limiting than more
satisfied performers.
Concluding, the international channel might not need specific adaptation,

especially when management has an entrepreneurial outlook. Besides, the firms are
probably better off keeping property (and, thus, control) in their own hands, but,
when outsourcing some chain functions, the possibilities should be screened and
evaluated (although evidence for this conclusion is rather weak).
A second part of export channel issues is the relationship between the seller and
the buyer (or import customer), which is investigated quite ad hoc. Leonidou &
Kaleka (1998) use the relational paradigm to explore several aspects within the
relational atmosphere. They find the commitment to, communication in, co-
operation in, dependence in, trust in, and the satisfaction with the relationship to
impact performance positively. On the other hand, the adaptation to, conflict in,
distance in and understanding in the relationship were non-significant57. The
authors state that “…this, in turn, implies that harmonious exporter-importer
57
It must be noted that the respondents in this study reported a low level of distance and
conflict anyway, reinforcing the above results.
71
relationships go hand-in-hand with the firm’s expansion in overseas markets.”

Harmonious relationships are meant to be long lasting. Beamish, Craig &
McLellan (1993) found that maintaining ongoing distribution arrangements is
important for Canadian SMEs, but not for UK firms. Similarly, De Luz (1993)
observes that managers find motivating the distributor or offering promotions
directed at the dealer non-significant in achieving success (although the sign of the
first is positive). In building and maintaining relationships, offering support to
channel members could be a crucial aspect. Beamish, Craig & McLellan (1993)
only find a positive impact of the provided assistance to the foreign representative
for UK SMEs and not for Canadian ones, while both Koh & Robicheaux (1988),
and Koh (1991) find non-significant relationships for the level of dealer support.
Shoham & Kropp (1998) conclude that channel support in the form of visits and
trained personnel positively impacts export sales and the growth in international
profits, although the international profits or the growth in export sales are not
altered. In addition, Thirkell & Dau (1998) find that managers perceive meeting
deadlines, after sales maintenance, physical demonstrations, and training customer
employees to enhance export performance. De Luz (1993) also finds that meeting
the delivery dates is influencing the export sales growth positively. Evangelista
(1994) shows that 'delivery' can pose a problem, by finding that poor performing
exporters think the problem of production and delivery to be more limiting than
more satisfied performers. Contrary, Katsikeas, Piercy & Ioannidis (1996) find that
managers do not perceive the difficulties and costs of exogenous logistic operations
as problematic. On the whole, the relationship between the firm and his (foreign)
channel members seems to be important, and has to be fed by continuing support
and effort, stressing the importance of inter-firm relationships in exporting.
International Product Policy

In exporting, one of the main questions is whether the firm should consider product
adaptation for the foreign markets (e.g., Levitt 1983; Yip 1989; Usunier 2000).
Quite some studies reviewed deal with this question, either by investigating the
adaptation of the product itself or of the product line. Besides, specific aspects of
the product received interest.
Several studies fail to find a significant relationship between product adaptation

(or the counterpart standardization) and export performance (Diamantopoulos &
72
Inglis 1988; Bourantas & Halikias 199058; Koh 1991; Beamish, Craig & McLellan
1993; Shoham 1996). Other studies urge for adaptation or standardization, partly
depending on the performance measure in question. For example, De Luz (1993)
advices firms not to adapt their product to the foreign market, as a nearly
significant positive impact from product standardization on the five-year export
sales growth is found. In addition, the same study concludes that matching
customer specifications or offering special warranties does not relate significantly
with performance, a second argument against adaptation. Refining these results by
using various performance measures, Kaynak & Kuan (1993) conclude that “…
product standardization is conductive to export-oriented corporations pursuing a
bigger share of sales and/or profit from export business while product
differentiation is the driving force that makes the export profitable” for Taiwanese
firms. Yet, Donthu & Kim (1993) do find a customized product to affect export
sales growth positively. In addition, Nakos, Brouthers & Brouthers (1998) find
product adaptation to heighten firm’s export ratio and perceived profitability of
exporting versus domestic business. Shoham (1999) also deduces that a
standardized product worsens static export performance (which is a composite of
export sales, ratio, profitability, and satisfaction), although the link with growth in
performance is non-significant. Johnson & Arunthanes (1995) distinguish between
ideal and actual adaptation of the product. In this, ideal product adaptation
(instigated by government regulations and the market lag, but, surprisingly, not by
differences in end-user preferences) influences actual product adaptation
positively. In turn, actual product adaptation affects sales growth positively, but
bears no relationship with export profit or export market share. As a specific aspect
of product modification, Koh & Robicheaux (1988), and Koh (1991) state that it
does not matter whether a firm uses his own brand name or not.
One could also adapt to market needs, by adding other products or services to the
product line especially for the foreign countries. For instance, Beamish, Craig, &
58
Here, ‘product adaptation’ is a multifaceted factor consisting of the items ‘product
quality control’, ‘experienced and trained personnel’, ‘forecasting market growth’,
‘products in high price market segments’, ‘development of new products’, ‘information
about foreign markets’, and ‘improving existing products’. Therefore, the non-significant
result could be due to the varying impact of the underlying variables on export ratio.
73
McLellan (1993) found that offering complementary products in the foreign market
benefits performance for Canadian exporters, but not for UK firms. According to
the authors, the explanation for this result lies in the ‘blind spot’ that Canadian
exporters have for the limited range of their offering, as they focus on U.S.
markets, versus UK exporters who already offer a much broader range of products
for various countries. Shoham (1996) also focuses on the standardization of
product lines, concluding that adapting the number of product lines enhances
profits, while adapting the number of items in product lines harms profits. Other
studies show that export sales increases with an extensive product line
(Diamantopoulos & Inglis 1988; Beamish, Craig & McLellan 1993; Kaynak &
Kuan 1993; Leonidou & Kaleka 1998).
Other product-related variables are the uniqueness and quality of the export
product. It appears that the product uniqueness is not important, and neither are
sufficient inventories (Beamish, Craig & McLellan 1993; De Luz 1993; Katsikeas,
Piercy & Ioannidis 1996; Thirkell & Dau 1998). Koh (1991) also finds the
management’s perception of the uniqueness of the product offer to be unrelated to
the management’s perceptions of the exporting profitability of exporting versus
domestic. Concerning the product quality, the results are mixed. In Katsikeas,
Deng & Wortzel (1997) the factor ‘product competitiveness’ does not have any
impact on the degree of internationalization, encompassing ‘product quality
control’, but also ‘competitive pricing’, and ‘packaging’. This non-significant
result can be due to the somewhat different nature of the underlying items, such as
price and product packaging. Similarly, Katsikeas, Piercy & Ioannidis (1996)
unexpectedly conclude that a poorly designed product with a poor quality does not
influence the achievement of export objectives. Thirkell & Dau (1998) find
‘product quality’ and ‘meeting the delivery deadline’ to improve an export
performance composite, while the respondents in De Luz’s study (1993) do not
perceive the quality of the product to be important for export sales growth. Yet, De
Luz (1993) does find that meeting delivery dates is of utmost importance in the
managers’ eyes (service in general is not). Therefore, maybe it is not the technical
quality of the product that is important, but the consumer quality, reflected in
keeping deadlines and customer service. To back this up, Beamish, Craig &
McLellan (1993) find positive and very significant results for the perceived
importance of customer service on both performance measures in both countries.
74
Yet, Kaynak & Kuan (1993) advise high performers on export profitability to
localize after sales service to a lesser extent. They also find that with a
technologically intense export product export sales drop dramatically, although an
innovative export product impacts all export performance proxies positively:
“Taiwanese firms successful in exporting turn out to be those who are innovation-
conscious”59. De Luz (1993) confirms this negative importance of an advanced
technology product, although Beamish, Craig & McLellan (1993) find mostly non-
significant results.
Overall, the results concerning product adaptation are far from conclusive,
although the inclination is towards adapting and extending the product offer
towards the needs of the foreign market. Besides, the product should not be too
technological for a fast penetration of the market. Diamantopoulos & Inglis (1988)
do remark that a distinction should be made between product adaptation due to the
product offering versus the pressure of legal requirements of the market. After all,
these last are unavoidable adaptations, and, as such, no discriminating
determinants. Johnson & Arunthanes (1995) adhere to this distinction in the
antecedents, but only find proof for the impact of legal requirements, and not for
market pressures.
International Pricing Policy

The next market instrument discussed is the price of the product. Again, the results
are mixed. For instance, Shoham (1996) shows that companies should adapt their
price to improve profits and profit growth, although there is no evidence for sales
or sales growth. In a later study, Shoham (1999) investigates a composite
performance measure, including both sales and profit variables, leading to the
advice to standardize price to improve performance growth, although static
performance is not impacted at all. These contradictory results are probably due to
the non-consistent operationalization of performance. Firms can adapt their price to
59
Yet, the export-related R&D expenditures do not have a significant influence in their
study. These results are somewhat contradicting, but we might say that Taiwanese exporters
should offer innovative products, that are not too technologically intense (this could also be
related to the negative impact they found on the image of Taiwanese products).
Unfortunately, the authors do not provide any further reason for these findings.
75
the competitors in the market, or to consumers’ wishes. Bourantas & Halikias

(1990) conclude that being price-competitive is rewarding for export ratio, which is
backed up by Holzmüller & Kasper (1991), and by Holzmüller & Stöttinger
(1996). Kaynak & Kuan (1993) also find that ‘pricing discrimination’ improves
export profitability, although the study does not define ‘discrimination’, i.e. higher
or lower than domestic price or competition. Contrary, De Luz (1993), and Shoham
& Kropp (1998) find that a low price negatively influences export sales. According
to Koh & Robicheaux (1988) and Koh (1991), setting foreign prices higher than
domestic prices also contributes to the profitability of the firm. Yet, Katsikeas,
Piercy & Ioannidis (1996) find that managers do not perceive the lack of
competitive prices to be significant in achieving export goals60. The review also
shows that it does not really matter which export pricing currency or quotation
strategy is chosen and whether the prices are set by the manufacturer or by the
channel members (Koh & Robicheaux 1988; Koh 1991; Beamish, Craig &
McLellan 1993; Kaynak & Kuan 1993). In short, adapting the price for the foreign
market and the relative price asked might affect various performance measures
differently.
International Promotion Policy

The fourth, and last traditional marketing instrument is promotion. Questions
researched involve the promotion adaptation and the effectiveness of certain media.
Concerning the adaptation of promotion, Shoham (1996) shows that the budget
should be standardized: “[I]t may be that budgets have a more universal appeal.”
Although the budget calls for standardization, Shoham (1996; 1999) clearly shows
the positive impact of adapting promotion (in advertising content) on export sales,
export profits, and the growth herein. This is partly backed up by Kaynak & Kuan
(1993). Yet, Beamish, Craig & McLellan (1993) demonstrate the strong positive
effect of using the own sales force on export ratio, pleading for standardization.
When choosing for adapting promotion, knowledge of the market is indispensable.
Thirkell & Dau (1998) find cultural affinity, which includes both ‘command of
foreign languages’ and ‘effectiveness of the promotion campaign compared to
60
We have to keep in mind that the authors place this item under a total factor that also
includes cost- and finance-related items, as well as the level of international competition.
This could be the reason for the non-significance of the results.
76
competitors’, to be an important determinant of export performance. They say that

“…higher levels of language and cultural skill may translate into a more effective
communications program”, leading to an adaptation view. Exporters do perceive
information communication with the foreign market problematic in achieving their
export goals (Katsikeas, Piercy & Ioannidis 1996). After the decision for
adaptation or standardizing the promotion, the next step to take in the decision
process is which specific media to use in the promotion campaign. It appears that
managers do not perceive promotion directed at the end-user or trade show exhibits
important for export sales growth (De Luz 1993). Similarly, advertising
(encompassing ‘heavy advertising’, ‘building brand awareness’, and ‘building a
reputation’) seems not important for export sales or profit. It might even harm
international profit growth for US manufacturing exporters (Shoham & Kropp
1998). The reason for this finding might be the self-selection effect, as the sample
mainly existed of firms with low overall advertising and promotion levels.
Considering the above, the results on the international promotion policy are still
rather fragmented and do not point to a specific direction.
Summary Export Activities

Table 2-11 gives a short outline of the relationships found between the activities
associated with exporting and export performance. The exporting activities that
(possibly) affect export performance is quite diverse, leading to a long list of items.
Yet, the results for most items are rather sparse and often ambiguous.
Table 2-11 Summary export activities.
tested tested (on)
Active / reactive orientation * Not tested
Competitive strategy * Not tested
Expansion strategy * Not tested
Commitment of resources * Not tested
Planning * Not tested
Information collection * * (on quality decisions)
Information use * * (on information collection)
Manpower devoted * Not tested
Special export structure * Not tested
Marketing mix (reputation) * Not tested
International channel management
Channel adaptation * Not tested
Channel length * Not tested
77

tested tested (on)
Channel members * Not tested
Relationships in channel * Not tested
Support to dealers * Not tested
Meeting delivery dates * Not tested
International product policy
Product adaptation * Not tested
Product line adaptation * Not tested
Export product uniqueness * Not tested
Export product quality * Not tested
Export product technology * Not tested
International pricing policy
Price adaptation * Not tested
Price level * Not tested
International promotion policy
Promotion adaptation * Not tested
Media * Not tested
Most results can be found for product adaptation, the structure of the organization,
and the channel used. Yet, drawing explicit conclusions is almost impossible in
view of the differing operationalizations of the items and/or export performance. A
possible conclusion is that the impact of these strategic export activities on export
performance is at least direct, except for information, which can also improve other
strategic decisions. Unfortunately, other indirect effects have not been tested.
2.4 Towards an Integral Model

The next step is to integrate the findings from the review into an integral export
performance model. For simplicity, only the overall categories are shown in the
integral model, the detailed items specifying each concept can be found in Table
2-8 through Table 2-11.
78
Firm Characteristics
See table 2-8
Objective Managerial
Characteristics
See table 2-9
Export Performance
See section 2.2.1.5
Subjective Managerial
Characteristics
See table 2-9
Export Activities
See table 2-10
Environment
See table 2-7
Figure 2-4 An Integral Export Performance Model.
Compared to the specific frameworks that came forward from the reviews in
section 2.1 (i.e. by Madsen 1987; Aaby & Slater 1989; Gemünden 1991; Leonidou,
Katsikeas & Samiee 2002), this model is more complete in both the items
pertaining to the concepts, and the relationships between the concepts. Table 2-12
shows the specific differences between the present integral model, which is
specifically based on the empirical studies reviewed, and the previous frameworks.
Table 2-12 Comparison present integral model with previous frameworks.

Concepts Relationships (with concept labels as used by the
original authors)
Madsen (1987)
Added: manager (obj. & subj.) Added: ‘E-structure’ => ‘Strategy’
Added: subj. measures in ‘perform.’ Added: ‘O-structure’ => ‘Strategy’
Added: ‘Manager’ => ‘Strategy’
Added: ‘Manager’ => ‘O-performance’
Aaby & Slater (1989)
Added: manager (obj. & subj.) Added: ‘environment’ => ‘strategy’
79
Added: ‘firm char.’ & ‘manager’ for ‘firm Added: ‘firm char.’ => ‘perform.’
char.’ & ‘competencies’
Not: non-outcome variables in ‘perform.’
Gemünden (1991)
‘Managers’ into obj. & subj. part Added: ‘obj. man. char.’ => ‘subj. man. char.’
Added: subj. measures in ‘export success’ Added: ‘manager’ => ‘export success’
Added: ‘home country’ => ‘export success’
Not: ‘home country’ => ‘managers’
Not: ‘export market char.’ => ‘managers’
Not: ‘firm char.’ => ‘managers’
Leonidou, Katsikeas & Samiee (2002)
Added: ‘obj. man. char.’ => ‘subj. man. char.’
Added: ‘managerial char.’ => ‘export perform.’
Added: ‘organizational factors’ => ‘export perform.’
Added: ‘environmental forces’ => ‘export perform.’
Abbreviations: obj. = objective, subj. = subjective, char. = characteristics, perform. = performance,
man. = managerial.
Shortcomings in the existing reviews (and the empirical studies) are, firstly, the
neglect of the manager, and the lumping together of all items under the heading
‘manager’, although our review clearly shows that there is a notable distinction
between objective and subjective managerial characteristics. Especially in the case
of SMEs, the manager maintains a pivotal place in the company. Secondly, the
measurement of export performance in mere objective outcomes (Madsen 1989;
Gemünden 199161) or even in non-outcome variables (e.g. ‘export problems’ in
Aaby & Slater 1989) is salient. Thirdly, various direct or indirect relationships
between environment, firm, or manager-related factors on the one hand and export
activities or export performance on the other hand found in the present review are
neglected in the existing frameworks, although Gemünden (1991) does add three
indirect relationships through ‘managers’. These are not only shortcomings in the
frameworks, but also in the studies reviewed in this synthesis. Lastly, often not all
items in a category are examined, or are misplaced.
So, most studies use only a sub-part of all possible determinants, and do not
consider valid indirect effects of determinants on export performance only by
estimating all these variables and relationships simultaneously, a valid inference
61
It must be mentioned that Gemünden (1991) does talk about attitudinal measures for
export performance, but he does not incorporate them in his ‘theoretical framework’.
80
can be made concerning the influential factors for export performance. This could
also be a reason for the fact that the results found in the empirical studies, even on
one and the same variable, are so dispersed. If not all relationships and items are
estimated simultaneously, results are biased, as important aspects of the model are
left out.
81
82
Chapter 3
The Interstratos Data Set
3.1 Introduction
The empirical part of this dissertation draws on data gathered through an

international mail survey by the INTERSTRATOS group (INTERnationalization of
STRATegic Orientation of Small- and medium-sized enterprises), with which I
operationalize the integral framework found in the previous chapter. Appendix 1
displays the variables in this survey. As mentioned in chapter 1, the purpose of
INTERSTRATOS is to explore and describe the strategies for adjustment to changes
in environments for small- and medium-sized firms in Europe. This longitudinal
research project into the internationalization of manufacturing SMEs in Europe
covers an annual survey research in seven European countries, i.e. Austria,
Belgium, The Netherlands, Switzerland, Norway, Sweden and Finland62. The
project is designed in such a way that participating researchers from each country
independently collect and analyze data from the commonly designed
questionnaire63. After data collection, the researchers have access to the common
database for different analyses. This INTERSTRATOS data set deems an SME as a
private, independent company with less than 500 employees64.
62
Originally, Great Britain also participated in the survey, but they dropped out in 1993.
63
The Dutch data are collected and financed by EIM Small Business Research and
Consultancy, Zoetermeer, the Netherlands.
64
This is a somewhat broader range than the definition in the EU (250 employees)
nowadays, but when starting the INTERSTRATOS project this was 500 employees.
83
3.1.1 Research Model

As the name of the project implies, the focus is on the internationalization process
of small- and medium-sized enterprises. The founders of the INTERSTRATOS group
state the following on the research model behind this internationalization process:
“From the outset we accept that the performance of SMEs is to a large extent
dependent on external conditions, e.g., resource availability, but we also
accept the view that through a planned, rational process of strategy formation
a firm will increase the odds in its favor. Given less than perfect competitive
conditions, it is the entrepreneur’s ability to take advantage of market
imperfections that determines the firm’s competitive advantages.”
In this, elements of the resource-based view can be found (Pfeffer & Salancik
1978), the strategic choice approach (Child 1972), and an emphasis on the
entrepreneur (e.g., Kuratko & Hodgetts 2001).
The specific research objectives of the research group are to explore and describe
four broad topics (Haahti, Hall & Donckels 1998):
• How do changes in competitive conditions, changes in market factors and
constraints and inducements to undertaking international operations influence
changes in product/market strategies and especially export orientation and the
degree of cooperation of small- and medium-sized manufacturing firms?
• What patterns of behavior relevant to modes and methods of
internationalization can be observed and how do adaptive changes in this
behavior affect changes in the export success of firms?
• What kinds of resource formation in terms of learning capability can be
observed in the SMEs in their growth vector from local to overseas markets
and how do these learning capabilities influence their choices and mode of
operations in the local market and in their international operations?
• Are changes in managers’ values and attitudes related to changes in their
adjustment to the demands of their task environment, and what type of learning
behavior can we observe?
3.1.2 Limitations
The origin of the data automatically implies that the author had no scientific
control over the type of variables or the number of variables included in the
questionnaire. Nor did I have any hand in the sampling procedure. Therefore, the
84
The INTERSTRATOS Data Set
data set is considered secondary data, with all shortcomings of this type of data.
Secondly, due to the broad interest of the research group and the broad set of
research questions, the survey is rather fragmentary in some instances.
3.2 Methodology
3.2.1 Population and Sampling
The population for this project includes small and medium-sized manufacturing
firms in Europe, with size varying between zero (only the entrepreneur him- or
herself) and 500 employees. The definition of the industry sectors included is in
two digit NACE65 codes, valid version 1991.
The following five manufacturing branches are included66:

NACE 43 & 45: textile and clothing
NACE 34: electronics
NACE 41 & 42: food, drink & tobacco
NACE 46: wooden & timber products, including furniture
NACE 31 & 32: metal products & mechanical engineering
The selection of these sectors leads to a diverse representation of industries,

consisting of sectors already internationalized to a large extent and sectors that are
not; sectors vulnerable for technological developments and sectors that are not;
sectors that produce consumer goods and sectors that produce capital goods;
sectors susceptible to economies of scale and sectors that are not. With this
selection, the mix of branches is thought to be representative of other branches than
the five included in the sample. Quoting one of the initiators of the Interstratos
project on the choice for these five industries (Haahti, Hall & Donckels 1998):
“The choice of industries selected for sampling was based mainly on pragmatic
considerations. Most members of our team had experience of three industries
(Stratos 1990), namely electronics, food, and clothing and textiles. To ascertain
65
NACE = Nomenclature des Activités Economique, valid version 1991.
66
In the Netherlands, this coincides with SBI-codes 20/21 (‘voeding’), 22/23/24
(‘textiel/kleding/leder’), 25 (‘houten meubelen’), 34/35 (‘metaalproducten/machine-
industrie’), and 36 (‘elektrotechnische industrie’), respectively.
85
differences in industry growth trends, strategic behavior and forms of

international operations, mechanical engineering and furniture making were
included. All these industries are sufficiently well represented in the respective
countries to form a basis for international comparisons.”
Next, the SMEs were drawn in a stratified manner, which means that the design
guarantees that at least 20 companies are present in strata consisting of the five
sectors, and five different size classes (1 – 9, 10 – 19, 20 – 49, 50 – 99, 100 – 499
full time employees) in the seven countries. For every country 500 observations
were expected.
The population universe was accessed through country-wise sampling frames

composed of individual firms as sampling units, and listed in national statistical
bureaus or other similar sources where available.
3.2.2 Survey
A master questionnaire was developed through a detailed outline of variables listed
in English, followed by a German questionnaire tested in an Austrian pilot study.
Subsequently, this was translated into all participating languages, i.e. Dutch,
Finnish, French, Norwegian, and Swedish. In some questionnaires, country-
specific questions are included, but these are not taken into consideration in this
thesis. The basic questionnaire contains the following subjects:
- The company;
- The manager/owner;
- Market position;
- Use of external support systems;
- Product/market strategies;
- Environmental changes, national and international;
- International activities;
- National and international co-operation;
- Changes in international activities;
- Attitudes on international activities.
The questions have been gathered yearly, over five years, covering the period 1991
– 1995, trying to gather a panel data set.
86
3.3 Response
As is common with longitudinal data gathering, some companies are lost along the
way, due to non-response or mortality (closure of the enterprise) (Menard 2002).
Therefore, the sample does not have the same size in all years, and the sample does
not consist of the same companies each year. Table 3-1 represents the total number
of participants per year, with this number varying between 3557 (1994) and 5231
(1992).
Table 3-1 Total Number of Respondents per Year.
Year 1991 1992 1993 1994 1995
Number of respondents 4744 5231 4373 3885 3557
Apart from the division of enterprises over the five years, it is interesting to look at
the distribution of SMEs over the three strata, namely country, sector and size
class. Table 3-2, Table 3-3, and Table 3-4 show these distributions.
As Table 3-2 shows, Austria, the Netherlands, and Sweden are the countries
represented most in the data set. The distribution of respondents over the countries
has remained rather stable, apart from some deviations (e.g., the number of Dutch
respondents in 1992).
Table 3-2 Number of Observed Enterprises per Country in Yearly Surveys.
Country 1991 1992 1993 1994 1995
Austria 1239 (26.1%) 1312 (25.1%) 1131 (25.9%) 1084 (27.9%) 868 (24.4%)
Belgium 502 (10.6%) 248 (4.7%) 315 (7.2%) 327 (8.4%) 301 (8.5%)
Netherlands 978 (20.6%) 1614 (30.9%) 1026 (23.5%) 647 (16.7%) 597 (16.8%)
Switzerland 419 (8.8%) 466 (8.9%) 434 (8.3%) 438 (11.3%) 505 (14.2%)
Norway 355 (7.5%) 354 (6.8%) 344 (7.9%) 373 (9.6%) 329 (9.2%)
Sweden 908 (19.1%) 838 (16.0%) 835 (19.1%) 720 (18.5%) 727 (20.4%)
Finland 343 (7.2%) 399 (7.6%) 288 (6.6%) 296 (7.6%) 230 (6.5%)
Total 4744 5231 4373 3885 3557
Clearly, the mechanical engineering sector is best represented in the

INTERSTRATOS data set over the five years (see Table 3-3). The electronics
businesses are somewhat less represented, while the other three sectors are more or
less evenly distributed over the data sets.
87
Table 3-3 Number of Observed Enterprises per Sector in Yearly Surveys.

Sector 1991 1992 1993 1994 1995
Textiles 948 (20.4%) 885 (17.2%) 724 (17.1%) 717 (19.3%) 620 (18.1%)
Electronics 703 (15.1%) 763 (14.9%) 656 (15.5%) 601 (16.2%) 516 (15.1%)
Food 839 (18.0%) 906 (17.6%) 800 (18.9%) 675 (18.2%) 631 (18.4%)
Furniture 936 (20.1%) 986 (19.2%) 825 (19.4%) 741 (19.9%) 649 (19.0%)
Engineering 1223 (26.3%) 1594 (31.0%) 1237 (29.2%) 983 (26.4%) 1005 (29.4%)
Total 4649 5134 4242 3717 3421
Approximately half of the SMEs in the data set employ less than 20 employees,
while only on average 12% of the SMEs belong to the largest size class (Table
3-4). Another remark concerns the growth in average number of full time
employees working in the enterprises over the five years. Naturally, the size
categories of the individual companies change over time: firms grow. On the other
hand, some firms experience a size-down. This can also be related to the
‘mortality’ or attrition problem of longitudinal data, representing drop outs due to
not completing or returning the questionnaires, bankruptcies, liquidation or close
down. Especially the participants that responded more than once can be labeled
‘survivors’. As said before longitudinal and especially panel studies in general
suffer from survey non-response (Finkel 1995; Menard 2002).
Table 3-4 Number of Observed Enterprises per Size Class in Yearly Surveys.
Size 1991 1992 1993 1994 1995
1–9 1447 (31.2%) 1709 (33.2%) 1369 (31.8%) 1146 (30.1%) 976 (28.0%)
10 – 19 925 (19.9%) 1017 (19.7%) 869 (20.2%) 729 (19.2%) 683 (19.6%)
20 – 49 1052 (22.7%) 1132 (22.0%) 956 (22.2%) 921 (24.2%) 845 (24.2%)
50 – 99 629 (13.5%) 684 (13.3%) 572 (13.3%) 547 (14.4%) 523 (15.0%)
100 – 499 590 (12.7%) 612 (11.9%) 533 (12.4%) 462 (12.1%) 464 (13.3%)
Total 4643 5134 4299 3805 3491
This non-response problem is clearly present in the INTERSTRATOS data set as well:
only a small part of all participants co-operate during all five years. Table 3-5
represents the participation rate, displaying that almost 50% of the respondents
answered more than once. Logically, more than 50% of the total number of
enterprises that entered the data set did only return the survey once.
88
Table 3-5 Participation Rate of the Respondents.

Participation rate Number of enterprises Percent of participants
1 year 5430 50.4
2 years 2253 20.9
3 years 1339 12.4
4 years 948 8.8
5 years 809 7.5
Total 10779 100
This attrition problem can be caused by discontinuation of companies (which is

natural, especially in the highly volatile SME population), but moreover and more
seriously, by survey non-response. As Table 3-5 shows, only 809 out of 10,779
SMEs responded consequently over five years. Unfortunately, the secondary nature
of the data set makes it impossible for the present research to uncover which part is
caused by survey non-response or by disclosures.
Next to this survey non-response, the data set suffers of substantial item non-
response, especially on objective economic performance items such as ‘total sales’
and ‘export sales’. Considering the fact that this dissertation focuses on export
performance, this item non-response poses serious problems.
A last problem is the lack of Dutch data in 1994. In the Netherlands, data collection
only took place in 1991, 1992, 1993, and 1995, due to financial considerations. To
solve this problem, the Dutch data for 1994 (n = 647) has been extrapolated using
the other years, but this reduces the reliability of these values greatly67.
3.4 Choice for Data Set

All these problems with survey and item non-response have to be taken into
consideration, when choosing the sample for this dissertation. Therefore, in this
dissertation, I only use those SMEs that responded in 1991, 1993 and in 1995, to be
called a panel data set. This leads to a larger data set than the five-year panel data,
and to the inclusion of valid data of Dutch SMEs. A total number of 1125
enterprises entered the database in these three years, a substantially larger number
than the 809 that participated for five years. The reason for choosing three years
67
So, the true net sample for 1994 is 3885 – 647 = 3238 enterprises.
89
instead of four years is the equality of the intervals between 1991 and 1993, and
between 1993 and 1995. Next, this data set includes the most recent data (1995).
For the one-year integral model in Chapter 5, I use the 1995 data, while the study
on information behavior in Chapter 6 uses the whole 1991 – 1993 – 1995 data set.
Table 3-6 up to Table 3-8 show the descriptives of this panel data set used in this
dissertation. These descriptives are compared with the cross-sectional data sets of
1991, 1993, and 1995.
Table 3-6 clearly shows that the quality of data collection differs considerably
between the seven countries. Although some percentages do not deviate too much
from the cross-sectional percentages over the three years (e.g., Switzerland,
Norway, Finland), some countries noticeably have few panel respondents (i.e. the
Netherlands), while others have a very loyal respondent base over the years (i.e.
Austria, and Sweden). As there are hardly enough SMEs in some countries to make
valid inferences on countries separately, I consider the panel data base as
comprised of European SMEs, instead of as a data base with seven segments.
Table 3-6 Number of Observed Enterprises per Country in Data Set 91-93-95.
Country Panel data set 1991 1993 1995
Austria 427 (38.0%) 1239 (26.1%) 1131 (25.9%) 868 (24.4%)
Belgium 36 (3.2%) 502 (10.6%) 315 (7.2%) 301 (8.5%)
Netherlands 40 (3.6%) 978 (20.6%) 1026 (23.5%) 597 (16.8%)
Switzerland 129 (11.5%) 419 (8.8%) 434 (8.3%) 505 (14.2%)
Norway 94 (8.4%) 355 (7.5%) 344 (7.9%) 329 (9.2%)
Sweden 336 (29.9%) 908 (19.1%) 835 (19.1%) 727 (20.4%)
Finland 63 (5.6%) 343 (7.2%) 288 (6.6%) 230 (6.5%)
Total 1125 4744 4373 3557
Concerning the distribution of the panel data set over the sectors, Table 3-7 reveals
that the differences between the panel data set and the three cross-sectional data
sets are less obvious than with the countries, maybe with the exception of the
textiles (less respondents in panel) and the furniture sector (more respondents in
panel).
90
Table 3-7 Number of Observed Enterprises per Sector in Data Set 91-93-95.
Sector Panel data set 1991 1993 1995
Textiles 176 (15.6%) 948 (20.4%) 724 (17.1%) 620 (18.1%)
Electronics 148 (13.2%) 703 (15.1%) 656 (15.5%) 516 (15.1%)
Food 220 (19.6%) 839 (18.0%) 800 (18.9%) 631 (18.4%)
Furniture 287 (25.5%) 936 (20.1%) 825 (19.4%) 649 (19.0%)
Engineering 294 (26.1%) 1223 (26.3%) 1237 (29.2%) 1005 (29.4%)
Total 1125 4649 4242 3421
Table 3-8 displays the number of respondents in the respective size classes within
the panel data set each year, compared with the cross-sectional data sets. Small
differences exist, mostly that the panel represents the size class 1-9 more, while the
opposite applies to the largest size class (100-499).
Table 3-8 Number of Observed Enterprises per Size Class in Data Set 91-93-
95.
Size Panel 1991 1991 Panel 1993 1993 Panel 1995 1995
1–9 393 (35.2%) 1447 (31.2%) 422 (37.5%) 1369 (31.8%) 415 (37.1%) 976 (28.0%)
10 – 19 235 (21.0%) 925 (19.9%) 228 (20.3%) 869 (20.2%) 210 (18.8%) 683 (19.6%)
20 – 49 250 (22.4%) 1052 (22.7%) 256 (22.8%) 956 (22.2%) 270 (24.2%) 845 (24.2%)
50 – 99 135 (12.1%) 629 (13.5%) 119 (10.6%) 572 (13.3%) 119 (10.6%) 523 (15.0%)
100 – 499 105 (9.4%) 590 (12.7%) 99 (8.8%) 533 (12.4%) 104 (9.3%) 464 (13.3%)
Total 1118 4643 1124 4299 1118 3491
This panel data set of 1125 respondents, consisting of the years 1991 – 1993 –
1995 is the basis for chapter 4 (operationalization of the theoretical framework),
chapter 5 (estimation one-year integral export performance model), and chapter 6
(estimation of the effects of managerial personality and information behavior on
export performance, both static and longitudinal).
91
92
Chapter 4
Operationalizing the Integral
Export Performance Model.
In Chapter 2, I presented an integral export performance model, based on an
extensive literature review. This framework describes four areas, which affect the
company’s export performance. Next, chapter 3 introduced the INTERSTRATOS
project. This project resulted in a three-year empirical panel data set of 1125 cases,
which I use to estimate the theoretical model. The present chapter aims firstly to
combine the literature and the empirical data, by operationalizing the concepts
found in literature using the variables in the INTERSTRATOS questionnaire. This is a
first, exploratory, step in analyzing the suitability of the empirical data for the
theoretical model, resulting in a data set with only those INTERSTRATOS variables
that operationalise the items in the theoretical framework from Chapter 2. In the
second step, the missing values in the panel data set are imputed. As explained in
the previous chapter, the INTERSTRATOS panel data set suffers from item non-
response, especially on the performance measures. Therefore, the second aim of
this chapter is to make the operationalization variables full rank (i.e. without
blanks) by imputing the missing values, using the NORM program (Schafer 1997).
Combining these two aims, this chapter results in multiple full-information panel
data sets containing the INTERSTRATOS variables used to operationalize the
theoretical model.
4.1 Operationalizing the Theoretical Model

4.1.1 Methodology
First, the variables in the INTERSTRATOS panel data set are examined for their
suitability in operationalizing the theoretical concepts found in Chapter 2:
• Environment;
93
• Firm Characteristics;
• Managerial Characteristics;
• Export Behavior, and
• Export Performance.
For this, I use the original panel data set, encompassing all possible variables from
the INTERSTRATOS survey, with variables possibly incomplete due to item non-
response. In this manner, only information available in the original data matrix
actually provided by the respondents is used. This exercise not only gives a good
indication of the structure in the original response, but also leads to a list of specific
survey variables for operationalization. As only these variables need to be imputed,
this is more efficient than imputing the comprehensive variable list68. All variables
are examined for their theoretical and analytical suitability in measuring the
concepts found in the literature. Subsequently, some items in the comprehensive
variable list are excluded from further analysis, either on theoretical grounds (i.e.,
the intrinsic meaning of the item is not linked to any of the concepts in the
proposed conceptual model), or on analytical grounds, as will be explained in the
following sections. In the end, the panel data set to be imputed consists of only
those variables, that are to be used either as stand-alone items, as manifest items of
constructs (or latent variables), or as items to be used in constructing an index.
4.1.1.1 Stand-Alone Items and Indices

Depending on the nature of the theoretical concepts in the integral model (i.e.
individual item, index or construct) and the matching variables in the panel data
set, the variables are examined by using descriptive statistics, factor analysis, or
reliability analysis. First, if a theoretical concept is operationalized with a stand-
alone item or as an index using individual items, the variable(s) is (are) assessed
using descriptive statistics, i.e. variance, (the amount of) missing values, and
68
The program used for multiple imputation (NORM) needs considerable computation
capacity (see section 4.2). So, to be more efficient, not the full INTERSTRATOS data matrix is
imputed, as this procedure takes considerable effort, but also complicates the imputation
procedure, with the threat of non-convergence of the estimation procedures. After all, with
every extra variable to be included in the imputation procedure, the number of
interrelationships between all variables increases exponentially.
94
Operationalizing the Integral Export Performance Model
deviations from normality (skewness statistics). Besides, if applicable, the

correlations between single items are measured, to control for multicollinearity in
the resulting data set.
4.1.1.2 Multiple Item Constructs

Concerning multiple item constructs, the appropriate methods are either reliability
analysis (RA) (using Cronbach’s alpha), or exploratory factor analysis (EFA)
followed by RA. The factors found are confronted with similar results found in the
literature, to check whether other studies support the results found in this data set.
The choice between RA and EFA depends on whether the assumption of one-
dimensionality is plausible. After all, when calculating Cronbach’s alpha the
researcher assumes one-dimensionality in the items to be analyzed, which is only
plausible if the items have an acceptable fit on a single-factor. After all,
Cronbach’s alpha does not ensure one-dimensionality but instead assumes it exists.
As there is not always enough theoretical ground to justify this assumption, Hair et
al. (1998) encourage researchers to perform one-dimensionality tests on multiple-
indicator constructs before assessing their reliability.
Reliability Analysis
Consequently, if the set of items share a common theoretical ground, we use RA
and start calculating Cronbach’s alpha on the items representing those facets
straight away. In this, alpha will be calculated together with the loss or gain in
alpha when a specific item is dropped, which is a assessment of the necessityt to
include the item or not. Nevertheless, when purifying the constructs, the theoretical
context is considered as well, as one should not pursue a blind filtering of items.
For the preferred value of alpha, the literature mentions several cut-off points.
Although the generally agreed upon lower limit for alpha is .70, mostly these are
for well-researched and validated constructs. Therefore, Nunnally (1978) indicates
that acceptable reliability may decrease to .60 or .50 in exploratory or early stages
of research. As most of the constructs used in export marketing research are still
(very) exploratory69, the cut-off point used here is .50.
69
Just consider the variety of ways measuring a concept as export performance or other
concepts as explained in Chapter 2.
95
Exploratory Factor Analysis

If a set of manifest items does not have an obvious common theoretical ground,
EFA is used to group variables together70. The suitability of the set of items for
EFA is tested by means of Bartlett’s test of sphericity (p-value should be lower
than .05) and the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO),
which should be over .7071 (Hair et al. 1998). Next, Principal Component Analysis
(PCA) is used for the extraction of factors72, with VARIMAX rotation to simplify
the columns of the factor matrix (i.e. variables scoring high on only one factor),
easing the interpretation of factors73. The number of factors to be extracted is based
on the assessment of the eigenvalues (preferably over 1), the amount of variation
(preferably over 60%, which is satisfactory in social sciences), and the ease of
interpretation (Cattell & Vogelmann 1977; Hair et al. 1998; Churchill 1999,
p.849)74. Furthermore, we use Horn’s parallel analysis (Lautenschlager, Lance &
Flaherty 1989). In this, observed eigenvalues are compared with criterion
eigenvalues of a correlation matrix of random variables. One retains only those
factors with eigenvalues greater than the corresponding criterion eigenvalues of the
comparison random matrix. The criterion eigenvalues for PCA are computed by
ln(λk) = ak + bk ln(N-1) + ck ln{(p-k-1)(p-k+2)/2} + dk ln(λk-1) + ek p/N, where ln =
natural logarithm, N = sample size, p = number of items, k indexes the kth
eigenvalue λk, and ak-ek are regression coefficients reported in Lautenschlager et al.
(1989, p. 343). In combining the various criteria, the following is taken into
consideration: a simulation study by Zwick & Velicer (1986) for PCA shows the
70
As the sample size is large (n = 1125), the chances of “over-fitting” the data can be
neglected (Hair et al. 1998).
71
Kaiser (1970) indicates that a KMO of .80 or higher is meritorious, .70 or higher is
middling, .60 or higher is mediocre, .50 or higher is miserable, and below .50 is
unacceptable. Yet, for exploratory research a KMO over .70 is satisfactory.
72
Green, Tull & Albaum (1988, p.573) pose PCA to be less vulnerable for
misinterpretation, an event more likely to happen in exploratory research. Moreover, with
more than 15 variables, PCA & Common Factor Analysis have the same results.
73
The factors are interpreted based on loadings greater than 0.5 (Hair et al., 1998).
74
Although the scree plot is a criterion used often in determining the number of factors (i.e.
scree -/- 1), establishing the location of the scree is often a rather subjective assessment of
the plot. Therefore, I choose to rely more on the objective criteria mentioned.
96
eigenvalue rule to be accurate 22% of the time, against the parallel method being
accurate 92% of the time. When the first rule is wrong, it always overestimated the
number of components, while with the parallel method, when wrong, there is only
a slight tendency to overestimate the number of components. Therefore, in
choosing for the number of factors, the parallel method is given more weight than
the eigenvalue rule.
Besides, recognizing that a ‘blind’ EFA can result in factors that lack substantive
meanings and are inappropriate for theory development, a thorough analysis of the
substantive meanings of factors and corresponding items is performed to purify the
uncovered factors. Using RA, after EFA, the justification of the factors extracted is
established, and only those factors with alpha’s higher than .50 are included in the
ultimate measurement model75. For the EFA, the 1995 data is displayed only, as
EFA is used to define the structure in the data set. Only if the 1991 or 1993 results
deviates substantially from the 1995 results, these are mentioned as well. For RA,
the alphas for each year are displayed.
The next five subsections are organized according to the various categories
proposed in the conceptual model. Analogous to Chapter 2, a separate sub-section
discusses each block of variables. Each subsection ends with comparing the
specific concept as found in literature and the empirical item(s) used to
operationalise the concepts. The data set with the items and factors found by
applying the analyzes explained above is the input for the multiple imputation.
4.1.2 Export Performance

First, the concept of export performance is operationalized. The literature reviewed
(see section 2.2.1.5) provides a wide array of variables suitable for operationalizing
export performance. In the INTERSTRATOS survey, two possible measures are
75
As explained, EFA is used to fine-tune the variable list used to operationalise the model,
after which NORM will be used to impute the data. Yet, NORM uses all information in the
variables (i.e. pairwise deletion), while the listwise deletion procedure is default option in
EFA in SPSS. Hence, I conduct EFA both with listwise- and pairwise deletion. The results
display only marginal differences, with the same interpretation (i.e. the variables load high
on the same factor). Therefore, only the listwise EFA results are reported.
97
available, i.e. export sales and export ratio (export sales divided by total sales, also
named export intensity), which are both objective, economic measures. These
measures are clearly favorite in export performance studies (see section 2.2.1.5,
and Katsikeas, Leonidou & Morgan 2000).
No subjective performance measures are included in the survey, which might raise
the question whether this mere inclusion of objective measures leads to a too
narrow measurement of export performance. Yet, Powell (1992, p. 126) finds that
the correlations between subjective and objective measures of export performance
are rather high, namely .58 up to .69. In particular, objective performance is a key
element of managers’ subjective assessment of their firms’ performance. Shoham
(1998) also documents high correlations between objective and satisfaction-based
measures of export performance, although he does admit that using both objective
and subjective measures reflects the thinking world of managers more realistically
(Shoham 1996). Madsen (1998) concludes that managers use (and, thus, act upon)
sales-related measures most, which is sometimes criticized for being somewhat
short-term minded. Hence, although subjective measures do add to the power of the
concept, objective measures capture a large part of the export performance concept.
So, acknowledging the shortcomings, I use the objective measures ‘export ratio’
and ‘export sales’ as separate measures for export performance, which are
practically useful indicators and enables building upon the existing body of
evidence. Table 4-1 through Table 4-3 provide some descriptives on ‘total sales’
(as a sub-part of ‘export ratio’), ‘export sales’, and ‘export ratio’.
98
Table 4-1 Descriptives Total Sales (in MIO).

Year Valid Min Max Mean SE mean Skewness
1991 801 41 79997 3988,55 255.04 4.10
1993 852 16 93330 3594.69 234.99 5.28
1995 888 11 80533 3861.69 244.44 4.54
Valid = valid response, Min = minimum value entered, Max = maximum value entered, SE mean =
standard error mean
Table 4-2 Descriptives Export Sales (in MIO).

1991 371 0 31888 2448.44 233.19 3.34
1993 394 1 31756 2426.92 238.15 3.52
1995 436 2 43101 2431.76 230.34 4.12
standard error mean
Table 4-3 Descriptives Export Ratio.

1991 371 0% 100% 31.37% 1.45 .913
1993 394 0% 100% 32.14% 1.42 .795
1995 436 0% 100% 31.69% 1.33 .839
standard error mean
The descriptives portray a few issues. First, the variables are rather skewed, which
has to be taken into consideration when imputing the variables, and when using
them in the analysis. Second, the number of valid cases is much lower than 1125 in
each of the years, especially when inspecting ‘export sales’. This item non-
response is a well-known characteristic of questions pertaining to objective
performance measurement, as respondents are often reluctant to provide this
information. As Nakos, Brouthers & Brouthers (1998) state: “Obtaining
performance data has been one of the major problems for empirical studies dealing
with small businesses as they tend to be privately owned and are, as a rule, very
reluctant to disclose sensitive financial information to strangers.” An even larger
reduction in valid cases occurs, when looking at firms who provided the
information for three years in succession: 609 out of 1125 cases display no item
non-response on ‘total sales’, compared with only 250 entrepreneurs who provided
the researchers with their ‘export sales’ for all three years.
99
This non-response behavior could also have an impact on the size of average
‘export sales’ (and, thus, ‘export ratio’) if the response behavior is contingent on
firm size. The average ‘total sales’ of an SME lie between 3.5 and 4 million MIO.
The average amount of ‘export sales’ lays somewhat under 2.5 million MIO. This
is a rather large part of the total sales, which might be due to non-response bias.
That is, it is possible that entrepreneurs responding to the questions on total sales,
and particularly on export sales, are among the larger companies. Besides, the
larger the company (both in employees and sales) the more obligations a firm bears
to publicity, and the more the entrepreneur will be used to revealing sales figures.
Therefore, the average number of employees of responding and non-responding
SMEs are compared. Table 4-4 clearly shows that there are significant differences
in size between responding and non-responding SMEs (see also 4.2.2.1 where we
discuss the pattern of missing values).
Table 4-4 Average Size Responding vs. Non-Responding SMEs.
Total sales Export sales
Size in employees 1991 1993 1995 1991 1993 1995
Responding 41.03 36.85 40.26 66.56 58.39 62.65
Non-responding 32.76 35.10 31.77 24.92 24.57 23.08
Significance * NS * *** *** ***
Table 4-4 shows for each year the average size of the SMEs that did provide their
(export) sales, and that of the SMEs that did not provide this information. For
example, in 1991 the average number of employees in the SMEs that responded to
‘total sales’ is 41, whereas in non-responding SMEs only 33 employees work on
average. Not only do larger companies report ‘total sales’ more often (two out of
three years significant on a 90% level); this contrast is even sharper for ‘export
sales’ (all years significant on a 99% level). Yet, as I will show in section 4.2, this
covariance between size and sales is covered in the imputation technique that I
apply before I estimate the model.
For ‘export ratio’, the effects of ‘total sales’ and ‘export sales’ might be averaged
out. The results show positive and significant correlations between total sales and
export ratio76 and between the number of employees and export ratio77, indicating
76
.13, .25, and .22 for the years 1991, 1993, and 1995, respectively.
100
that larger firms do have larger export ratios. Ergo, if the data set contains more
responses of larger firms, the left-hand side of the export ratio distribution might be
underrepresented. Yet, as Chapter 4 explains, the imputation technique partly
solves this non-response bias.
4.1.3 Environment
In the framework, an important step has been made towards incorporating the
influence of the environment on export performance and export strategy. The
INTERSTRATOS questionnaire includes twelve questions regarding the development
of the market inducing or prohibiting companies to go abroad. Besides, the
respondents were asked whether certain external stimuli (or barriers) encouraged or
discouraged them when going abroad.
4.1.3.1 Developments in the Market

Concerning the developments in the market, respondents are asked how they
consider recent changes on four different markets from their firm’s point of view
(labor market, supply market, sales market, and capital/credit market). In each
market, three specific developments are examined (see Table 4-5). The answer
categories follow a 5-point Likert scale, ranging from one (‘Very positive’), to five
(‘Very negative’). Although the questions are already ordered into four sub-
markets, these are not based on theoretical underpinnings. Therefore, to verify this
structure in the panel data set, EFA is performed, followed by RA. Both EFA and
RA result in the predicted four-factor solution with all acceptable alphas (i.e.
higher than .50). The item ‘capital and interest costs’ is dropped from the last
factor, for the low inter-correlations with the other variables in the factor.
Subsequently, the four factors are labeled as the changes on the supply, labor,
sales, and capital market, respectively (see Table 4-5).
77
.16, .25, and .23 for 1991, 1993, and 1995, respectively.
101
Table 4-5 Environment: EFA78 and RA.

Environment Change on the … market
Factor Label Supply Labor Sales Capital
Eigenvalue 2.88 2.11 1.48 1.19
Percent of Variance 24.01% 17.56% 12.31% 9.91%
Variables Loading
Market for raw materials .81
Market for products bought for resale .85
Market for technologies .78
Availability qualified workers .81
Availability of unskilled workers .80
Availability of apprentices .81
Local markets .68
National markets .87
International markets .69
Availability finance .77
Availability & quality of financial advice .69
(Interest and other capital costs) (.69)
Cronbach’s alpha
1991 .79 .74 .6480 .60
1993 .81 .78 .63 .58
1995 .7779 .61 .65 .69
In parentheses, the items that are excluded after RA.
4.1.3.2 External Stimuli

Respondents are also asked to indicate whether certain external stimuli encouraged
or discouraged their present level of activities abroad. In handling these, I follow
the classification of Albaum, Strandskov & Duerr (2002), and Leonidou (1995;
1998). These theoretical classifications (see first two columns Table 4-6) serve as
the starting point for categorizing the INTERSTRATOS items into the respective
factor labels that touch upon the export market attractiveness, the home market or
government support (the three topics found in Chapter 2).
78
As said before, only the 1995 data are displayed; the 1993 and 1991 data are only
mentioned if they deviate substantially from the 1995 results.
79
RA indicates to drop ‘market for technologies’ in 1995, but the increase is marginal, so to
be consistent with the other two years, the variable is retained.
80
RA indicates to drop ‘international market’, but the variable is included, for theoretical
reasons.
102
In this section on the environment, only the external stimuli are operationalized.
The internal stimuli are either firm- or management-related, and will be dealt with
in the respective sections. The external stimuli are either proactive or reactive by
nature. Albaum, Strandskov & Duerr (2002) distinguishes between two external
proactive factors, (1) ‘foreign market opportunities’ and (2) ‘change agents’, while
Leonidou (1995; 1998) uses a more specific operationalization with ‘better
opportunities’, ‘possession of exclusive information’, and ‘orders from trade
fairs/missions’ for the first factor, and ‘external agents/organizations’, and
‘government incentives’ for the second factor. The INTERSTRATOS survey
encompasses five items for these external proactive categories (see the third
column in Table 4-6). ‘Better opportunities’ is operationalized with ‘larger market’,
and ‘government incentives’ with ‘subsidies’, ‘political stability’, and
‘entrepreneurial climate’. The variable ‘raw materials supply’ is not available in
Leonidou’s study, but has the nature of a ‘foreign market opportunity’ (Albaum,
Strandskov & Duerr 2002). To determine whether the variables show enough
cohesion for subsequent analyzes, Cronbach’s alphas are displayed for the external
proactive factor for 1991, 1993, and 1995 (as the external reactive factor consists
of only one item, no alpha is calculated). The alphas for 1993 and 1995 are
reasonable to good, especially for exploratory research. The external proactive
factor is a combination of export market attractiveness (‘larger market’, ‘raw
materials supply’, and ‘political stability’), home market (‘entrepreneurial
climate’), and government (‘subsidies’). The external reactive factor can be related
to the situation on the home market: The competition at home is already present on
the foreign market.
103
Table 4-6 Export Motives Classification.

Albaum, Strandskov & Leonidou (1995; 1998) INTERSTRATOS81
Duerr (2002)
External Proactive Factor82
α 1991 .45
α 1993 .60
α 1995 .67
Foreign market opportunities Better opportunities Larger market
Possession of exclusive information
Orders from trade fairs/ missions
NA Raw materials supply
Change agents External agents/ organizations
Government incentives Subsidies
Political stability
Entrepreneurial climate
External Reactive
single-item construct
Unsolicited orders Unsolicited orders NA
Small home market Intense competition home market NA
Stagnating/ declining home Saturation/ shrinkage home market NA
market
NA Initiation export by competitors Following market leader
NA Unsolicited orders from abroad NA
NA Favorable foreign exchange rates NA
NA = not available.
4.1.3.3 The Concept of ‘Environment’ and Operationalization

Overall, six sets of variables are used to operationalise the concept of
‘Environment’. As stated before, the environment is often not included or only
fragments are incorporated in survey research. Eventually, four factors on market
attractiveness (without discriminating specifically between the export market or the
home market) are found, together with two broad motivational factors, which touch
upon macro-, and micro market factors and national policy (government
incentives). As can be seen easily, these fit the theoretical found aspects of
81
Some variables are dropped from further analysis for their lack of theoretical linkages
with the stimuli literature.
82
Although RA indicates to drop ‘subsidies’ and ‘larger market’ in 1991 and 1993, for
theoretical reasons and consistency they are still included over the three years.
104
‘Environment’ only partially. Figure 4.1 represents the concept of ‘Environment’

as described in Chapter 2 and the empirical operationalization in this dissertation.
Literature Our model

Export Market Attractiveness Attractiveness
Macro-economical situation Labour Market (3 items)
Political situation Supply Market (3 items)
Foreign competition Sales Market (3 items)
(Similarity) market Capital /Credit Market (3 items)
Home Country Attractiveness Stimuli
Domestic competition External proactive (5 items)
Home Country Support External reactive (1 item)
Export assistance
National policy
Figure 4-1 The Concept of ‘Environment’ and Operationalisation.
4.1.4 Firm Characteristics

This sub-section focuses on the concept of ‘firm characteristics’ as determined in
section 2.4. These can be divided into objective characteristics (i.e. firm size,
company age, industry, structure, and culture), competencies (i.e. experience,
managing competencies, marketing competencies, and exporting competencies),
and firm-originated stimuli. In the INTERSTRATOS survey, the following company-
related items can be found: firm size, international experience, competitive
advantages, linguistic skills of the management team83, and internal stimuli.
4.1.4.1 Firm Size

In the INTERSTRATOS variable list, the two common measures for firm size are
included, i.e. ‘total sales’ and ‘number of full time employees’. I choose to use only
the last measure for ‘firm size’, as (1) this is a much more researched measure than
‘total sales’, with more consistent results, (2) there appears to be some dependency
83
Although this last item is not found as such in the literature (most focus on the linguistic
skills of the manager), here the proficiency of the total management team is asked for. As
this cannot be attributed to just one manager, we treat the item as a firm capability.
105
between the total sales and the export sales, and (3) the dependent variable ‘exprt
ratio’ is computed using ‘total sales’. Due to our definition of SMEs, the size of the
employee force is restricted between the boundaries of no employees and 500
employees. Table 4-7 shows the 3-year descriptives of ‘number of employees’.
Table 4-7 Descriptives Firm Size.

Year Valid Min Max Mean SE mean Std. Dev. Skewness
1991 1118 0 456 38.66 1.91 63.96 3.46
1993 1124 0 500 36.43 1.87 62.65 3.73
1995 1118 0 500 38.51 2.02 67.67 3.85
standard error mean, Std. Dev. = standard deviation
The minimum number of employees is zero, which implies that the entrepreneur is
the only person working full time in the company. On average, SMEs employ
nearly 40 full time employees, and total sales of the responding SMEs is almost
4000 MIO. The variables are rather skewed (a large right-hand tail); therefore, a
log-linear transformation might be necessary for imputation and further analysis.
4.1.4.2 International Experience

As concluded in section 2.4, an important competence for an exporting firm (often
labeled ‘export experience’) is the international experience it gained. International
experience is mostly measured as the number of years in export (i.e. duration of
export activities), although some authors do expand this myopic operationalization
(such as the scope of the international activities). Therefore, I replace this export-
focused experience with international experience, which develops from both
outbound (i.e. export) and inbound (i.e. import) international activities. Although
the INTERSTRATOS survey does not provide any items to measure how long a firm
is international (the ‘duration’ dimension of experience), quite a few ‘scope’
variables are available.
With these scope variables, I compute two indices to measure both import- and
export-related experience, which together are a proxy for international experience.
The variables measuring the inbound dimension are ‘Did you buy foreign goods
from a supplier abroad’, ‘Which proportion of your supplies came from abroad’,
106
‘Did you make or sell anything under license from abroad’ and ‘Countries you
drew supplies from’ (see Table 4-8).
Table 4-8 Import-Related Experience.
Import Experience Categories Score on scale
Variables
Buy foreign goods? No 0
Yes 1
Which proportion from abroad?84 Nothing 0
(Percentage recoded into quartiles) 0% - 10% 1
11% - 25% 2
26% – 60% 3
> 60% 4
Anything under license? No 0
Yes 1
Number of countries drew supplies from85. No countries 0
(Number recoded into quartiles) 1 country 1
2 countries 2
3 – 4 countries 3
> 4 countries 4
To attribute a score to the answering categories, quartiles are used. The minimum
possible score on inbound experience is zero (no import of foreign goods, nothing
under license from abroad, and, logically, no countries where the SME drew
supplies from), while the maximum score is 10 (the SME bought foreign goods in a
relatively large quantity from several countries, and had (a) product(s) under
license from abroad).
An analogous index is computed for the export-related experience, comprising of

‘Number of countries you sold supplies to’, and ‘Number of channels used abroad’
(see Table 4-9). The more countries (channels) served (used), the more
international contacts it will have, improving international experience.
84
The (25 50 75) quartiles are 1991: (10 30 60), 1993: (10 25 60), and 1995: (10 20 60).
85
The (25 50 75) quartiles are 1991: (1 2 4), 1993: (1 2 4), and 1995: (1 2 4).
107
Table 4-9 Export-Related Experience.

Export Experience Categories Score on scale
Variable
Number of countries sold to.86 No countries 0
(Number recoded into quartiles) 1 country 1
2 – 3 countries 2
4 countries 3
> 4 countries 4
Number of channels?87 No channels 0
(Number recoded into quartiles) 1 channel 1
2 channels 2
> 2 channels 3
The number of channels is a recoded summation of the entry modes88 the SME
indicates to use in the survey:
Did you sell through an agent in your country?; As supplier to an exporting
producer?; To a local firm which eventually exports?; Through agents
abroad?; Directly to customers?;
Did you license any of your products abroad?;
Did you have any subsidiaries for distribution?;
Did you have any subsidiaries for manufacture?
Therefore, the minimum score on export experience is zero (exporting to no
countries at all, and using none of the mentioned channels at all), while the
maximum score is seven (exporting to more than four countries, while using more
than two channels).
4.1.4.3 Linguistic Skills

As mentioned before, the linguistic skills of the management team (instead of just
the owner-manager) is a proxy for the export-related competencies of the SME (see
Table 4-10 for the descriptives). On average, the management team masters 2.5
86
The (25 50 75) quartiles are respectively 1991: (1 3 5), 1993: (1 3 4), and 1995: (1 3 4).
87
The (25 50 75) quartiles are respectively 1991: (0 1 2), 1993: (0 1 2), 1995: (0 1 3).
88
The entry modes ‘licensing’ and ‘having a manufacturing subsidiary’ are broader than
just exporting modes. Although these fall outside the export focus of this dissertation, these
modes do increase a firm’s international experience. Therefore, these entry modes should
be included in an overall measure of outbound international experience.
108
foreign languages, with a minimum of zero and a maximum of 13. Looking at the
skewness of the variable, again a transformation might be necessary before using it
for further analysis.
Table 4-10 Descriptives Number of Languages Spoken by Management Team.
1991 1014 0 10 2.56 .04 1.21 1.05
1993 995 0 13 2.65 .04 1.19 1.10
1995 1112 1 9 2.49 .04 1.19 .80
4.1.4.4 Competencies
Next to the competence firm’s export experience, the empirical variable list
contains a set of questions in which the respondents are asked how important 19
capabilities are for being successful in their specific line of industry, using a 5-
point Likert scale ranging from 1 (no importance) up to 5 (very high importance).
To facilitate interpretation and further analysis, a data reduction exercise was
undertaken using the procedure described in 4.1.1.
An initial EFA, based on eigenvalues over one, results in five factors explaining
59.9% of the total variance. Including a sixth factor increases percentage variance
explained to 64.7%, but the eigenvalue of the extra factor is rather low (.86).
Horn’s parallel analysis (with p = 18 items, and N = 1125 cases) shows criterion
eigenvalues of 1.24, 1.20, 1.18, 1.16, 1.14, and 1.13, for six factors respectively.
This indicates a four-factor solution, which is also easy to interpret. Therefore, a
four-factor solution is selected, explaining 54.3% of all variance.
109
Table 4-11 Competencies: EFA and RA.

89 Factors
Competitive Advantages
Factor Label MC PS CP PC
Eigenvalue 5.48 1.57 1.45 1.28
Percent of Variance 30.46% 8.73% 8.03% 7.08%
Variables
Creativity .54
Distribution & selling staff .75
Prod. design & brand image .81
Financial strength .48
Modify products .42
Workers’ skill .62
Product quality .64
After sales service .53
Flexibility .68
Reliability of delivery .72
Administration .46
Pricing .70
Low cost level .75
Market share .55
(Customer relations) (.54)
(Quality of management) (.55)
Technology .81
Solve technical problems .85
KMO = .87
Bartlett’s chi = 5379.94 (p = 0.00 and df = 153)
Cronbach’s alpha
1991 .74 .64 .6390 .7691
1993 .72 .70 .61 .75
1995 .75 .76 .64 .78
In parentheses, the items that are deleted after RA. MC = Marketing capability, PS = Product
superiority, CP = Competitive Pricing, and PC = Production Capability.
Table 4-11 shows these results, including the factors’ 3-year alphas, which are all
acceptable. These four factors also bear close resemblance to the four competitive
advantage factors that Katsikeas (1994) finds, i.e. ‘Production Capability’ (PC),
‘Marketing Capability’ (MC), ‘Product Superiority’ (PS), and ‘Competitive
Pricing’ (CP). Backed up by other empirical research, I label the factors as such.
89
The variable ‘reputation, local image, personal contacts’ is deleted (communality < .4).
90
Without ‘customer relations’.
91
Without ‘quality of management’.
110
4.1.4.5 Internal Firm Stimuli

As mentioned before, stimuli to enter or pursue exporting stem from outside and
inside organizations. Part of the internal stimuli asked for in the INTERSTRATOS
survey stem from firm characteristics (versus managerial characteristics). As Table
4-14 shows, a proactive, and a reactive factor are operationalized, both with
acceptable alphas. The proactive factor can be seen as a capability or competence,
while the reactive factor is somewhat of a converse of a capability.
Duerr (2002)
Internal Proactive
α 1991 .65
α 1993 .69
α 1995 .71
Marketing Advantages Technical competitive advantage Workers’ skills
Technology
Financial competitive advantage Price
Marketing competitive advantage Costs
Customer relations
Supplier relations
Export know how
Product range93
Economies of scale Economies of scale NA
Unique product/technology Unique products NA
Internal Reactive
α 1991 .56
α 1993 .61
α 1995 .55
Risk diversification Reduce dependence on risk Spreading of risks
Extend sales seasonal product Offset sales seasonal product NA
Excess capacity of resources Unsold inventory/ overproduction NA
Unutilized production capacity Full use of capacity
NA Stagnation/decline dom. sales/profits NA
NA = not available
92
93
Although ‘product range’ has some links with the product offer uniqueness. It
encompasses much more, such as customer choice. Thus, I categorize it as an overall
marketing capability.
111
4.1.4.6 The Concept of ‘Firm Characteristics’ and Operationalization

All variables used to operationalise the concept of ‘Firm Characteristics’ are
summarized in figure 4.2. The left-hand side of this figure displays the variables as
found in the theory (see also chapter 2), while the right-hand side shows the
operationalization of these variables as explained above.
Literature
Our model
Firm Size
Firm Size
Company age
Competencies
Structure
International Experience
Culture
Linguistic skills
Technology level / Industry
Marketing capability
Competencies
Product Superiority
Export Experience
Competitive Pricing
General capabilities
Production capability
Marketing capabilities
Stimuli
Product competencies
Internal Proactive – firm
Internal Reactive – firm
Figure 4-2 The Concept of ‘Firm Characteristics’ and Operationalisation.
4.1.5 Managerial Characteristics

The managerial characteristics are the second internal contextual determinant of
export performance, next to the firm characteristics. Together these determine the
internal context in which (export) behavior is acted out. In the following three sub-
sections, the INTERSTRATOS items on objective and subjective managerial
characteristics, and on internal managerial stimuli are examined.
4.1.5.1 Objective Managerial Characteristics

Regarding the objective characteristics, two items are available, namely the age of
the manager, and the education of the manager (measured as the age of leaving full
time education). Table 4-13, and Table 4-14 show the descriptives on these two
variables. On average, the managers are between 40 and 50 years old, and left full
time education at the age of 22. The INTERSTRATOS variable list does not contain
112
any information on the international experience or language fluency of the

manager.
Table 4-13 Descriptives Manager Age.
1991 1037 21 80 45.73 .30 9.68 -.04
1993 1093 19 82 46.89 .29 9.72 -.11
1995 1088 17 82 47.62 .30 9.90 -.12
Table 4-14 Descriptives ‘Age of Leaving Full Time Education’.

Year Valid Min Max94 Mean SE mean Std. Dev. Skewness
1991 1021 15 50 22.45 .12 3.92 0.98
1993 893 15 45 22.02 .13 3.98 0.95
1995 979 15 60 22.41 .15 4.57 2.13
4.1.5.2 Subjective Managerial Characteristics

With respect to the subjective characteristics, the survey postulates nine attitudinal
statements, touching upon different issues in (international) business. The
respondents are asked to indicate whether they agree with these statements on a 5-
point Likert scale, ranging from 1 (‘I strongly disagree’) to 5 (‘I strongly agree’).
Compared with the theoretical constructs found in subsection 2.5.2, the statements
used bear no specific references towards exporting as a risky business, pertaining
more to the general personality of the manager.
Using EFA, I obtain three factors that explain 57.0% of the total variance.
Including an extra factor improves the variance explained with 10.7%, but the
accompanying eigenvalue is only .86. Besides, Horn’s parallel analysis (with N =
1125 and p = 8 items) results in criterion eigenvalues of 1.08, 1.04, 1.01 (1.005),
and .98, indicating a three-factor solution. After RA, the factors are labeled ‘Risk
94
These maximums seem somewhat strange, as we are talking of ‘full time education’.
Respondents could have answered this question from the starting-point that ‘you never
leave full time education’, and that ‘you never stop learning’.
113
attitude towards change’ (RTC), ‘Risk attitude towards planning’ (RTP), and
‘Government Restrictions’ (GR) (see Table 4-15)95.
Table 4-15 Subjective Managerial Characteristics: EFA and RA.

Managerial Characteristics: Attitude Factors
Factor Label RTC RTP GR
Eigenvalue 2.26 1.29 1.01
Percent of Variance 28.30% 16.07% 12.58%
Statements
Changes should be avoided .78
Not leave region of establishment .71
(Not hesitate to do business with large firms) (-.46)96
Jobs should be clearly defined .75
Managers should plan instead of follow intuition .68
Only introduce proven office procedures .61
Government should not restrict free market .74
Not only provide assistance to members .68
KMO = .71
Bartlett’s chi = 892.66 (p = 0.00 and df = 28)
Cronbach’s alpha
1991 .63 .5597 .23
1993 .57 .60 .34
1995 .66 .55 .30
In parentheses, the items that are deleted after RA. RTC = Risk Attitude towards Changes, RTP =
Risk Attitude towards Planning, GR = Government Restrictions
Only the first two factors RTC and RTP are included, as both have acceptable
alphas in all three years. The third factor GR is be dropped for its very low alphas.
When looking at theoretical justification of these factors; the factors included relate
to the propensity of the respondent to take specific risks98 (see also Kuratko &
Hodgetts 2001).
95
The statement ‘Manager should take personal responsibility for recruiting all employees’
has been dropped, as the communality is under .4 (i.e. .36).
96
Although this variable also loads on GR (.52), both theory and RA indicated that the
variable should not be used in labelling and measuring that factor.
97
Although RA indicates the omission of ‘jobs should be clearly defined’, the increase in
alpha is only marginal, and is contra theory. Therefore, the variable is still included.
98
Risk taking behavior is defined as “every conscious or unconscious controlled behavior
with a perceived uncertainty concerning the outcome and/ or concerning the possible
114
4.1.5.3 Internal Managerial Stimuli

As mentioned before, internal stimuli for exporting stem both from firm and the
manager. Table 4-16 shows the one managerial stimulus factor in the
INTERSTRATOS survey, namely an internal proactive factor, translated into the wish
for ‘prestige’ and ‘growth’ by management. This can be related to the attitude of
the manager towards ‘exporting’.
Duerr (2002)
Internal Proactive
α 1991 .58
α 1993 .55
α 1995 .63
Managerial Urge Special management interest Prestige
Extra growth Growth
Extra profits NA
Extra sales NA
NA = not available
4.1.5.4 The Concept ‘Managerial Characteristics and Operationalization

All variables used to operationalise the concept of ‘Managerial Characteristics’ are
summarized in figure 4.3. The left-hand side of this figure displays the variables as
found in the literature (see also Chapter 2), while the right-hand side shows the
operationalization of these variables as explained above.
benefits or costs for the physical, economical, or psychosocial well-being of somebody or

others” (Trimpop 1994).
99
115
Literature
Objective Characteristics
Age Our model
Education Objective Characteristics
Export experience Age
Age leaving ft education
Language fluency
Subjective Characteristics
Subjective Characteristics
General personality Risk attitude towards changes
Attitude towards exporting in general Risk attitude towards planning
Attitude towards specific aspects of Stimuli
exporting Internal Proactive - Manager
Priority of exporting
Figure 4-3 The Concept of ‘Managerial Characteristics’ and Operationalisation.
4.1.6 Export Activities

This subsection deals with the actual behavior firms displays in exporting, i.e. all
activities SMEs perform to improve the export-related performance, termed
‘Export Activities’. Combining the literature review in section 2.6, and the
INTERSTRATOS survey, the empirical items measuring the firm’s behavioral
orientation on exporting, the behavioral commitment of resources (information
acquisition, and general resource commitment), and a new relational item ‘co-
operation abroad’ are covered in the following subsections.
4.1.6.1 Behavioral Orientation on Exporting

The behavioral orientation of exporting SMEs can be either proactive or reactive,
where an active behavioral orientation is hypothesized to have a positive effect on
export performance (see subsection 2.6.1). In the present survey, the question ‘Did
you search actively for orders abroad’ appears, but is a very crude index, being
binominal (1 = ‘yes’, 0 = ‘no’). Table 4-17 shows the descriptives of this variable.
116
Table 4-17 Descriptives 'Did You Search Actively for Orders from Abroad'.
Year Valid Mean SE mean Std. Dev. Skewness
1991 1125 .27 .01 .44 -1.04
1993 1125 .28 .01 .45 -.96
1995 1125 .28 .01 .45 -.96
4.1.6.2 Expansion Strategy

As mentioned in Chapter 2, ‘the number of countries a firm exports to’ has been
used as a proxy for export performance, firm size, international experience, and the
expansion strategy. In this, we follow the criticism brought forward by Katsikeas,
Leonidou & Morgan (2000), when they refer to the “persisting debate on export
market expansion, suggesting that the number of foreign markets is not an end in
itself but is contingent on the specific company, product, market, and marketing
factors.” By treating the number of export countries as a proxy for market
expansion, and not for export performance or firm size, this does not mean that its
merit for international experience deteriorates. Therefore, the number of export
markets is both used in the index for international experience, as in the estimation
of the exporting firm’s expansion strategy.
Table 4-18 Descriptives ‘Number of Export Markets’.
1991 547 1 12 3.11 .10 1.96 .84
1993 535 1 13 3.16 .10 1.95 .84
1995 551 1 10 3.24 .10 2.08 .88
On average, the SMEs export to three different countries. This figure remains
rather stable over the three years of study. As the statistic for skewness is rather
high, a transformation might be needed when imputing the data in NORM and
analyzing the models.
4.1.6.3 Behavioral Commitment of Resources: Information Acquisition

A second category of export activities pertains to the behavioral commitment of
resources, either by planning, by looking for information, or by restructuring the
117
organization for exporting. Literature tells us that higher commitment of resources

improves export performance. A proxy for the commitment of resources, or the
extent of involvement, is the amount of information collected on the international
operations100. The INTERSTRATOS survey contains several items related to the
information acquisition of the exporting SME. Specifically, respondents are asked
whether they used twelve specific information sources either at home or abroad
concerning international activities. As mentioned in Chapter 2 ,the information
sources can be divided into formal and informal information sources (e.g., Hart,
Webb & Jones 1994). The twelve information sources asked for in the present
survey are also categorized according to this classifications (see Table 4-19)101.
Table 4-19 Classification Information Sources.
Hart, Webb & Jones (1994) INTERSTRATOS
Formal information Training institutions
Business consultants
Credit agencies
Research institutions
Chambers of Commerce
International organizations
Public promotion fairs
Informal information Suppliers
Customers
Export clubs
National trade fairs
International trade fairs
100
Another proxy available is the number of employees that is working abroad.
Unfortunately, the variable has almost no valid cases, and the variance is deflated due to the
fact that here appears to be an inconsistency in the data: Some countries entered a ‘0’ when
a respondent did not enter a value. So, it is not clear whether all 0’s (respectively 89.4%,
94.0%, and 93.4%) define ‘no employees abroad’ or a missing value. Therefore, this
variable is not considered suitable for further analysis.
101
See for a more extensive explanation of these categorizations, chapter 6, where I also
use the more specific classification of export market research, assistance, and intelligence.
For efficiency reasons, the higher aggregation level of formal and informal sources is used
in the integral model. After all, every new variable increases the number of relationships
exponentially.
118
In the present study, a respondent can consult a maximum of 24 information

sources: twelve at home and twelve on the foreign market. Of these, 14 are
considered formal information sources and ten are considered informal. New index
variables are being computed, counting the number of formal and informal
information sources consulted by the respondent either at home or abroad.
4.1.6.4 Restructuring the Organization

As explained in subsection 2.6.3, the behavioral commitment of resources is also
present in the way the organization is restructured for exporting. Mostly
respondents are questioned whether they have their own export department or not.
Another way of looking at this is by considering the entry mode chosen as a
combination of risk, control, and capital aspects, with entry modes higher on
financial investments implying a higher commitment of resources. Basically, the
fundamental choice for SMEs is whether to expand through an integrated channel
(company owned) or through a channel with independent intermediaries (Anderson
& Coughlan 1987). This concept of involvement can be translated into a continuum
ranging from export entry modes assuming virtually no control and risks, accepting
lower returns, to modes with highest risk and control, and, thus, higher expected
returns. On the one end one could imagine the firm selling its products to another
(domestic or foreign) firm that gains possession and subsequently will export or
sell the goods. On the other end, a firm could be placed that starts a wholly owned
distributing subsidiary in a foreign country. Pursuing this thought, the entry modes
in the INTERSTRATOS survey are categorized and scores are attributed (Table 4-20).
This new scale is made according to the level of capital, control and risk that an
entry mode incorporates102.
102
This scale was computed on the base of literature (e.g., Albaum, Stradskov & Duerr
2002) and a discussion with Prof. S. Tamer Cavusgil (Michigan State University).
119
Table 4-20 Entry Modes: Commitment Scores.

Entry Mode103. 104
Score
Sold anything through:
an agent in your country? 1
as supplier to an exporting producer? 1
to a local firm which eventually exports? 1
through agents abroad? 3
directly to customers? 4
Any subsidiaries for distribution? 6
The order in which the various modes are organized in the survey will be the same
on this scale, i.e. selling to a local firm who eventually exports encompasses a
lower degree of control and risk than selling through agents abroad. Besides, as
some of the entry modes are very close (e.g., selling through an agent in your
country and selling as a supplier to an exporting country), and others are more than
just one unit apart from each other (e.g., selling directly to customers or owning a
subsidiary abroad for distribution), the seven categories will be scaled curvilinear.
4.1.6.5 Cooperation Abroad

Lastly, although not mentioned in the export performance literature as such, but
mentioned often in networking theory (Johanson & Mattsson 1988, Chetty &
Blankenburg Holm 2000), building relationships with other partners and working
together can be beneficial. In the interaction approach, a firm’s position is
determined not only by its’ resources, but “position is the product of a firm’s
interactions, adaptations, and relationships with other entities in the network”
(Evangelista 1996). Although some studies press the importance of relationships
with channel members (e.g., Leonidou & Kaleka 1998) or customers (Evangelista
103
Respondents were also asked whether they licensed any of their products abroad, and
whether they had a subsidiary for manufacturing. These are excluded, as these are both
non-export entry modes, and, thus, alternatives for exporting. Besides, license contracts are
very heterogeneous, with a varying degree of control possible in different license contracts,
which makes it impossible to assign a single score to this entry mode (thanks to the remarks
of Prof. S. Tamer Cavusgil).
104
Some respondents entered multiple answers. In this case, only the highest-ranking entry
mode is used to enter a score on the commitment scale for that particular respondent.
120
1996), no export performance study reviewed examined the effect of relationships

on a more horizontal level (i.e. with businesses in the same industry) to enhance the
business position. By co-operating, an SME can build up a network of “value laden
relationships” (cf. Styles & Amber 1994), improve the business process, and
increase export performance. Most authors stress the social value of networks; by
forming relationships firms gain experiential knowledge (e.g., Styles & Amber
1994; Evangelista 1996). In addition to this social value, co-operation relationships
can have more formal gains, in obtaining access to resources of relations not owned
by the SME itself. For instance, Ling-Yee & Ogunmokun (2001) stress the
strategic impact of resources and capabilities that can be shared in marketing’s co-
operative relationships. “In short, due to the very nature of the export channel
environment, exporting firms cannot rely on their internal competencies for
crafting competitive advantages and achieving export sales, they must tap into the
distinctive competencies of importers if they want to reali[s]e the full potential of
their product offerings.” Therefore, co-operating with other actors in the
international field is hypothesized to impact export performance positively.
Returning to the INTERSTRATOS survey, respondents are asked whether they co-
operated on specific activities abroad with domestic or foreign partners (i.e. the
extension of the product range, R&D, raising funds, sales, market research, after-
sales service, advertising or promotion, purchase or supply, transport or
warehousing, manufacturing, administration or electronic data processing) 105. By
counting the number of times SMEs indicate to have co-operated with domestic or
foreign partners, I compute two new indices: (1) extent of co-operation with
domestic partners abroad, and (2) extent of co-operation with foreign partners
abroad. These indices range from zero (no co-operation at all) to twelve (co-
operation sought for all twelve activities listed).
105
Respondents were also asked whether they co-operated with domestic or foreign
partners at home on specific activities. Yet, the survey is not explicit in this question.
Implicitly, the INTERSTRATOS survey relates to activities carried out to improve the
company’s international operations, but it is not clear whether the respondents interpreted
the present question as such. Chances are that the respondent ticked any activity on which
he co-operated, no matter whether to improve national or international performance. To
avoid the risk of misinterpretation, these questions are excluded from further analyses.
121
4.1.6.6 The Concept of ‘Export Activities’ and Operationalization

All variables available to operationalise the concept of export activities are
summarized in figure 4.4, adding a new strategic activity, i.e. the extent of co-
operation abroad.
Literature
Firm s’ behavioural orientation on exporting
Competitive and expansion strategy
Resource commitment
Planning O ur model
Search actively for orders abroad
Information behaviour
Number of export countries
Structure
Acquisition of information
M arketing M ix
Channel design
International channel managem ent
Co-operation abroad
International pricing policy
International product policy
International promotion policy
Figure 4-4 The Concept of ‘Export Activities’ and Operationalisation.
4.1.7 Variables in Operationalized Model

The data set with the variables used to operationalize the constructs in the
theoretical framework is represented in Table 4-21. Three different types of
variables can be distinguished, i.e. single-item variables, constructs, and indices.
That is, ‘export sales’ is measured with only one variable, and is labeled a single-
item concept. On the other hand, ‘export experience’ is a summation of several
variables: an index. Lastly, a construct is a scale build up from several reflective
variables. These are the input for the imputation phase, described in the subsequent
section.
122
Table 4-21 Constructs, Indices and Single-Items for Operationalization.

Concept Type
Export Performance
Export sales Single-item
Export ratio Single-item
Environment
Labor market attractiveness Construct (3 items)
Supply market attractiveness Construct (3 items)
Sales market attractiveness Construct (3 items)
Capital/ credit market attractiveness Construct (2 items)
External proactive stimulus Construct (5 items)
External reactive stimulus Single-item
Firm
Firm size Single-item
International experience Index
Linguistic skills Single-item
Marketing capability Construct (5 items)
Competitive pricing Construct (6 items)
Reliability Construct (3 items)
Production capability Construct (2 items)
Internal proactive stimulus (firm) Construct (8 items)
Internal reactive stimulus (firm) Construct (2 items)
Manager
Age Single-item
Age leaving full time education Single-item
Risk attitude towards changes Construct (2 items)
Risk attitude towards planning Construct (3 items)
Internal proactive stimulus (manager) Construct (2 items)
Export Activities
Actively searching for orders from abroad Single-item
Number of export markets Single-item
Acquisition of information Index
Commitment resources through entry mode Single-item
Co-operation abroad Index
4.2 Multiple Imputation: NORM

As mentioned in the introduction section , the second aim of this chapter is to make
the operationalization items full rank by imputing the missing values. The
exploratory analysis in the previous section came up with the items to be imputed
(see Table 4-21 for the constructs, indices and items used). The following sub-
sections introduce this concept of multiple imputation as a remedy for missing
123
values, the method used in this imputation (i.e. the Norm program by Schafer
1997), and the actual imputation of the data set. With the resulting data sets the
operationalized model is estimated in the next chapter.
4.2.1 Multiple Imputation - Theory106

Item non-response in surveys can lead to serious problems, as missing values result
in less efficient estimates and the impossibility of using complete-data methods. A
solution is using only full information rows in the data set (i.e. respondents in the
survey that have no item non-response, discarding cases with missing values), but
this reduces the available cases drastically. For example, many SMEs refused to
provide the researcher with their total, or export sales, as already pointed out in
sub-section 4.1.2. Another solution is the imputation of missing values, either
single or multiply. In multiple imputation, several probable values are imputed for
every missing value, resulting in various imputed data sets. On the other hand,
single imputation results in one imputed data set as only one value is imputed for
each missing. This sub-section goes deeper into the different methods to impute
missing values.
A popular single imputation method is to replace the missing values with the mean
of observed values (Rubin, 1987). Although a simple procedure, the disadvantages
are considerable, due to the underestimation of the variability in the data set and
the possibility of distorted relationships due to a possible non-response bias.
Another procedure is the Hotdeck procedure in which a closely matching
respondent is found for each non-respondent using variables observed for both.
Unfortunately, this method also underestimates the variability in the data set,
because it treats imputed values as if they were known with certainty. A last
possible single-imputation method is to use the mean of followed-up non-
respondents (also called the double sampling estimation technique). Still, the
disadvantages of the first two procedures are not completely solved. Moreover,
non-respondents are not re-approached in the INTERSTRATOS project.
In contrast to single imputation, multiple imputation allows the analyst to use

knowledge to reflect the uncertainty about which values to impute. Multiple
106
This section is largely based on Rubin (1987).
124
imputation (MI) is a statistical technique designed to take advantage of the

flexibility in modern computing to handle missing data. With it, each missing value
is replaced by two or more imputed values in order to represent the uncertainty
surrounding the value to impute. Next, when multiple imputations represent
repeated random draws under a model for non-response, valid inferences can be
obtained by combining complete-data inferences in a straightforward fashion.
Besides, if the model for non-response is not known, MI allows studying the
sensitivity of inferences to various models for non-response by using the complete-
data methods repeatedly. On the other hand, the drawbacks of MI are the
imputation and analyzing time needed and the storage size for multiple data sets.
An important tool in MI is Bayesian theory. Here, the posterior distribution of Ymis

is simulated under an explicit Bayesian model. In this, three formal phases can be
defined: the modeling task, the imputation task, and the estimation task. In the
modeling task, a specific model for the data is chosen, resulting in the posterior
distribution for Ymis, Pr (Ymis| X, Yobs). Next, in the imputation phase, the goal is to
draw Ymis from its posterior distribution under the model. Finally, in the estimation
task, the posterior distribution of θY|X is computed, so that a random draw of θY|X
can be made. In more general wording, a data set consists of two types of variables:
X and Y. The X variables are totally observed, and contain no missings. In
contrast, the Y variables are not completely observed and contain missings,
splitting each variable up in observed Y values for other cases: Yobs, and missing
Y-values for some cases: Ymis. The program uses the information in the completely
observed X-variable and the non-missing Y values (Yobs), to impute Ymis m times
(Ymis| X, Yobs).
4.2.2 The Method - NORM107

Based on the theory above, Schafer (1997) developed several statistical packages to
deal with missing values. NORM is one of those packages108, performing multiple
imputation using the multivariate normal distribution. The program gives
researchers the opportunity to impute missing values multiply, using Bayesian
approaches. Using information obtained from the observed part of the data set,
107
This section is largely based on Schafer (1997).
108
NORM is the only package that is produced as a stand-alone Windows package well.
125
NORM simulates the missing part m times, with m > one, creating m equally
plausible versions of the complete data. These m sets can be analyzed with
complete-data techniques, combining the m sets of results to produce one set of
estimates and standard errors109.
NORM exists of two steps to impute Ymis based on X, and Yobs. The first is an EM
algorithm (EM-phase) for efficient estimation of mean, variances, and covariances
(or correlations) using all of the cases in the data set, including those that are
partially missing. The EM algorithm is a general method for obtaining maximum-
likelihood estimates of parameters from incomplete data. The second is the data
augmentation procedure (DA-phase) for generating multiple imputations. The DA
algorithm simulates random values of parameters and missing data from their
posterior distribution in two steps, the I-step (i.e. imputing the missing data by
drawing them from their conditional distribution given the observed data and
assumed values for the parameters), and the P-step (simulate new values for the
parameters by drawing them from a Bayesian posterior distribution given the
observed data and the most recently imputed values for the missing data). EM will
provide a set of starting values for the parameters. By running DA for a large
number of cycles, and storing the results of a few I-steps along the way, one
obtains proper multiple imputations of the missing data. Before entering the EM-
phase, some pre-processing of the data might be necessary to handle non-normal
variables. Several transformations can be applied, if variables are not normally
distributed, that is, power transformations (for correcting skewness), and logit
transformations (useful for a variable that takes values only within a limited range),
and dummy coding (for including categorical variables that have no missing values
in the model110).
109
NORM is based on three statistical assumptions, i.e. the missing values are assumed to
be missing at random (MAR), the variables in the data set are jointly normally distributed,
and the data augmentation procedure is a Bayesian procedure and, thus, relies on a prior
distribution for the unknown parameters (means, variances, and covariances).
110
If the categorical variable y does have missing values or if the values of y represent
ordered categories, you can include y in the model without recoding; rounding the
continuous imputed values to the nearest observed values of y.
126
4.2.2.1 Assessing Non-Ignorable Missingness111

In general, two models for non-response can be distinguished; the ignorable and
the non-ignorable model. The first model assumes that the unobserved distribution
of Y for non-respondents is only randomly different from the observed distribution
of Y for respondents. The second model assumes systematic different values for
non-respondents compared to respondents with the same values of observed
variables. Without some external knowledge, the assumption of ignorable non-
response is rather plausible since no extra information is available on the
population distribution of Y. In this case, imputation methods adjust for all
observed differences between respondents and non-respondents and assume that
unobserved differences are random. Besides, including as many observed
characteristics of each individual as possible when defining the data set enhances
the plausibility. Increasing the pool of observed variables decreases the degree to
which missingness depends on unobservables given the observed variables
(Gelman et al. 1995).
So, before continuing, we have to establish that the pattern of missingness is

ignorable, as NORM assumes a MAR-pattern (i.e. the probability that an
observation is missing may depend on Yobs but not on Ymis (Schafer 1997)). If this
assumption is violated the imputed values are biased and useless. Yet, Schafer
(1997) states that the MAR hypothesis cannot be tested from data at hand; doing so
would require knowledge of the missing values themselves. “When such an
external source is unavailable, deciding whether or not MAR is plausible will
necessarily involve some guesswork, and will require careful consideration of
conditions specific to the problem at hand.” Keeping this in mind, I try to establish
some insights into the pattern of missingness in the INTERSTRATOS data set. This
investigation helps, in the end, to draw conclusions with more certainty. In
particular, the pattern of missingness in the dependent variables ‘total sales’ and
‘export sales’ is examined, using the background variables ‘country’, ‘branch’, and
111
I would like to thank Professor Ton Steerneman of the University of Groningen (Faculty
of Economics, Department of Econometrics) for his useful remarks on this section.
127
‘size’112. In addition, the pattern over the years is examined, as longitudinality is a

specific condition of the INTERSTRATOS data set.
Table 4-22 and Table 4-23 present the assessment of MAR by country and sector,
respectively, by showing both the percentage of non-response on total sales and
export sales within the country and sector. In a similar manner, in parentheses the
percentage of each country and sector within the non-respondents is shown. Next,
the column ‘Sign.’ shows the significance level of the chi-square test statistic
conducted on the percentages response and non-response on total sales and export
sales in the countries/branches. Lastly, the second row shows the percentage of
firms coming from the respective country or sector in the data set.
Table 4-22 Assessing MAR: Non-Response by Country.
Austria Belg. Netherl. Switz. Norway Sweden Finland N
Data set 38.0% 3.2% 3.6% 11.5% 8.4% 29.9% 5.6% 1125
Total sales Non-response Sign.
1991 34.2% 19.4% 25.0% 46.5% 10.6% 25.3% 9.5% ***
(45.1%) (2.2%) (3.1%) (18.5%) (3.1%) (26.2%) (1.9%)
1993 27.9% 22.2% 12.5% 27.1% 20.2% 23.5% 12.7% *
(43.6%) (2.9%) (1.8%) (12.8%) (7.0%) (28.9%) (2.9%)
1995 27.9% 16.7% 10.0% 20.2% 17.0% 16.7% 15.9% ***
(50.2%) (2.5%) (1.7%) (11.0%) (6.8%) (23.6%) (4.2%)
Export sales Non-response Sign.
1991 88.1% 30.6% 62.5% 67.4% 52.1% 50.6% 57.1% ***
(49.9%) (1.5%) (3.3%) (11.5%) (6.5%) (22.5%) (4.8%)
1993 84.3% 38.9% 50.0% 60.5% 59.6% 50.0% 55.6% ***
(49.2%) (1.9%) (2.7%) (10.7%) (7.7%) (23.0%) (4.8%)
1995 82.9% 27.8% 47.5% 55.0% 53.2% 46.7% 44.4% ***
(51.4%) (1.5%) (2.8%) (10.3%) (7.3%) (22.8%) (4.1%)
* = significant at 90%, ** at 95%, *** at 99%
Examining at the output, it appears that Austrian SMEs are consistently less
inclined, while Belgium SMEs are consistently more inclined to reveal export
sales. Similar conclusions can be drawn from the output on total sales in the top
112
Although other variables also contain missing values, these are assumed to be random,
and only determined by Yobs en X. For example, there is no reason to assume that the
response on questions concerning whether companies did or did not use certain information
types or entry modes, is non-ignorable.
128
half of the table. Again, Austrian firms are consistently less inclined, while Finnish
firms are more open concerning their total sales.
Table 4-23 Assessing MAR: Non-Response by Sector.
Percentage Textiles Electronics Food Furniture Mechanical
in data set
Data set 15.7% 13.0% 19.6% 25.6% 26.2%
Total sales Non-response Sign.
1991 36.9% 18.6% 36.4% 24.6% 27.2% ***
(20.1%) (8.0%) (24.7%) (21.6%) (25.6%)
1993 28.0% 16.1% 39.2% 18.8% 20.1% ***
(17.9%) (8.4%) (31.9%) (19.8%) (22.0%)
1995 26.7% 19.6% 26.4% 18.8% 16.7% **
(19.8%) (12.2%) (24.5%) (22.8%) (20.7%)
Export sales Non-response Sign.
1991 59.1% 47.1% 85.9% 72.9% 61.6% ***
(13.8%) (8.8%) (25.1%) (27.5%) (24.9%)
1993 52.6% 47.6% 87.8% 71.8% 57.0% ***
(12.6%) (9.3%) (26.7%) (28.2%) (23.3%)
1995 52.3% 48.0% 79.5% 70.7% 50.3% ***
(13.4%) (10.3%) (25.4%) (29.5%) (21.5%)
* = significant at 90%, ** at 95%, *** at 99%
Regarding the sectors, high non-response percentages can be found in the food
branch, while the electronics branch is relatively open, both on total sales and on
export sales113.
Another interesting variable that could impact the responsiveness to questions on

sales is the size of the company. After all, larger firms might be more ‘successful’
in terms of sales, and, thus, more open on these figures. Besides, the larger the
company (both in employees and sales) the more obligations a firm bears to
publicity, and the more the entrepreneur will be used to revealing sales figures.
Table 4-24 represents the average firm size in employees of responding versus non-
responding SMEs in 1991, 1993, and 1995. Clearly, SMEs reporting their export
113
I also executed this exercise for ‘country’ and ‘sector’ simultaneously. Although the
results indicate associations between country’, ‘sector’, and (non-)response behavior, too
many cells contain too few cases, which makes the chi-square results unstable, and, thus,
unreliable. Therefore, I only look at the first-order associations of (non-)response behavior
versus country, and sector.
129
sales are on average twice until three times as large as their non-responding
counterparts.
Table 4-24 Assessment of MAR: Firm Size Respondents & Non-Respondents.
Total sales Export sales
Size in employees 1991 1993 1995 1991 1993 1995
Responding 41.03 36.85 40.26 66.56 58.39 62.65
Non-responding 32.76 35.10 31.77 24.92 24.57 23.08
Significance * NS * *** *** ***
* = significant at 90%, ** at 95%, *** at 99%, NS = non-significant
Examining these tables on the non-response patterns in the original data set with
1125 cases, one can see that the missingness varies over country, over sector, over
size (measured as number of employees), and also over years. This amount of
missing information is larger in some parts of the data set; e.g., Austrian SMEs or
companies in the food business providing hardly any export sales figures. In these
cases, the imputation program has less information to base the imputation upon
than in other cases, which might mean that these values are less stable. An option
would be to loose the cases with little information on important variables. On the
other hand, this means introducing a selection bias in the data set.
Three options prevail:

1) Only full ranked cases are included (listwise deletion): no imputation needed;
2) Only cases with full rank on both export sales and total sales are included (i.e.
no missing values in the dependent variable): other missing values are imputed;
3) All cases are included: imputation for all Y variables.
Several criteria are important when making a choice. Firstly, the breadth of
conclusions to be made. Secondly, the rate of missing information and, thirdly, the
number of cases left. As discussed earlier, NORM allows for the missing
information rate in the data set by adding variability between the three data sets
imputed. Moreover, when including all 1125 cases, we improve the reach of the
conclusions we draw. Another solution could be to only include those 250 cases
that have no missing values on export sales or total sales in either year. This might
answer the need for a data set with a smaller rate of missing information leading to
stronger conclusions (owing to more stable imputations); but the breadth of
conclusions to be drawn can not reach any further than the firms in this part of the
130
data set. For instance, only few Austrian firms are included, few food
manufacturers, whereas larger SMEs are overemphasized. The most rigorous
option is not to impute at all and use list-wise deletion. Although this is the easiest
approach, it is also the most unwanted procedure, as this would reduce the data set
enormously (i.e. a reduction from 1125 cases to far less than 250 cases). Besides,
Schafer (1997) shows the benefit of general ignorable procedures, such as the EM
algorithm used in NORM over ad hoc procedures such as case deletion. “Even
when the missing data are not precisely MAR, however, general ignorable
procedures still tend to be better than ad hoc procedures such as case deletion for
the following reason: general ignorable procedures remove all of the non-response
bias explainable by Yobs, whereas ad hoc procedures may not.” He states, “[T]he
crucial assumption made by ignorable methods is not that the propensity to respond
is completely unrelated to the missing data, but that this relationship can be
explained by data that are observed. Whether this is plausible in a particular setting
is therefore closely related to the richness of the observed data Yobs, and the
complexity of the data model P (Y | θ). … Even if they do not perform well, they
still may provide an important and useful baseline for assessing and comparing any
available alternatives.”
All things considered, using only the 250 cases with full rank on export sales (and,
thus, export ratio) implies that we can only draw conclusions related to larger
companies, or survivors. Therefore, the 1125 SMEs are a much better
representation of European SMEs. Another issue to consider is that an imputation
model is a predictive model. Its sole purpose is to create plausible imputations of
the missing values. The goal is not to predict the missing values with the greatest
accuracy or to describe the data in a scientifically meaningful way, but to properly
reflect uncertainty, to preserve important aspects of the data distributions, and to
preserve important relationships. If that is the goal, the choice for 1125 cases is
justified even more.
4.2.2.2 Variables to be Included in the Imputation Model

An important question to answer before running NORM is which variables to
include in the model NORM uses to impute. Schafer (1997) states, “An imputation
model should be chosen to be (at least approximately) compatible with the analyzes
to be performed on the imputed data sets. In particular, the model should be rich
131
enough to preserve the associations or relationships among variables that will be

the focus of later investigation114. In general, any variable(s) that may prove
important in subsequent analyzes should be present in the imputation model.115”.
For this reason, the variables to be used in the imputation model are the same as the
variables used in the model operationalized in this chapter, supplemented by the
background variables ‘country’, ‘sector’, and ‘size’. After all, MAR is defined
relative to the variables in a data set. For example, if the response pattern on export
sales depends on the country-of-origin, we can still have MAR, as ‘country’ is
included in the imputation as an X-variable (fully observed). In that way, the
covariances between ‘country’ and ‘export sales’ are taken into account when
imputing ‘export sales’. Accordingly, if a variable X is related both to the
missingness of other variables and to the values of those variables, and X is
removed from the data set, then MAR will no longer be satisfied. For this reason, it
is good practice to include variables that are predictive of missingness, increasing
the plausibility of MAR (source: help-function NORM116). Therefore, although one
can see differences in responsiveness due to country-of-origin, sector, and size of
the SME (see 4.2.2.1), we can assume MAR, as the imputation procedure includes
these variables. Besides, simulations show that multiple imputation is quite robust
to model misspecification (Schafer, 1997).
All variables are checked on deviations from normality and are transformed if
needed, to satisfy the normality assumption of the program. After running EM, in
114
He goes on: “For example, suppose that you use NORM to impute a variable Y under a
model that includes the variable X. After imputation, an analyst uses linear regression to
predict Y from X and another variable Z that was not in the imputation model. The
estimated coefficient for Z from this regression would tend to be biased toward zero,
because Y has been imputed without regard for its possible relationship with X.”
115
He ends: “The converse of this rule, however, is not at all necessary. If Y has been
imputed under a model that includes Z, there is no need to include Z in future analyzes
involving Y unless the YZ relationship is of substantive interest. Results pertaining to Y
cannot be biased by the inclusion of extra variables in the imputation phase. Therefore, a
rich imputation model that preserves a large number of associations is desirable because it
may be used for a variety of post-imputation analyzes.”
116
http://www.stat.psu.edu/~jls/misoftwa.html
132
the DA step the number of iterations is set as three times the number of iterations
needed at the EM-step (i.e., three times k), as k is an indication for the number of
iterations that is needed to converge. Every kth iteration the data augmentation step
imputes a data set, until three imputed data sets exist at the end of the process.
At the start of the imputation phase in my PhD project, the computational capacity
of the computers available did not allow for multiply imputing the whole data set at
once (all variables for 1125 cases for three years). Therefore, the data set is first
split up in three smaller data sets, i.e. one per year, and the 1995 part of the panel
data set is imputed. About half a year later, stronger computers became available,
allowing me to impute the 3-year data set in one time117.
Table 4-25 presents the number of iterations NORM needs for convergence (k), the
ridge prior used in this procedure, and the EM observed-data log-likelihood118 for
both the 1995 data set and the 3-year data set.
Table 4-25 Iterations Needed for Convergence (k) and Log-Likelihood 1995.
Year k Ridge prior hyper- Log-likelihood
parameter
1995 600 100 -16775.73
1991-1993-1995 803 100 -9202.03
4.2.2.3 Original and Imputed Original Variables

Table 4-26 and Table 4-27 display the original mean values versus the imputed
mean values of total sales, export sales and export ratio for both the 1995 data set,
and the 3-year panel data set119.
117
Both the EM-step and the three imputation steps took more than 40 days each, to
indicate the computational effort asked from the computer.
118
The EM-step converges if the log-likelihood improves with less than 0,0001 in the
following iteration.
119
After deleting outliers.
133
Table 4-26 Means in the Original and Imputed Values 1995 Data Sets.
Variable Original data set 1st imputed data 2nd imputed data 3rd imputed data
set set set
Total sales 3861.69 3819.16 3925.47 3794.87
Export sales 2431.76 1507.76 1978.99 1551.85
Export ratio 31.69 22.34 24.01 21.63
Table 4-27 Means in the Original and Imputed 3-Year Panel Data Sets.
Variable Original data set 1st imputed data 2nd imputed data 3rd imputed
set set data set
Total sales
1991 3988.55 4102.43 3990.49 5256.78
1993 3594.69 3871.78 3842.67 4612.59
1995 3861.69 3944.09 4156.79 4065.47
Export sales
1991 2448.44 1739.49 1425.97 2279.43
1993 2426.92 1781.72 1979.23 1861.83
1995 2431.76 1544.85 1512.27 1426.90
Export ratio
1991 31.37 23.09 24.25 25.38
1993 32.14 21.54 23.25 22.90
1995 31.69 21.60 22.68 23.31
As the figures show, the more missing information appears in a variable, the more
deviation appears from the original mean value. What is especially remarkable, is
the lower mean values of the performance measures in the imputed data sets. Yet,
considering the findings in 4.2.2.1 (remember, in general larger firms responded on
total and export sales), these findings are not so far-fetched as the purpose of
imputation is to impute item non-response. As these probably belong to smaller
SMEs, a lower value is imputed (consistently over the three years). This puts the
panel data set more in line with reality. Besides, and moreover, we have to keep in
mind that it is not the goal to precisely reflect the mean values of the variables, but
to mirror the covariance matrix as good as possible, allowing for uncertainty due to
the amount of missing information in the respective variable.
The three imputed data sets are the basis for the subsequent chapters in which (1)
the one-year integral export performance model is estimated (1995-data), and (2) a
longitudinal model is estimated focusing on information behavior (3-year data).
The procedure is the following: the first imputed data set is used to find and
134
estimate the models, after which the found model structure is imposed on the other
two imputed data sets to establish the amount of uncertainty or the stable nature of
the models, and parameters found. After all, if the model is replicated exactly in all
three imputed data sets, the solution is rather robust. On the other hand, it is
possible that some parameters deviate in size and direction over the three data sets.
In that case, the conclusions according to the specific relationship will be less
strong. Therefore, in presenting the results of the models in Chapter 5 and Chapter
6, I take all three imputed data sets into consideration.
135
136
Chapter 5
The Integral Export
Performance Model -
Estimated
In the previous chapter, the proposed theoretical model (stemming from Chapter 2)
is operationalized using the variables available in the INTERSTRATOS data set.
Several theoretical constructs have been left out due to the non-comprehensiveness
of the empirical data set; proxies have been worked out for some constructs, with
other constructs readily available in the INTERSTRATOS variable list. In this
chapter, I describe the estimation of this operationalized model. Section 5.1
introduces the operationalized model, as it came forward from the previous
chapter. In section 5.2, the data analysis procedure is introduced. In section 5.3 and
5.4, the indices and constructs in the model are validated, followed by the results of
the structural model (section 5.5).
5.1 The Operationalized Model

Figure 5.1 depicts the proposed export performance model to be tested. Based on
the theoretical analysis and discussion in the previous chapters, I can state the
model in a formal fashion. To render the graphical representation of the model not
too complicated, the figure assumes that all constructs in one category relate in the
same manner to constructs in another category. The categories, as mentioned
previously, are:
• Environment;
• Firm characteristics;
• Managerial objective characteristics;
137
• Managerial subjective characteristics;

• Export activities;
• Export performance.
Firm Characteristics:
- Firm size
- International experience
- Linguistic skills
- Marketing capability
Objective Managerial - Product superiority
Characteristics: - Competitive pricing
- Age - Production capability
- Age leaving full time education - Internal proactive stimulus
- Internal reactive stimulus
Characteristics: Export Performance:
- Risk attitude towards changes - Export sales
- Risk attitude towards planning - Export ratio
- Internal proactive stimulus
Export Activities:
- Actively searching for orders from abroad Environment:
- Number of export markets - Labour market attractiveness
- Acquisition of information - Supply market attractiveness
- Commitment resources through entry mode - Sales market attractiveness
- Co-operation abroad - Capital/ credit market attractiveness
- External proactive stimulus
- External reactive stimulus
Figure 5-1 The Operationalized Integral Export Performance Model.
138
The Integral Export Performance Model - Estimated
5.2 Method of Analysis

5.2.1 Two Step Procedure
As the model combines both aspects of multiple regression (examining dependence
relationships), and factor analysis (representing unmeasured concepts with multiple
variables), structural equations modeling (SEM) is the preferred multivariate
technique, enabling the estimation of a series of interrelated dependence
relationships simultaneously (Hair et al. 1998). Specifically, with SEM two models
have to be specified: the measurement model, and the structural model (e.g. Byrne
1998). The measurement model defines relations between the observed and
unobserved variables. The structural model defines relations among unobserved
variables. The data are analyzed using LISREL 8.50, a popular software package
for SEM (Jöreskog & Sörbom 1999). In this, I adopt a two-step approach, by first
establishing the quality of the measures of the constructs (i.e. the measurement
models), followed by estimating the whole conceptual or structural model. This
staged approach is recommended, if the theory underlying the constructs is
tentative (as is the case in this research), or if (resulting) measures are less reliable
(Hair et al. 1998, p.600). A second reason pertains to the wish to interpret the
findings for the conceptual model apart from the effect of the measurement models.
Adding to this discussion, Steenkamp & Van Trijp (1991) indicate that the fit of a
model depends on the number of parameters in the measurement and the structural
model, with the fit deteriorating if there are relatively many parameters to be
estimated in the measurement models.
Therefore, the two-step approach is adopted, following the steps proposed by

Anderson & Gerbing (1988): estimate the measurement models for the latent
variables first, using confirmatory factor analysis (CFA), and fix these in the
second stage when estimating the structural model (see also Netemeyer et al. 1997,
and Willenborg 2001 for an application). Yet, not all constructs in the
operationalized models can be fixed using measurement models and reflective
indicators, as is default in LISREL. Contrary to reflective indicators, some indices
are measured with formative indicators.
139
5.2.2 Reflective versus Formative Indicators

Considering the different nature of the three types of variables in the
operationalized model, i.e. single-items, constructs, and indices, a distinction is
being made between reflective (in the case of constructs) and formative indicators
(in the case of indices). In conventional covariance structure modeling, it as
assumed that the latent variable causes the observed variable, and the indicator is a
manifestation or reflection of a construct (Fornell & Bookstein 1982). The
alternative approach reflects the notion that “in many cases, indicators could be
viewed as causing rather than being caused by the latent variable measured by the
indicators” (MacCallum & Browne 1993, p.533). These measures are termed
formative, indicating that the construct is formed or induced by its measures
(Fornell & Bookstein 1982). More mathematically, with formative indicators, the
latent variable is defined as a linear function of the indicators, plus a disturbance
term, contrary to reflective indicators, which are measured as a linear function of
the latent variable. Other differences between reflective and formative indicators
are the following:
• Omission of a formative indicator may alter the construct itself;

• No specific pattern of relationships is necessary among formative indicators of
the same construct;
• Formative indicators do not have error terms;
• Identification of formative models is only possible when embedded in a larger
model, which includes consequences of the latent variable, and may even then
prove to be problematic (Diamantopoulos 1999, Diamantopoulos &
Winklhofer 2001).
Regarding the above, formative models cannot be tested using the well-known
CFA route for measurement models as these only pertain to reflective indicators.
So, in the case of formative models the first step of the two-step procedure has to
be shaped differently. In this, I follow the steps proposed by Diamantopoulos &
Winklhofer (2001).
In the present model, some constructs are measured with reflective indicators
(labeled constructs in Chapter 4), while other constructs are single-item or
measured with formative indicators (labeled indices in Chapter 4). Attitude or
140
perception variables often can be regarded reflective, whereas behavior or

performance often are indices. For instance, Diamantopoulos (1999) argues that
export performance is of a formative rather than the often assumed reflective
nature, pointing out that performance measures often conflict and touch upon
different aspects of export performance (see also section 2.2.1.5, where I point this
out as well). It follows that performance cannot be reflective, as reflective items are
supposed to be interchangeable, while formative indicators can touch upon
different aspects of the composite variable. Following MacCallum & Browne
(1993), I define constructs measured with reflective indicators as latent variables,
and constructs with formative indicators as composite variables; an appropriate
term “in that these causally indicated constructs are, in fact, linear composites of
their indicators, plus a disturbance term.” In the present case, this implies that the
variables related to ‘experience’, ‘level of co-operation’, ‘information acquisition’,
and ‘export performance’ are all composites. On the other hand, the constructs
relating to the attractiveness of the various markets (environment), the external and
internal stimuli, the competencies of the firm and the attitude of the owner-
manager are latent variables.
5.3 Measurement Models for the Latent Variables

In this section, the measurement models for reflective indicators are introduced, as
well as the procedures for combining these reflective indicators into latent
variables (as opposed to composite variables), followed by the findings for these
reflective measurement models.
5.3.1 The Reflective Measurement Models

In LISREL, measurement models for the reflective indicators (i.e. latent variables
and the indicators or items) are formulated in a general form, respectively as (Hair
et al., 1998):
X = Λ xξ + δ (5.1)
Y = Λ yη + ε ( 5.2)
Where:
141
X = a (q × 1) vector of manifest exogenous indicator variables;

Y = a ( p × 1) vector of manifest endogenous indicator variables;
Λ x = the ( q × n) matrix of loadings, showing which manifest indicator variable
loads on which latent exogenous variable;
Λ y = the ( p × m) matrix of loadings, showing which manifest indicator variable
loads on which latent exogenous variable;
ξ = a (n × 1) vector of latent exogenous variables;
η = a (m × 1) vector of latent endogenous variables;
δ = a vector of error terms with expectation zero, and uncorrelated with ξ ;
ε = a vector of error terms with expectation zero, and uncorrelated with η .
In Chapter 4, the data was already subjected to exploratory factor analysis and
conventional reliability analysis. The resulting latent variables and measures are
further evaluated by applying CFA to the measurement models using LISREL.
5.3.2 Procedure for the Measurement Models

The CFA procedure for the evaluation of the measurement models of the latent
variables, consists of two parts. First, the appropriateness of the CFA (do the items
belong to the constructs as hypothesized) is assessed. When the CFA comes up
with satisfactory results, the separate latent variables within the CFA are examined
for the reliability, and validity. Only those latent variables with acceptable
reliability and validity levels are included in the structural model, eventually.
5.3.2.1 Confirmatory Factor Analysis

For each set of indicators that belong to a specific topic, CFA is performed to
assess the one-dimensionality of all multiple-indicator constructs. To determine the
overall fit of the proposed measurement models I use the Satorra-Bentler χ2-
statistic, the Root Mean Square Error of Approximation and its p-value (RMSEA),
the goodness-of-fit statistic (GFI), and the comparative fit index (CFI).
142
All χ2-statistics are absolute fit measures, and indicate the fit between the model
and the data. If the χ2-statistic is significant, the model does not fit the data very
well. In other words, the differences between the original covariance matrix and the
estimated covariance matrix are too large. Hair et al. (1998) indicate that the
traditional χ2-statistic is very sensitive to non-normality and large samples (over
200). As a solution, the Satorra-Bentler scaled χ2-statistic (χ2SB) is used as it
corrects for violations of multivariate normality (Hoogland 1999). Yet, it is still
susceptible to large sample sizes, in that the statistic becomes too sensitive to
departures of fit. Therefore, two other absolute fit indices are taken into account,
namely the RMSEA and GFI. The RMSEA is one of the most informative criteria
and tests whether the model fits the population covariance matrix, if it were
available. Values less than .05 indicate good fit, values of .08 to .10 indicates
mediocre fit, and values larger than .10 indicate poor fit (Byrne 1998, p. 112).
Besides, Rigdon (1996) finds that RMSEA is best suited to use in a confirmatory or
competing models strategy with larger samples. The GFI statistic compares the
model with a model with only an intercept. The higher this statistic, the better the
fit of the model. Sharma (1996, p. 159) indicates a cut-off point of .90. The CFI
statistic, also known as Bentler’s normed comparative fit index, is an incremental
fit measure of the proportionate improvement in fit (defined in terms of non-
centrality) as one moves from the baseline to the target model. The higher this
statistic, the better the fit of the model. Bentler (1992) advises a cut-off point of
.90.
Together, these fit measures provide a good evaluation of the fit of the
measurement models120, and indicate which set of items reflects the constructs in
the CFAs.
5.3.2.2 Outliers
Before we can proceed, we must identify the influential observations, i.e.
observations that lie outside the general patterns of the data set or are likely to
influence the results. Influentials are based on one of four conditions:
120
I do not use any parsimonious fit measures, as their use in an absolute sense is limited to
comparisons between models, which is not the purpose of the present exercise (Hair et al.
1998, p. 658).
143
1) An error in observations or data entry;

2) A valid but exceptional observation that is explainable by an extraordinary
situation;
3) An exceptional observation with no likely explanation;
4) An ordinary observation in its individual characteristics but exceptional in its
combination of characteristics.
Yet, there is still no standard strategy for locating, and more importantly for
dealing with, outliers (Hayduk 1996). In this case, I act as follows. As I have no
backwards control on the data entry, I discard condition 1. Also, influentials as
defined by condition 4 are very hard to detect in an integral model with many
variables that impact each other in one way or the other. Besides, the whole
purpose of this exercise is to find out the relationships between the variables. That
leaves condition 2 and 3.
In order to detect these observations, a box-plot has been inspected for both the
dependent variables ‘export sales’ and ‘total sales’ (after all, export ratio is a
restricted variable, falling between 0 and 100%). Based on the box-plots on total
sales 1995 (first data set), two cases are dropped, and one case based on the plot for
export sales121. After examining the other answers provided by those cases, no
extraordinary situation can be indicated (i.e. condition 2). In case of condition 3
Hair et al. (1998) suggest “…there is no reason for deleting the case, but its
inclusion cannot be justified either”. Therefore, these cases are dropped.
5.3.2.3 Which Specific Measures to Include?

Subsequently, the specific latent variables in the CFAs are assessed on their
internal reliability, convergent validity, and discriminant validity, to determine
which specific latent constructs should be included in the structural model.
Construct reliability is established by calculating composite reliability (CR)

(Sharma 1996, p.165; Hair et al. 1998, p. 612) and the variance extracted (VEE)
(Hair et al. 1998, p. 612), using the formulas below:
121
Case number 847, 984, and 986, respectively.
144
CR =
( λcs )
2
( 5.3)
( λcs ) +
2
εj
VEE =
(λ ) cs
2
(5.4)
(λ ) + ε
cs
2
j
In both equations, ε j represents the measurement error for each indicator (i.e. 1 –
the reliability of the indicator).
“Convergent validity assesses the degree to which two measures of the same
concept are related” (Hair et al. 1998, p. 118). As such, convergent validity is
closely related to the construct’s reliability. After all, if the items show high
internal coherence, the items ‘converge’ to one underlying construct. Therefore, the
lambda’s are supposed to be significant (t ≥ 1.96, and the standardized loadings
( λ cs ) should exceed 0.70. As Hair et al. (1998, p. 612) point out: Indicator
reliabilities should exceed 0.50, which roughly corresponds to a standardized
lambda that is 0.70 or higher.
“Discriminant validity is the degree to which two conceptually similar concepts are
distinct. Now the correlation between the summated scales should be low,
demonstrating that the summated scale is sufficiently different from the other
similar concept” (Hair et al. 1998, p.119). So, to determine discriminant validity,
the measures of constructs are summed and correlations between composite
measures of the constructs of the model were calculated. This is also a first
indication for the existence of multicollinearity. A second test of multicollinearity
is performed, using (1) the Variance Inflation Factor, and (2) the Condition Index,
combined with variance-decomposition matrix (Hair et al. 1998, p. 220).
Summarizing, only those constructs and item are included in the structural model,
that pass the above mentioned assessment criteria.
145
5.3.3 Results of the Measurement Models
5.3.3.1 Goodness-of-Fit of the Measures

Table 5-1 portrays the results of the measurement models for the latent variables
(and reflective indicators). The measurement models are adjusted by examining
double loadings, levels of variance in the indicators explained by the construct
(R2), and high standardized residuals or modification indices on measurement error
covariances122 (e.g., Steenkamp & Van Trijp 1991; Byrne 1998). Notice that the
use of the two-step approach implies that error covariances between indicators
loading on different constructs should be set to zero, even though standardized
residuals or modification indices indicate an improvement of fit when freeing these
parameters. As step two includes fixing the measurement models by using non-
weighted summated scores, the correlation of these error covariances cannot be
incorporated: the indicators only relate to the latent construct they pertain to
according to theory.
As Table 5-1 portrays, almost all model fit measures agree with the norms set (χ2SB
= non-significant, GFI ≥ 0.90, CFI ≥ 0.90, RMSEA < 0.05). Taking into account
the sensitivity of of the fit-indices χ2SB, GFI, and RMSEA to high sample size and
the sample size of 1122, which is well above the optimal sample size of
approximately 200 (Boomsma 1983; Hair et al. 1998, p. 605), the fit levels of the
CFA for Competencies are deemed acceptable as well.
122
These measurement error covariances represent systematic, rather than random,
measurement error in item responses and may derive from characteristics specific either to
items or to the respondents (Aish & Jöreskog 1990). According to Byrne (1998, p. 147)
these can indicate a small omitted factor or a high degree in item overlap in item content
(“such redundancy occurs when an item, although worded differently, essentially asks the
same question”).
146
Table 5-1 Validity Reflective Models.

Reliability Set of Reliability Latent Measurement Model
Measurement Construct Goodness-of-Fit Statistics
Items (N = 1122)
1. CFA Environment χ2SB = 15.20 (p = 0.23, df = 12)
GFI = 1.00
CFI = 1.00
RMSEA = 0.02 (p = 1.00)
1a. Attractiveness Labor Market α = 0.54 CR = 0.63
(2 items, Labor) VEE = 0.49
λav = 0.66
λlow = 0.42
1b. Attractiveness Supply Market α = 0.79 CR = 0.82
(2 items, Supply) VEE = 0.69
λav = 0.83
λlow = 0.73
1c. Attractiveness National Sales NA NA
Market (1 item, Natsal)
1d. Attractiveness International Sales NA NA
Market (1 item, Intsal)
1e. Attractiveness Capital Market α = 0.68 CR = 0.69
(2 items, Capital) VEE = 0.53
λav = 0.73
λlow = 0.71
2. CFA External Stimuli χ2SB = 6.25 (p = 0.25, df = 5)
GFI = 1.00
CFI = 0.99
RMSEA = .02 (p = 0.97)
2a. External Proactive Stimuli α = 0.56 CR = 0.56
(4 items, Extpro) VEE = 0.24
λav = 0.49
λlow = 0.46
2b. External reactive Stimuli NA NA
(1 item, Extre)
3. CFA Competencies χ2SB = 54.13 (p = 0.00, df = 14)
GFI = 0.99
CFI = 0.98
RMSEA = 0.051 (p = 0.45)
3a. Marketing Capability α = 0.73 CR = 0.79
(2 items, Mrktng) VEE = 0.65
λav = 0.81
λlow = 0.70
3b. Product Capability α = 0.78 CR = 0.79
(2 items, Prodcap) VEE = 0.65
λav = 0.81
λlow = 0.77
3c. Competitive Pricing α = 0.65 CR = 0.66
(2 items, Pricing) VEE = 0.50
λav = 0.71
λlow = 0.68
147

Items (N = 1122)
3d. Product Superiority α = 0.65 CR = 0.65
(2 items, Prodsup) VEE = 0.49
λav = 0.70
λlow = 0.69
4. CFA Attitude χ2SB = .53 (p = 0.47, df = 2)
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.84)
4a. Risk Attitude towards Changes α = 0.66 CR = 0.67
(2 items, RTC) VEE = 0.51
λav = 0.71
λlow = 0.64
4b. Risk Attitude towards Planning α = 0.53 CR = 0.59
(2 items, RTP) VEE = 0.43
λav = 0.64
λlow = 0.48
5. CFA Internal Stimuli123 χ2SB = 17.09 (p = 0.25, df = 14)
GFI = 1.00
CFI = 1.00
RMSEA = 0.01 (p = 1.00)
5a. Firm Proactive Technology α = 0.68 CR = 0.68
Internal Stimuli124 VEE = 0.52
(2 items, Ftechn) λav = 0.72
λlow = 0.70
5b. Firm Proactive Financial Internal α = 0.82 CR = 0.83
Stimuli VEE = 0.72
(2 items, Ffinanc) λav = 0.84
λlow = 0.73
5c. Firm Proactive Marketing Internal α = 0.57 CR = 0.57
Stimuli VEE = 0.40
(2 items, Fmrktng) λav = 0.63
λlow = 0.58
5d. Management Proactive Internal α = 0.54 CR = 0.55
Stimuli VEE = 0.38
(2 items, Manager) λav = 0.62
λlow = 0.60
NA = not applicable (latent consists of one indicator). Criteria in italic are below the cut-off point.
123
The inclusion of the reactive dimension of the internal stimuli greatly distorted the CFA.
The variable did not meet any of the criteria and was dropped. Besides, the alpha in chapter
5 was below .6, indicating a less than desirable reliability.
124
Although the reliability of the overall firm proactive factor is near acceptable, the
standardized residuals and the modification indices indicate that three sub-dimensions
should be distinguished, i.e. ‘technology’, ‘financial’, and ‘marketing’.
148
As explained in Chapter 4, the first imputed data set is used to estimate the
measurement models and the structural model. Next, the measurement and
structural model structures are imposed upon the other two imputed data sets. The
results of these two imputed data sets are represented in Appendix 2 (imputed data
set 2) and Appendix 3 (imputed data set 3). These are used to examine the
robustness of the solution given by the first imputed data set: Are the parameters
(sign and significance) stable over the three data sets, or are the imputed values
surrounded with a high extent of uncertainty (due to a high percentage of imputed
missing values) causing the three models to deviate from each other.
5.3.3.2 Reliability of the Constructs

Most Cronbach’s alphas exceed the norm of 0.60. The exceptions are Supply,
External Proactive Stimuli, Risk Attitude towards Planning, Firm’s Proactive
Marketing Stimuli, and Management Proactive Stimuli, although they are higher
than 0.50. Next, the largest part of the measures meet the norm set: CR ≥ .60
(Bagozzi & Yi 1988), and VEE ≥ .50 (Hair et al. 1998, p. 612). The measures for
Supply, Product Superiority, and Risk Attitude towards Planning deviate only
marginally from the norm, and are kept in the study. Three measures (i.e. External
Proactive Stimuli, Firm’s Proactive Marketing Stimuli, and Management’s
Proactive Stimuli) have unacceptable low CRs and VEEs, and are excluded from
the structural models.
5.3.3.3 Convergent Validity

Concerning the convergent validity of the remaining, internal reliable latent
variables, all lambdas are significant (t ≥ 1.96), representing significant
relationships between the indicator and the latent construct. Besides, almost all
standardized lambdas exceed the cut-off point of 0.70. So, based on the convergent
validity, no latent variables or items are excluded125.
125
Although these rules do not automatically apply for one indicator latent variables, such
as the external reactive stimuli, the national sales market and the international sales market,
I use the standardized lambda > .70 rule to determine whether I should include these latent
variables. As the standardized lambda for Extre is .48, which is far below the cut-off point
of .70, I decide to exclude this variable from further analysis. Natsal (standardized lambda
= .88 and Intsal (standardized lambda = .96) are for the same reason included.
149
5.3.3.4 Discriminant Validity and Multicollinearity

To determine discriminant validity, the items under the latent variables are summed
and correlations between these summated measures are calculated. In the present
case, the single-item measures resulting from the CFAs are taken into account as
well. The highest correlation exists between National Sales Market and
International Sales Market (0.457). Examining just the summated measures,
Competitive Pricing and Product Superiority show the highest correlation, i.e.
0.331. Hence, none of the correlations exceed 0.50, demonstrating good divergent
validity (Hair et al. 1998, p. 613). Secondly, the Variance Inflation Factor, and the
Condition Index, combined with variance-decomposition matrix (Hair et al. 1998,
p. 220) show no multicollinearity between the latent variables. Again, no constructs
or items are excluded from further processing.
5.3.3.5 Latent Variables to Include

Summarizing, the various validation and fit indices indicate to include the latent
variables related to the attractiveness of the labor-, supply-, national sales,
international sales-, and capital market, to the internal technological- and financial
firm stimuli, to the firm capabilities with regard to marketing, product superiority,
competitive pricing, and production, and lastly, two components of the owner-
manager’s attitude. The various validation- and fit statistics for the second and
third imputed data sets deviate only marginally from the first data set (see
Appendix 2 and 3). Therefore, these latent variables are to be included in the
structural model, taking into account the relationships found in the theoretical
review in Chapter 2.
5.4 Index Construction of the Composite Variables

In this section, the index construction for formative indicators is introduced (for a
definition see section 5.2.2, as well as the procedures for combining these
formative indicators into composite variables (as opposed to latent variables),
followed by the findings for these formative models.
150
5.4.1 The Formative Models

The direct formative model specifies the measures as correlated cause of a
construct (Edwards & Bagozzi 2000, p. 162), and is represented by the following
equation:
η= Γi X i + ς (5.5)
i
Where:
η = a (m × 1) vector of latent endogenous indices;
Γi = a (i × m) matrix of parameters, showing the magnitudes of the effects of the
X i on η ;
X i = a ( m × 1) vector of manifest exogenous indicators;
ς = a vector of error terms with expectation zero, and uncorrelated with X i .
The measures related to experience, co-operation behavior, and information
acquisition behavior are all formative indicators, and are evaluated as such.
5.4.2 Procedure for the Formative Models

As said before, the appropriateness of composite variables cannot be determined
with conventional reliability analysis, convergent and discriminant validity, due to
the special nature of formative models. Therefore, I follow the guidelines proposed
by Diamantopoulos & Winklhofer (2001, p. 272): “… four issues are critical to
successful index construction: content specification, indicator specification,
indicator collinearity, and external validity.” These four issues should be assessed
for every composite variable.
Content specification relates to the scope of the latent variable, or the “domain of
content the index is intended to capture”. Closely related is the indicator
specification, which should be a census of indicators for the domain chosen in
content specification. Thirdly, as with multiple regression, multicollinearity
between formative indicators results in problematic assessment of the indicator
validity, which can be assessed by each indicator coefficient (i.e. the direct
structural relation between the indicator and the index). Therefore, the
151
multicollinearity should be examined. The last issue is the external validity, which
can be investigated by correlating the individual indicators each to another variable
(external to the index). Only those indicators that have external validity (i.e.
significant correlation with the variable at interest) should be retained. Even better,
some reflective indicators of the construct could be included. Yet, the researcher
should not just go about eliminating indicators, as this will change the conceptual
meaning of the index clearly. To determine nomological validity, the index should
be linked to other constructs with which it is supposed to be linked126. These
systematic steps prove an alternative to scale development in the case of the
composite variables.
5.4.3 Results of the Formative Models

Table 5-2 shows the results for the formative models (‘international experience’,
‘international co-operation’, ‘information acquisition at home’, and ‘information
acquisition abroad’) on the issues mentioned in the previous section.
For the multicollinearity diagnostics, I use the same procedure as described in sub-
section 5.3.2.3, reporting the highest correlation between the formative indicators,
the highest condition index (CI), and the lowest tolerance value. The external
validity can only be examined at a nomological level, as the INTERSTRATOS data
set does not provide other indicators that reflect the composite variables. The
variables used to test this nomological validity are firm size, the number of
countries the firm sold products to, and the export sales. The last column indicates
which part of the correlations are significant (with alpha 5%), and in the
hypothesized direction, in determining nomological validity.
126
This construct should be measured with reflective indicators, to avoid identification
problems.
152
Table 5-2 Validity Formative Models.

Content Indicator Specification Indicator Collinearity Validity
Specification
International Export experience Corrhigh = 0.53 6/6
experience Import experience CIhigh = 6.67
Tolerancelow = 0.72
International Co-operation domestic partners Corrhigh = 0.37 6/6
co-operation Co-operation foreign partners CIhigh = 1.86
Tolerancelow = 0.86
Information Informal information at home Corrhigh = 0.512 5/6
acquisition at home Formal information at home CIhigh = 3.43
Tolerancelow = 0.74
Information Informal information abroad Corrhigh = 0.37 6/6
acquisition abroad Formal information abroad CIhigh = 2.42
Tolerancelow = 0.87
On a whole, all formative models comply with the tests they have been subjected
to, and are included in the structural model.
5.4.4 Issues with Formative Models

Although the theoretical meaning of indices calls for a formative model, this brings
about several programming problems. A necessary condition for the identification
of the residual variance in a composite variable is for that variable to emit more
than one path (MacCallum & Browne 1993). In the present case, the information
and cooperation indices emit only one path to export performance, thereby not
satisfying that condition. This condition is not sufficient; “it is desirable for
composite variables to emit at least two paths that are received by different latent
variables with effect variables” (MacCallum & Browne 1993). This only applies to
the experience index. So, only one of the formative models can be modeled as
such. This will change the relative effect of this index versus the other indicators to
be measured as indices.
Bollen (1994, p.311) provides a solution: “Analysts can incorporate ‘cause’ or

‘formative’ indicators into the current model structure. To do so, define each cause
indicator, x, as equal to a ‘latent’ variable (i.e. x = Iξ`where x is a vector of cause
indicators).” Therefore, to be able to have a theoretically sound model, which
allows for the formative nature of some indicators, and compares the formatives in
a similar way, an extended MIMIC (Multiple Indicators and MultIple Causes)
153
model is chosen, with every formative indicator causing a latent variable (x = 1*ξ,
ε = 0) (see also Bollen, 1994, p. 398, fig. 9.1.a).
5.5 The Structural Model

5.5.1 The Conceptual Model
Figure 5.2 represents the one-year integral export performance model to be tested.
In this model, both the latent variables and composite variables as determined in
the previous sections are incorporated. Although very similar to the original
operationalized model, some measures did not meet the validation standards and
were dropped as described in the previous sections. Next, the depicted model only
displays the relationships between the categories, although there might be
differences within a category as far as the associations go. To be more specific,
within the category ‘Objective Managerial Characteristics’ only the variable ‘age
leaving full time education’ is found to affect the ‘Export Activities’ in the
literature. Thus, the association between ‘Objective Managerial Characteristics’
and ‘Export Activities’ is only modeled for ‘education’ and not for ‘age’ (of
course, ‘age’ is modeled as an antecedent of other categories). Secondly, regarding
the influence of the category ‘Firm Characteristics’ on ‘Export Activities’, Chapter
2 showed that only the ‘firm size’, the ‘international experience’ (in the present
case measured by ‘export experience’ and ‘import experience’), and the ‘linguistic
skills’ might affect the activities conducted in exporting. Therefore, only these are
included in the estimation phase. Moreover, the association between ‘Firm
Characteristics’ and the ‘Subjective Managerial Characteristics’ is established by
‘firm size’ and ‘international experience’ only. Although Chapter 2 did show that
the linguistic skills of the manager might affect the attitude of the owner-manager,
in this case ‘linguistic skills’ is operationalized as the number of languages
mastered by the management team, and is excluded, for that reason. Lastly, within
the category of ‘Export Activities’, proof was found that the information behavior
induced better decisions concerning other activities (see Chapter 2). Consequently,
I include an association between the information variables and the cooperation
behavior within the category of activities. After all, the information collected can
help managers to decide whether they want to cooperate or not. The resulting
model is the structural model to be tested with LISREL.
154
- Firm size
- Export experience
- Import experience
- Linguistic skills
- Marketing capability
Objective Managerial - Product superiority
Characteristics: - Competitive pricing
- Age - Production capability
- Age leaving full time education - Internal technological stimuli
- Internal financial stimuli
Characteristics: Export Performance:
- Risk attitude towards changes - Export sales
- Risk attitude towards planning - Export ratio
Export Activities:
- Actively searching for orders from abroad Environment:
- Number of export markets - Labour market attractiveness
- Acquisition of informal information at home - Supply market attractiveness
- Acquisition of formal information at home - National sales market attractiveness
- Acquisition of informal information abroad - International sales market attractiveness
- Acquisition of formal information abroad - Capital/ credit market attractiveness
- Commitment resources through entry mode - External reactive stimulus
- Co-operation abroad with domestic partners
- Co-operation abroad with foreign partners
Figure 5-2 The Operationalised Integral Export Performance Model – Validated.
The common notation for the structural model is as follows (e.g., Byrne 1998, p.
12-13; Hair et al. 1998, p.646):
η = Γξ + Β η + ς (5.6)
155
Where:
Γ = an ( m × n) matrix of coefficients that relates the n exogenous factors with the
m endogenous factors;
ξ = a (n × 1) vector of latent exogenous variables;
Β = an (m × m) matrix of coefficients that relates the m endogenous factors to one
another;
η = a (m × 1) vector of latent endogenous variables;
ς = an (m × 1) vector of residuals with expectation zero, and uncorrelated with
η and ξ .
5.5.2 Procedure for the Conceptual Model

As said before, LISREL assumes the variables to be multivariate normal. In
practice researchers often work with variables that do not meet this condition127.
Therefore, to ensure a higher level of (multivariate) normality, some variables have
been transformed, using the logarithm. The skewness improves drastically after this
operation. Next, PRELIS is used to compute the covariance matrix, which serves as
the input in the subsequent LISREL analyzes, using the Maximum Likelihood
estimation method128. As mentioned before, separate analyzes are performed for
‘export ratio’ (model 1) and ‘export sales’ (model 2), as I expect different results
and effects for either measure. First, the model is estimated on one of the imputed
data sets for 1995 (chosen at random). The other two imputed data sets validate the
127
One option to take non-normality into account is by using the χ2SB, which requires the
computation of the asymptotic covariance matrix (ACM, see also 5.3.2.1). Yet, the size of
this matrix increases exponentially with the number of observed variables. In our case, the
size of the sample (n = 1122) is too small to calculate the ACM, as PRELIS indicates.
Therefore, the logarithm of several variables has been used.
128
With continuous and categorical variables, the Weighted Least Squares (WLS) method
is probably the best estimation method, being distribution-free. This also demands
estimating the ACM. Yet, the size of the sample might inhibit the computation of the ACM.
In that situation, the ML method is a good alternative (Boomsma 1983).
156
model found in the first imputed data set. The models are evaluated using the
following goodness-of-fit statistics: χ2 = non-significant, GFI ≥ 0.90, CFI ≥ 0.90,
RMSEA < 0.05, keeping in mind the sensitivity of χ2 and CFI for large sample
sizes (for a explanation of these statistics, see 5.3.2). Using the modification
indices indicated by the program, changes have been made to improve the fit of the
model. Only those alterations with theoretical ground, and with a significant
change in χ2 have been incorporated. Although the program sometimes indicates to
include other indicators for a latent variable, I did not incorporate these, as this
would alter the structure between the indicators as found in literature.
5.5.3 Results Structural Model

Table 5-3 shows the results of the estimated structural model using the first
imputed data set for model 1 (with export ratio as dependent variable) and in model
2 (with export sales as dependent variable)129. The goodness-of-fit statistics are
displayed, as well as the standardized parameters130, and the t-value of that
respective parameter (in parentheses). The results for the second and third imputed
data set are represented in a similar way in Appendix 2 (second imputed data set)
and Appendix 3 (third imputed data set). In describing the results, the first data set
is used as the point of departure, while I use the other two models (stemming from
the two other imputed data sets) to see whether these results are robust over the
129
The variables ‘number of export markets’, ‘channel’ and ‘proactive’ are dropped for
computational problems with the covariance matrix. The first two for their multicollinearity
with ‘export experience’, which is a much stronger and more often used variable (see
Chapter 2), and the second for the weakness of the measurement scale (i.e. binomial)
compared with the other variables and its subsequent disturbing effect on the model.
Besides, the program indicates several modification indices that are significant. Some of
these are not theoretically justifiable, and are, therefore, not adopted. Other modification
indices indicate logical relationships between the error variances of the information,
cooperation, and attitude variables. Also, the latent variables ‘export experience’, import
experience’, ‘firm size’, and ‘languages’ seem to be correlated. These are included in the
model.
130
“The standardized coefficients indicate the relative importance of the causal
relationships; they closely approximate effect sizes comparable to standardized beta
coefficients in regression analysis” (Hair et al., 1998).
157
data sets131. After all, if all three data sets show a significant positive relationship
between two variables, this is stronger support for the existence of this relationship
compared with a relationship only found in one or two of the data sets. Deviations
between the three data sets can derive from a reversed significant sign, a parameter
that is significant in the second and/or third data set versus non-significance in data
set 1, or is now non-significant while significant in the imputed data set used for
model building132. Both the consistencies and the deviations over the data sets are
discussed in the results section, either to strengthen the support for a association (if
this support is found consistently in all three data sets) or to refine the support
found in data set 1.
Table 5-3 Results Structural Models Integral Export Performance.

Model 1 (export ratio)133 Model B (export
sales)
Goodness-of-fit statistics
χ2 = 2087.89 (.00, df = 528) χ2 = 2081.53
RMSEA = .051 (.35) (.00, df = 528)
GFI = .91 RMSEA = .051
CFI = .84 (.36)
GFI = .91
CFI = .86
Size and significance of effects
On ‘risk attitude towards changes’ from
Age .15 (4.44) .58 (4.44)
Age leaving ft education -.06 (-1.67) -.04 (-1.67)
Firm size -.18 (-4.76) -.14 (-4.76)
Export experience -.19 (-4.40) -.10 (-4.40)
131
As explained in Chapter 4, the first imputed data set is used to estimate the correct
model, which is subsequently imposed on the two other imputed data sets. Herewith I check
whether the fact that the data sets are imputed due to missing values impacts the
parameters, i.e. to see whether the results are robust.
132
Therefore, the appendices show deviations between the parameter found in the second or
third imputed data set from the one found in the first data set (parameter is shaded).
133
Although including a correlation between the latent constructs ‘risk’ and ‘formal’
improves the fit-statistics, the resulting standardized parameters deteriorate significantly
(values which are over 1). Therefore, although suggested by the modification indices, this
alteration is not included.
158
Import experience -.18 (-4.44) -.09 (-4.44)

On ‘risk attitude towards planning’ from
Age .12 (3.54) .12 (3.54)
Age leaving ft education .04 (1.19) .04 (1.19)
Firm size -.09 (-2.21) -.09 (-2.21)
On ‘acquisition of informal information at home’ from
Supply market attractiveness -.06 (-1.59) -.06 (-1.59)
Labor market attractiveness .07 (2.00) .07 (2.00)
National sales market attractiveness .04 (1.25) .04 (1.25)
International sales market attractiveness .03 (.94) .03 (.94)
Capital market attractiveness -.01 (-.32) -.01 (-.32)
Age leaving ft education .00 (.06) .00 (.06)
Firm size .11 (3.12) .11 (3.12)
Export experience .09 (2.28) .09 (2.28)
Import experience .02 (.48) .02 (.48)
Linguistic skills -.01 (-.42) -.01 (-.42)
Risk attitude towards changes .00 (-.08) .00 (-.08)
Risk attitude towards planning .03 (.90) .03 (.90)
On ‘acquisition of formal information at home’ from
Labor market attractiveness .14 (4.03) .14 (4.03)
International sales market attractiveness -.02 (-.54) -.02 (-.54)
Capital market attractiveness -.07 (-1.84) -.07 (-1.84)
Age leaving ft education -.01 (-.42) -.01 (-.42)
Firm size -.13 (3.91) -.13 (3.91)
Linguistic skills -.05 (-1.41) -.05 (-1.41)
Risk attitude towards changes -.10 (-2.49) -.10 (-2.49)
On ‘acquisition of informal information abroad’ from
Supply market attractiveness .00 (-.09) .00 (-.09)
Labor market attractiveness .02 (.60) .02 (.60)
International sales market attractiveness -.05 (-1.68) -.05 (-1.68)
Capital market attractiveness -.04 (-1.27) -.04 (-1.27)
Age leaving ft education -.04 (-1.44) -.04 (-1.44)
Firm size .12 (3.76) .12 (3.76)
Linguistic skills .05 (1.64) .05 (1.64)
Risk attitude towards changes -.06 (-1.69) -.06 (-1.69)
Risk attitude towards planning -.06 (-1.72) -.06 (-1.72)
On ‘acquisition of formal information abroad’
159

Labor market attractiveness -.01 (-.16) -.01 (.16)
International sales market attractiveness .00 (.10) .00 (.10)
Capital market attractiveness .03 (.80) .03 (.80)
Firm size .22 (6.72) .22 (6.72)
Import experience .05 (1.31) .05 (1.31)
Linguistic skills .00 (-.15) .00 (-.15)
Risk attitude towards changes -.02 (-.64) -.02 (-.64)
Risk attitude towards planning -.03 (-.88) -.03 (-.88)
On ‘cooperation abroad with domestic partners’ from
National sales market attractiveness -.02 (-.50) -.02 (-.50)
International sales market attractiveness -.01 (-.17) -.01 (-.17)
Capital market attractiveness .04 (1.18) .04 (1.18)
Firm size -.10 (-2.62) -.10 (-2.62)
Risk attitude towards changes .00 (.11) .00 (.11)
Acquisition of informal information at home .10 (3.28) .10 (3.28)
Acquisition of formal information at home .07 (2.45) .07 (2.45)
Acquisition of informal information abroad .17 (5.22) .17 (5.22)
Acquisition of formal information abroad .15 (4.80) .15 (4.80)
On ‘cooperation abroad with foreign partners’ from
Labor market attractiveness -.08 (-2.62) -.08 (-2.62)
Capital market attractiveness .02 (.65) .02 (.65)
Age leaving ft education .06 (2.44) .06 (2.44)
Firm size .04 (1.07) .04 (1.07)
Risk attitude towards changes .07 (1.83) .07 (1.83)
Acquisition of informal information at home .13 (4.87) .13 (4.87)
Acquisition of formal information at home .14 (5.23) .14 (5.32)
On ‘export performance’ from
160
Supply market attractiveness .15 (4.84) .06 (2.65)

National sales market attractiveness .03 (.88) .02 (.81)
Capital market attractiveness .05 (1.32) .00 (.11)
Firm size .06 (1.66) .54 (23.35)
Internal technological stimuli -.01 (-.33) -.02 (-.90)
Internal financial stimuli .00 (.17) .00 (.18)
Marketing capability .08 (2.26) .05 (1.92)
Product superiority -.07 (-1.46) -.02 (-.58)
Competitive pricing -.15 (-3.45) -.08 (-2.61)
Production capability .04 (1.03) .05 (2.01)
Age leaving ft education .04 (1.54) .01 (.62)
Age .01 (.35) -.04 (-2.02)
Import experience .02 (-.48) .00 (.16)
Risk attitude towards changes .03 (.76) -.01 (-.37)
Risk attitude towards planning .05 (1.35) .01 (.27)
Acquisition of informal information at home .04 (1.27) -.01 (-.37)
Acquisition of formal information at home .08 (2.78) -.01 (-.32)
Co-operation abroad with domestic partners -.01 (-.20) .00 (.12)
Co-operation abroad with foreign partners .06 (1.88) .09 (3.80)
As Table 5-3 shows, the fit of the two models is acceptable to good. Both the
RMSEA and the GFI indicate good fit, although the CFI statistic is below .90
(indicating a mediocre fit), and the χ2-statistic is very significant134. In the other
two imputed data sets (see Appendix 2 and 3), the fit is somewhat better, with a
CFI approaching the cut-off point of .90. The models are inspected for patterns and
significant relationships.
Describing the specific results, only the significant parameters are explicated for
their size and direction. As the export-sales and export-ratio models show similar
results for the relationships between the ‘Environment’, ‘Firm Characteristics’,
‘Managerial Characteristics’, and ‘Export Activities’, I do not distinguish in
describing the interrelationships between these categories for the two models. As
134
As said before, one has to keep in mind the sensitivity of this statistic for large samples.
161
only the various associations with ‘Export Performance’ deviate between model 1
and 2, I treat these separately. This goes for all three imputed data sets.
Management and Firm Effects on ‘Subjective Managerial Characteristics’

Concerning the influence of the objective managerial and firm characteristics on
the two attitudinal components, the estimates show that the younger, higher
educated owner-manager in a larger SME with more export- and import experience
is more risk taking, both towards changes as towards taking more ad hoc decisions.
These patterns occur in each of the three data sets, so are rather stable. It should be
noted, that these relationships are stronger for the first attitudinal component. In
case of the planning-attitude, the education seems not to have any impact, and firm
size is a significant antecedent in only the first imputed set.
Environment Effects on ‘Export Activities’

Both environmental, managerial, and firm-related factors are antecedents of
‘Export Activities’. The environment impacts the export activities related to
information and cooperation in various ways135. First, the changes, either positive
or negative, on the market for capital seem not to have any significant effect on the
information or cooperation behavior of the respondents in either imputed data set.
In contrast, the other four markets do impact this behavior significantly, although
not in a consistent manner. The more positive the changes on the supply market,
the more formal information is collected abroad, and the more cooperation is
sought with foreign partners. Besides, the second and third imputed data set show
some proof that these changes positively impact the amount of formal and informal
information at home. The labor market specifically affects the amount of
information sought at home (both formal and informal): if this market aggravates,
more information providers are consulted. In addition, if the developments on the
labor market are more positive, more cooperation is sought with partners from
abroad. These findings are consistent over the three data sets. Looking at the
national sales market in the first imputed data set, the results show that SMEs turn
more to information providers from abroad (both formal and informal), when the
135
Keep in mind the coding of these environmental variables: a low score means positive
changes on the specific market, while high scores indicate a negative outlook of the
respondent on that market.
162
national market worsens. Nevertheless, these associations are much weaker in the
other two data sets. At home, more formal information is collected (according to
the first and second data set), while the informal information collected goes
unchanged with a less attractive national sales market. In addition, the first and
third imputed data sets show that more cooperation is sought with foreign partners
if the national sales market is disappointing (but not with domestic partners).
Positive developments on the international sales market marginally increase both
the informal information collected abroad (significant on 90% in set one and two
only), and the cooperation with foreign partners (only in set one and three).
Firm Effects on ‘Export Activities’

Based on literature, only the effects of certain firm characteristics are tested on the
export activities, i.e. the firm size, the international experience (stemming from
either exporting or importing), and the linguistic skills of the management team.
The size of the firm appears to be a strong determinant of the extent to which
exporting SMEs collect information: the larger the firm the more informal
information is collected both at home and abroad, and the more formal information
providers are consulted abroad. Conversely, the smaller SME collects more formal
information at home. Moreover, it is also the smaller SME that seeks cooperation
with domestic partners. The results for firm size are all consistent over the three
imputed sets, and are regarded as stable. With regard to the international
experience, experience developed by exporting is a stronger determinant than the
know-how gained by importing. In all three sets, a more experienced SME in
exporting collects more foreign information, with stronger impact on informal
information, and cooperates more with both domestic and foreign partners abroad
(only some sporadic positive results for domestic information is found in the three
sets, and are, therefore, not included). Import-related experience appears to have no
significant associations with information behavior at all, but does encourage SMEs
to seek cooperation abroad with foreign partners. For this cooperation with either
domestic or foreign partners abroad, the results of the three imputed data sets also
display that better linguistic skills (i.e. more languages mastered by the
management team) improve the extent of cooperation. Weak support is found for
the positive impact of this fluency in foreign languages on just one component of
information behavior, i.e. on collecting informal information abroad (90%
significant in only two data sets).
163
Management Effects on ‘Export Activities’

The next set of influentials on the export activities are the managerial
characteristics, in this case the education level and the risk attitudes of the owner-
manager. Consistently over all data sets, the level of education only impacts the
level of cooperation with foreign partners positively, the other export activities are
unaffected. Some weak support (in two out of the three sets) is found for the impact
of the respondent’s attitude. The less positive the respondent feels about changes,
the less formal information is sought at home and the less informal information is
collected abroad. Yet, there is some indication that change-averse managers want
to cooperate more with foreign partners (90% reliability in set one, 95% in set
two). The way the owner-manager feels about planning has no significant effect on
any of the behavior variables.
Effects within ‘Export Activities’

Within the category of ‘Export Activities’, I state that the information behavior
could induce cooperation behavior. The results of all three imputed data sets
clearly show this is true for all information categories and for both types of
cooperation. The strongest impact originates from the information collected
abroad, and the associations are stronger for the cooperation with foreign partners.
Environment Effects on ‘Export Performance’

Lastly, Table 5-3 shows the associations of all determinants with the two
performance measures. For the environmental factors, both export ratio and export
sales improve with deteriorated developments on the supply market, and with an
improved international sales market, while the capital market bears no impact at all
(consistent over the data sets). Besides, in the second and third data set, a
worsening national sales market improves the export ratio and export sales
significantly and directly.
Firm Effects on ‘Export Performance’

The findings pertaining to the firm characteristics demonstrate that the size of the
firm especially impacts export sales (the relationship with export ratio is barely
significant at 90%, and only in set three significant at 95%). Comparing all
standardized loadings, this is the single most important determinant of export sales.
164
Secondly, turning the attention to the capabilities of the SME, the results are not
very consistent over the imputed data sets except for export experience. This
variable associates strongly and positively with both export ratio and export sales
in all models. Combining the three sets of results, two sets do show a 95%
significant negative association between import experience and export ratio, but no
association with export sales. In addition, some support can be found that SMEs
who are capable in marketing have higher performance measures (two out of three
sets), and that export ratio suffers from competitive prices. For the production
capability, the product superiority, and the linguistic skills of the management team
the results are too dispersed to give a general sign or size.
Management Effects on ‘Export Performance’

The direct impact of the manager on performance is hardly noticeable. The results
with respect to the manager’s education and age are again so diffuse that no overall
conclusion can be drawn. The overall results for the risk attitudes show that these
have no association at all with neither performance measure.
Export Activities Effects on ‘Export Performance’

Lastly, the effects of the export activities on export performance are examined.
Clearly, the extent to which the SME consults foreign information sources
enhances export performance significantly, with an even stronger association with
export sales than with export ratio. The same goes for the cooperation with foreign
partners: export sales improves if the extent of this collaboration increases (export
ratio only at a significance level of 90%). For the other behavioral components the
results are more scattered over the three sets. Although the domestically acquired
informal information seems not significant in the results presented here, in the
second and third data set there is a negative association with export ratio.
5.5.4 Conclusions
The models clearly indicate which of the various determinants are important for
export performance, especially when combined over the three data sets. A good
way to summarize and order the total effects of the various determinants on export
performance is examining the total effects of the determinants on export
performance for their size and their significance. After all, some variables relate to
export performance only indirectly, by impacting other antecedents of export
165
performance136. Table 5-4 shows these total effects on export ratio and export sales
(t-values in parentheses).
Table 5-4 Total Effects of Variables on Export Performance.

Total effects of variable: Model 1: on export ratio Model 2: on export sales
Environment:
Supply market .18 (4.18) .10 (1.98)
Labor market .03 (.82) .01 (.13)
National sales market .08 (1.99) .09 (1.77)
International sales market -.10 (-2.37) -.17 (-3.56)
Capital market .04 (.95) .00 (.00)
Firm size .12 (3.86) 1.01 (26.82)
Internal financial stimuli .01 (.17) .01 (.18)
Product superiority -.09 (-1.46) -.04 (-.58)
Language fluency .15 (1.86) .16 (1.61)
Import experience -.01 (-.32) .02 (.94)
Objective Managerial Characteristics:
Education .27 (1.45) .16 (.73)
Age .09 (.59) -.42 (-2.25)
Subjective Managerial Characteristics:
Risk attitude towards change .01 (.21) -.03 (-.56)
Risk attitude towards planning .05 (1.03) -.01 (-.12)
Export Activities:
Domestic informal information .05 (1.64) .01 (.24)
Domestic formal information .08 (3.26) .01 (.32)
Foreign informal information .27 (7.30) .30 (6.84)
Foreign formal information .19 (3.49) .21 (3.31)
Domestic cooperation -.01 (-.20) .00 (.12)
Foreign cooperation .05 (1.88) .11 (3.80)
In parentheses the t-values. Bold figures are significant at 95%, italic figures are significant at 90%.
From this table, the most important determinants can be deduced. Merging the
three sets of results, key positive antecedents of both export ratio and export sales
136
“Total effects include the direct effects, as well as indirect effects that result from
correlations among exogenous variables and circular or reciprocal effects” (Jöreskog &
Sörbom 1993, p. 114).
166
are: Encouraging developments on the international market, the size of the SME
(although the impact is much larger for export sales compared to export ratio), the
international experience gained through exporting, the information collected
abroad (both formal and informal), and the extent to which SMEs cooperate with
foreign partners abroad. Negative antecedents are optimistically perceived
developments on the supply- and (to a lesser extent) the national sales market, and
an emphasis on competitive pricing. Possibly SMEs turn their attention more to the
national market when things worsen on the supply and national market. Weaker
results (two out of three sets) are found for the positive effect of linguistic skills
and a better-educated owner-manager on performance. Next to these antecedents
influencing both performance measures, it would be interesting to find some
variables that only influence one of the export performance measures.
Unfortunately, the results concerning many variables are so dispersed over the
three data sets, that no clear conclusion can be drawn on their importance. Very
cautiously, it might be concluded that it is especially profitable for export sales to
emphasize the marketing capabilities. The other associations found with export
performance are less probable over the three sets, and could very well be a product
of chance, as they only occur on one of the imputed data sets.
Labeling the top five antecedents in data set one for both performance measures,
for export ratio these are (1) export experience, (2) foreign informal information,
(3) supply market, (4) firm size, and (5) foreign formal information, while for
export sales these are (1) firm size, (2) export experience, (3) foreign informal
information, (4) foreign cooperation, and (5) the international sales market.
Especially the importance of the international sales market, the export experience
and the collection of informal information at home for both sales and ratio returns
in all sets, just like the impact of firm size on export sales. The other top five
antecedents in data set one are significant in the other two data sets, but not top
five. Other variables that enter the top five in the other imputed data sets are the
foreign formal information acquisition, the national sales market, domestic
cooperation and the marketing capability. The last two associations are not
considered very valid, as they come forward in only one imputed data set, which is
probably more to be attributed to chance than to a traceable pattern. The other two,
‘national sales market’, and ‘foreign formal information’ might not be in the top
167
five in all three sets for both export performance measures, but do have significant
and large total effects in all sets.
Concluding, the importance of the collection of information abroad, and to a lesser

extent of cooperation abroad (not in top five but consistently significant) is
established for SMEs. These are actual activities an owner-manager can decide
upon. Outside of the control of the manager (at least on the short term) are the fact
that export experience and a larger employee base increases export performance.
Also, there seems to be an outside pull on exporting, namely developments on the
national and international sales market.
168
Chapter 6
The Information Behavior of
Exporting SMEs
Chapter 5 intended to estimate a one-year integral export performance model,

including many variables and relationships. Unfortunately, only few relationships
were found to be significant. One reason for this finding is the multitude of
relationships estimated, and the fact that only a subpart of all relationships found in
literature (Chapter 2) could be operationalized (Chapter 4). Yet, this exercise did
result in establishing the importance of the collection of information for export
performance. By focusing on a specific module of the integral model, so that a
better insight can be gained on the effect of this component on export performance.
Therefore, in this chapter, we concentrate on the information behavior displayed by
European SME-managers, including the effect of this information on export
performance, and the influence of the manager on the information behavior.
Information is an increasingly important input in the internationalization process.

As Czinkota (2000) suggests: “Information and its management is even more
important in the international setting, where entirely new parameters and
environments are encountered”. International expansion makes business
environments more turbulent, more complex and, consequently, harder to predict.
A large body of market research literature (e.g., Douglas & Craig 1989) suggests
that proper use of market information reduces these uncertainties in the firm’s
decision process, improving the firm’s ability to cope with opportunities and
threats on the export market, and, consequently, the firm’s competitiveness.
Gemünden (1991) specifically dedicates part of his review to the important role of
information activity for exporters, as this is the only activity found to have a
positive relationship with all three export performance measures used (export
169
profitability, export growth, and export intensity). This significance of ‘the firm’s
utilization of international marketing research’ is also found in Zou & Stan’s
(1998) review, and in our review (Chapter 2)137. The overall conclusion clearly is
that information acquisition and use especially improves business in international
business, which is more uncertain than domestic business.
The need for and acquisition of foreign market information should be considerable
among (new) exporters. This is consistent with the findings of Rose & Shoham
(2002), who find that “acquiring and responding to market information are
particularly important in an export marketing context, where changes in the
economic, political, and consumer environment are likely”. They also state that
“exporting generally increases the complexity and dynamism of the external
environment, which increases the need for market intelligence and
responsiveness”. As strategic marketing literature shows, the search for
information is an important part of the market orientation concept (e.g., Kohli &
Jaworski 1990; Narver & Slater 1990). Lately, the interest for exporters,
international information behavior and market orientation has come together in the
attention for an international concept of market orientation. Therefore, the first
section elaborates on this conceptual background of international information
behavior. Based on this theoretical discussion, three questions are raised. First,
what does this international information behavior look like for the SMEs in our
longitudinal data set (section 6.2)? Second, I investigate whether information
behavior is impacted by the specific personality of the manager, in view of the
pivotal role of the manager in SMEs. Third, the impact of the acquisition on export
performance is explored, partly based on the findings in Chapter 2. These last two
topics will be investigated both in a static and a longitudinal manner (see section
6.3).
137
The dedication of a special issue of International Marketing Review (2002) to export
information acquisition and use emphasizes the importance and growing attention for the
role of information in exporting.
170
The Information Behavior of Exporting SMEs
6.1 Role of Information in Internationalization
If we consider exporting as an innovation (cf. Axinn 1988), the importance of

information in internationalization becomes even more obvious. For example,
Julien (1998, p.40) states “[I]n other words, innovation means listening to the
environment and thus requires implicit or fairly well-organized technological,
competitive, and commercial scanning”. In this section, the conceptual background
of information as part of an international market orientation is unfolded, including
the theoretical relationships between information behavior, managerial personality,
and export performance.
6.1.1 International Market Orientation

In marketing literature, market orientation is a well-known and often researched
construct (Kohli & Jaworski 1990; Narver & Slater 1990; Jaworski & Kohli 1993;
Kohli, Jaworski & Kumar 1993). Kohli & Jaworski (1990) define market
orientation as the generation and dissemination of information, and the subsequent
use hereof in adapting the firm’s offer to the needs of the customer. In a later study,
they find market orientation to be positively related to ‘overall performance’ as
seen by managers (Jaworski & Kohli 1993), comparable to the positive link of
‘marketing orientation’ on ‘business performance’ established in Narver & Slater
(1990), and in Slater & Narver (2000). Therefore, it might be imperative for
businesses to be market-oriented when doing business. The logical question is,
whether this applies to international business as well.
Based on these general constructs of market orientation, Cadogan &

Diamantopoulos (1995) offer an export-oriented measure of market orientation.
After all, as the international performance often is evaluated separately from
national performance, there is need for an international derivation of market
orientation. This export-oriented market orientation consists of four components,
namely export intelligence generation, dissemination, responsiveness, and a
coordinating mechanism. The first component concerns all actions that lead to the
establishment of export market intelligence by a firm. The other three components
of the concept concern the internal sharing of intelligence (dissemination) to come
to an integration and interpretation of the information on which actions are
171
designed and implemented (responsiveness), governed by an internal coordinating

mechanism that should solve all frictions.
6.1.2 International Information Collection

Despite the interdependency of the four aspects of international market orientation,
in this chapter I choose to focus on the first part of export market orientation, i.e.
the intelligence generation, or, in other words, the information acquisition behavior
of firms. As a first step in obtaining an export market orientation, this acquisition
of information on customers and competitors is a key element.
To obtain this intelligence, numerous sources of information are available. Several

researchers occupied themselves with the specific information used by exporting
firms, categorizing them according to the type of information. Both Johanson &
Vahlne (1977), and Seringhaus (1986; 1993) distinguish between objective and
experiential information. Objective information embodies published (often
statistical) information from primary or secondary sources, while experiential
knowledge is personally acquired through direct market or customer contact.
Another classification is that between formal and informal information (e.g., Hart,
Web & Jones 1994). More specifically, Souchon & Diamantopoulos (1996, 1999)
classify the information acquisition genres into export market research (i.e. formal,
systematic and objective information gathering), export assistance (i.e.
governmental export promotion), and export market intelligence (i.e. informal
information gathering, through day-to-day business). In a way, the three
categorizations can be easily integrated, as formal information incorporates both
export market research and export assistance, and informal information can be seen
as export market intelligence.
6.1.2.1 International Information Collection in Industrial SMEs

To be market-oriented a firm needs to collect information. Yet, often there appear
to be differences between large firms and SMEs in their acquisition behavior.
Julien (1998, p.40) states: “Technological and commercial information is
expensive, and changes quickly. It is essential to the long-term vision, and is taken
into account by large firms through their research centers and other specialized
departments. Small businesses are particularly vulnerable in this area because of
the limited resources they are able to devote to creating, obtaining and assessing
172
information”. The question arises, therefore, whether the size of the firm affects
information acquisition behavior. If we consider firm size as a resource, and
information as costly, larger firms will have more resources, and, thus, collect more
(formal) information (Mohan-Neill 1995). Diamantopoulos et al. (1990) conclude
that companies systematically using export market information are larger than
nonusers. Samiee & Walters (1990) find similar results; with large companies more
actively collecting export market information than smaller firms do.
More specifically, SMEs often find export market research too costly and,
therefore, mainly rely on export market intelligence (Souchon & Diamantopoulos
1996). Julien (1998) confirms this, and deems that “small businesses must join or
construct networks that allow them to obtain information they need at a lower
cost”. Belich & Dubinsky (1995) also study the extent to which small companies
use internal versus external sources. Their results suggest that this choice is partly
related to the management strategy, claiming that, due to the dominant position of
the owner-manager in SMEs, smaller companies sometimes give priority to
subjective and personal goals, resulting in a less objective use of external
information sources. The results of a later study also suggest that information
processing within small firms may not mirror patterns of activities in their larger
counterparts (Belich & Dubinsky 1999).
Resulting, smaller firms gather less information than their larger counterparts,
especially less formal information, partly due to the scarcity of resources and the
personality of the owner-manager.
A minor remark on the fact that the INTERSTRATOS data set consists of
manufacturing SMEs: it appears that industrial firms collect less intelligence than
consumer companies (Avlonitis & Gounaris 1997)138. Therefore, the question is
how much intelligence the industrial SMEs in the INTERSTRATOS data set collect.
138
Besides, consumer firms have a higher dissemination of information, and a higher
responsiveness; concluding, the behavioral component of market orientation is less
developed in industrial firms versus consumer companies (Avlonitis & Gounaris 1997).
173
6.1.3 Information and Export Performance

As mentioned in 6.1.1, firms holding a market orientation, collecting and using
information, have better firm output than firms who are less market-oriented: a
market orientation leads to a better use of core capabilities and transforms these to
competitive advantages, thereby boosting business performance (Slater & Narver
1994). In view of the relatively young line of research on international market
orientation, only little empirical proof can be found for the impact of international
market orientation on export performance.
Does & De Mortanges (1998) find a positive significant influence of overall

market-orientation only on subjective measures of export performance, but not on
objective measures. So, managers tend to believe that being market-oriented is
rewarding. Cadogan, Diamantopoulos & De Mortanges (1999) relate each of the
four dimensions of their international market orientation concept with export
performance. For this, they use three performance measures, i.e. export sales
revenue divided by total number of employees, a weighted performance index
(assessing the performance relative to management’s objectives and the firm’s
satisfaction with the achievement of each objective), and a global assessment of the
firm’s export success. All four dimensions correlate positively with the
performance measures. On the other hand, Rose & Shoham (2002) conclude that
export sales is neither significantly related to overall (general) market orientation,
nor to any of its components. Yet, change in export sales, export profits, and
change in export profits relate both significantly and positively to overall market
orientation, intelligence generation, and responsiveness, but not to intelligence
dissemination139. They adhere this result to the fact that dissemination of
information is less important in SMEs due to the size of these enterprises.
More empirical results can be found on the relationship between the first
component of international market orientation, i.e. intelligence creation through the
collection of information and on export performance (see also section 2.6.3). Julien
139
Rose & Shoham (2002) were aware of the four international dimensions proposed by
Cadogan, Diamantopoulos & De Mortanges (1999), but for a lack of good measurement of
the coordinating mechanism, they chose to apply Kohli & Jaworski’s general framework
(1990).
174
& Ramangalahy (2003) hypothesize for instance that “[T]he SMEs’ limited
capacity to acquire information and use sources is a major factor explaining their
low level of involvement and performance on export markets”. Therefore, there
seems to be a strong relationship between collecting information and output. In
this, researchers investigate either the effect of information search and the amount
of information sources consulted in general, by dividing the information collected
into formal/informal or into assistance/research/intelligence sources, or even by
examining specific information sources.
Related to the hypothesis stated above, Julien & Ramangalahy (2003) conclude that
the collected information does impact export performance positively through
competitive strategy. Years earlier, Donthu & Kim (1993) already find an increase
in the overall number of information sources consulted to improve export
performance. On the other hand, using the same proxy, Diamantopoulos & Inglis
(1988) conclude that more export intensive firms consult fewer information
sources. According to the authors, due to a learning effect these firms obtain a
higher internal capacity to cope with the complexity of exporting, diminishing the
need for external information sources.
Other studies focus on the effects of specific types of information. Bijmolt & Zwart
(1994) find a positive impact of export market research on export performance,
prescribing the necessity of an analysis of the export country, the competitors, and
the consumers. Similarly, Christensen et al (1987) find successful exporters to be
three times as likely to enter a foreign market based on detailed market studies than
non-successful ones. Dominguez & Sequeira (1993) conclude the same concerning
these formal methods of information acquisition. Moini (1995) also recommends a
systematic formal exploration of export market possibilities, next to informal visits
to foreign markets, as do Katsikeas, Piercy & Ioannidis (1996), and Hart & Tzokas
(1999). Koh (1991) even finds indirect positive results through the improvement of
strategic decisions, thereby enhancing performance. A few studies find
nonsignificant results for research activities (Madsen 1989; Bourantas & Halikias
1990; Koh 1991; De Luz 1993).
Another form of formal information acquisition is the use of export assistance. Yet,
the results are mixed. Reid (1984) finds that governmental promotion programs
175
raise the likelihood of exporting to a new foreign market within the next year
(which is positively related to export performance)140, but Cafferata & Mensi
(1995) conclude that SMEs experience the assistance provided by government and
local agencies to be inadequate. Christensen et al. (1987) also conclude that the
successful exporters in their sample exploited generalized market information less
than ex-exporters, although they do note that such generalized data can be
important bases for subsequent detailed market studies and, therefore, are not
without use. “As firms gain export experience and rely less and less on public and
private information sources, their information needs probably become too specific
for ordinary information sources to satisfy them” (Denis & Depelteau 1985).
Contrary, Cavusgil & Naor (1987) find that exporters search more information
through the US Department of Commerce and through the State Agency than non-
exporters. Furthermore, Bell et al. (1991) found 62 percent of the Irish SMEs in
their sample to have obtained export sales that were directly attributable to
participation in export marketing training programs. Lastly, Dominguez &
Sequeira (1993) do not find any significant results for export promotion agencies.
Many studies conclude that SMEs benefit greatly from informal- or export market
intelligence, including information acquired through relationships with foreign
distributors and customers, and through visiting trade fairs or the export market
(Reid 1984; Denis & Depelteau 1985; Cavusgil & Naor 1987; Cavusgil & Zou
1994; Styles & Amber 1994; Moini 1995). For example, Moini (1995) finds visits
to foreign markets to be imperative in the development of export markets. On the
other hand, Koh & Robicheaux (1988), Koh (1991), and Katsikeas, Piercy &
Ioannidis (1996) all find that the frequency (or regularity) of face-to-face contacts
with dealers is nonsignificant. Dominguez & Sequeira (1993) also fail to prove the
significance of informal market intelligence.
Summarizing, export market research and intelligence both tend to improve the
outcomes of the export planning process, while export assistance is often too
140
With regard to the positive relationship between the number of markets an SME serves
and the export performance, see Diamantopoulos & Inglis (1988), Holzmüller & Kasper
(1991), Beamish et al. (1993), Dominguez & Sequeira (1993), Naidu & Prasad (1994).
176
general to be of specific use, although a cautious conclusion might be that it does

enhance export performance. This is in accordance with Denis’ & Depelteau’s
(1985) conclusion on exporting SMEs: “The study reveals the overwhelming
influence on export expansion of information acquired from business transactions
as opposed to reliance on private or public information services”.
Based on the above, the following hypothesis is postulated:

H1: the amount of information collected by an SME has a positive effect on
export performance.
6.1.4 Personality of the Manager

Various antecedents precede strategic activities, such as the personality of the
manager (see Chapter 2). Yet, this is a rarely investigated concept in the studies on
international market orientation. In their first conceptualization of the international
market orientation, Cadogan & Diamantopoulos (1995) already criticize that most
researchers look upon market orientation from a behaviorist approach, foregoing
influences of unobservable phenomena such as ways of thinking. See for instance
Avlonitis & Gounaris (1999), who summarize the literature on the development of
market orientation, and conclude that the development of this concept is influenced
“only by company-specific factors but also by market-specific ones”, but do not
mention any management-related factors. Based on Aaby & Slater (1989),
Cadogan & Diamantopoulos (1995) plead for the inclusion of cognitions, given the
importance of manager’s in shaping export success. Avlonitis & Gounaris (1997)
also distinguish between market orientation as both behavior ánd attitude, finding a
close interrelationship between the two141. In 2000, Narver & Slater replicate their
famous 1990-study, extending it with a measure for ‘entrepreneurial orientation’,
next to ‘market orientation’ (Slater & Narver 2000). The ‘entrepreneurial
orientation’ is build up around the innovativeness, risk taking, and competitive
141
Although Avlonitis & Gounaris (1997, 1999) use the term ‘marketing orientation’
instead of ‘market orientation’, they conceptualise the behavioral component of ‘marketing
orientation’ closely related to the market-orientation concept by Kohli & Jaworski (1990).
The attitudinal operationalization of ‘marketing orientation’ is more related to the
management orientation. Actually, market orientation encompasses much more than a mere
orientation on marketing, so, to avoid confusion, I keep using the term ‘market orientation’.
177
aggressiveness of the enterprise. Although both orientations were tested as

independent regressors of business performance, there appears to be a strong
positive and significant correlation between the two concepts. Therefore, there
might be a link between the attitude of the owner-manager and the market-oriented
behavior that the firm displays.
As indicated by Belich & Dubinsky (1995; 1999), information behavior in

internationalizing small firms is partly determined by the decision-maker’s
priorities. This is in line with Churchill & Lewis (1983), who find that managers of
small companies not only act for the good of the firm, but also make decisions that
satisfy personal goals. Several other instances of the impact of managers on
(international) information behavior can be found. For instance, the individual
needs of the decision-maker should be used in shaping export assistance programs,
rather than the needs of the organizations where these managers are employed
(Gray 1997). Moreover, Walters (1996) concludes that the information acquisition
activity in the firms in his survey is heavily influenced by management’s
commitment to exporting. This is confirmed by Cadogan et al. (2001), who test
management’s commitment with exporting as part of ‘export leadership’, and find
that a higher commitment positively influence export market-oriented behavior.
The manager’s personality influences the information behavior not only through
shaping the goals and needs, but also through the way managers cope with
uncertainty (see also sub-section 2.5.2). As said before, (perceived) uncertainty in
international settings induces firms to collect information. For example, Wright &
Ashill (1998) hypothesize that the more uncertainty management perceives (due to
a highly diverse and volatile environment), the higher the information needs will
be. Related to this way of thinking, Menon & Varadarajan (1992) found the
volatility of the environment to influence the uncertainty as perceived by
management and, thus, to influence the propensity to seek and use information.
Lang & Calantone (1997) conclude that manager’s perception of the environment
influences the information seeking behavior: higher perceived threats or
opportunities in the environment increase small firms information seeking.
Therefore, the perception of uncertainty by managers influences the information
acquisition behavior. This notion of perceived uncertainty is closely linked to the
risk-taking propensity as defined by Ahmed (1985): Risk-taking propensity is
178
dealing with uncertainties and the degree of readiness to bear it. This risk
preference is one of the most important characteristics of small business
owners/entrepreneurs (e.g., Lumpkin & Dess 1996). As scanning the environment
can solve the uncertainty in the environment (Milliken 1987), the entrepreneur with
a low risk-taking propensity (or, a high perceived uncertainty), could be expected
to search for more information to solve this uncertainty. On the other hand, as
exporting is often an innovating activity (see Axinn 1988), with all potential
barriers, risk-avoiding managers could also decide not to search for information as
they might look upon exporting as a too risky strategy. Kohli & Jaworski (1990)
conceptualize senior management factors as an important antecedent of market
orientation, including the ‘risk aversion of top management’ (negative impact), and
‘top management attitude towards change’ (positive). Being responsive to changing
customer/client needs, can be seen as a continuous innovative behavior. Based on
this model, in a later study, Jaworski & Kohli (1993) hypothesize that a risk-averse
top management leads to employees generating fewer market knowledge for fear of
a failure142. Avlonitis & Gounaris (1999) actually prove that de extent of risk-
aversion negatively relates to the development of market orientation.
Therefore, if a market orientation improves performance, and risk-averseness

deteriorates market orientation, a negative attitude towards risk does not pay off.
Besides, the export performance literature (see also 2.5.2) shows that attitudes also
have direct effects on export performance as well. Concerning the risk attitude in
an international setting, this has been measured both as a general personality trait
and as a specific precondition towards the risk of exporting. For example, Dichtl,
Köglmayr & Müller (1990) find that a manager who is less rigid, more willing to
change, or to accept product-policy risks generates higher export sales growth. Yet,
other authors find non-significant direct effects on export performance from the
attitude towards risk in exporting (Kaynak & Kuan 1993; Bijmolt & Zwart 1994;
Moini 1995). On a general performance level, Hoskisson, Hitt & Hill (1991) find a
positive relationship between risk-taking and performance. Besides, they turn the
direction of the relationship between performance and managerial risk orientation
142
In the results, the relationship between ‘top management risk aversion’ and ‘intelligence
creation’ is non-significant. Yet, the risk posture of top management did affect the
‘responsiveness’ of the firm negatively.
179
around, hypothesizing that a poor performance induces managers to become less

risk averse (cf. Cyert & March 1963)143. In their study on Ghanaian manufacturing
companies, Pattillo & Soderbom (2000) confirm that “firms with more risk averse
managers who face high risks have lower profit rate variability and lower mean
profit rates”. According to the authors, an explanation can be found in theory,
which “predicts that the higher the risk aversion indicator, the greater should be the
attempt at profit smoothing (lower variance) with attendant lower expected
profits”. Moreover, Naman & Slevin (1993) establish that a lack of entrepreneurial
style (i.e. being risk-taking, proactive, and innovative) in a demanding environment
worsens financial performance144.
The following hypotheses result:

H2: SMEs with a risk-averse owner-manager collect fewer export market
information.
H3: SMEs with a risk-averse owner-manager display a lower export
performance.
6.2 Information Behavior INTERSTRATOS

In this section, a description of the information collection behavior of the
INTERSTRATOS respondents is given for all three years. Not only the type of
143
Although a stream of research on risk taking and general performance exists (e.g., Singh
1986, Bromiley 1991), one should be careful to compare these results with the risk-taking
propensity defined in this dissertation. In these studies, risk taking is often operationalized
as the uncertainty of outcomes, either ex post (by measuring the ex post variability of a
firm’s ROI), or ex ante by asking various managers to predict the outcome of operations
and taking the variability in these predictions (e.g., Bromiley 1991). I use the perception of
or attitude towards risk, i.e. risk-taking propensity, as discussed in the entrepreneurship
literature (e.g., Julien 1998), rather than the classical decision theory definition of risky
choice (March & Shapira 1987; Hoskisson, Hitt & Hill 1991). It must be noticed, however,
that the results on the economic approach of risk suggest that a poor performance induces
higher risk taking, but that risk taking deteriorates performance (Singh 1986; March &
Shapira 1987; Bromiley 1991).
144
This is the same measure for entrepreneurial style as Slater & Narver (2000) used.
180
information sources (informal or formal), but also the number of information

sources consulted by the respondents are described.
6.2.1 The Information Sources

Table 6-1 shows the twelve information sources in the INTERSTRATOS survey and
the categories they pertain to.
Table 6-1 Information Sources in the INTERSTRATOS Survey, Categorized cf.

Hart, Webb & Jones (1994), and Souchon & Diamantopoulos (1996).
Hart, Webb & Jones (1994) Souchon & Diamantopoulos INTERSTRATOS
(1996)
Formal information Export Market Research Training institutions
Business consultants
Credit agencies
Research institutions
Export Market Assistance Chambers of Commerce
International organizations
Public promotion fairs
Informal information Export Market Intelligence Suppliers
Customers
Export clubs
International trade fairs145
The respondents have been asked, whether they consulted these information
sources in the last twelve months, either at home or abroad. Table 6-2 shows the
percentage of respondents who indicate that they consulted the specific source in
1991, 1993, and 1995146. In parentheses, the rank order of the use of the
information sources is indicated. Besides, the top three sources are represented
bold.
145
The categorization of ‘export clubs’ and ‘(international) trade fairs’ as export market
assistance- or export market intelligence sources is open for discussion. However, taking
the description of these sources into account, these are ‘experiential knowledge’: These
sources evolve around the gaining and keeping business contacts, which can be labeled as
informal knowledge (cf. Dominguez & Sequeira 1993).
146
These descriptive statistics are based on the original data set, which still include missing
values.
181
Table 6-2 Frequency with which Information Sources are Consulted at Home
or Abroad, in 1991, 1993, and 1995.
Consulted at home Consulted abroad
Information source 1991 1993 1995 1991 1993 1995
(N=977) (N=721) (N=681) (N=722) (N=754) (N=515)
Training institutions 44,1 (4) 34,4 (5) 38,3 (4) 9,6 (6) 4,2 (8) 6,0 (8)
Business consultants 61,3 (1) 50,1 (1) 54,8 (1) 9,4 (7) 7,0 (5) 10,7 (5)
Credit agencies 44,6 (3) 32,7 (6) 35,5 (6) 7,6 (9) 4,2 (8) 3,9 (10)
Research institutions 14,1 (11) 11,9 (11) 13,7 (11) 3,9 (10) 3,7 (10) 5,8 (9)
Chambers of Commerce 38,9 (7) 37,3 (3) 40,9 (3) 10,7 (5) 6,5 (7) 11,3 (4)
International organizations 4,1 (12) 3,9 (12) 5,4 (12) 3,0 (12) 2,4 (12) 2,5 (12)
Public promotion fairs 23,3 (8) 16,4 (10) 16,3 (10) 3,9 (10) 3,3 (11) 3,5 (11)
Suppliers 43,6 (5) 36,1 (4) 37,9 (5) 21,1 (3) 12,4 (3) 19,8 (3)
Customers 39,1 (6) 32,5 (7) 33,0 (7) 21,7 (2) 16,6 (2) 22,9 (2)
Export clubs 19,5 (9) 19,3 (9) 17,9 (9) 8,3 (8) 6,9 (6) 8,3 (7)
National trade fairs 52,6 (2) 43,1 (2) 43,8 (2) 11,9 (4) 8,0 (4) 10,3 (6)
International trade fairs 17,2 (10) 19,8 (8) 21,7 (8) 36,3 (1) 22,9 (1) 35,0 (1)
As Table 6-2 shows, the domestic sources most often consulted over the three years
are the business consultants, and the national trade fairs, closely followed by the
Chambers of Commerce, the suppliers, the customers, the training institutions and
the credit agencies. More than one-third of all respondents consistently indicate to
use these information providers at home. Regarding the foreign sources, the first
striking issue is the overall much lower usage of foreign sources. International
trade fairs are an exception, being even more often visited abroad (on average
30%) than at home (on average 19%). Again, foreign suppliers and customers are
consulted comparatively often by firms (around 20%), as are foreign national trade
fairs, business consultants and Chambers of Commerce, although to a much lesser
extent (about 10%). Thus, both at home and abroad, SMEs consult mainly trade
fairs, followed by information collection at the customers and suppliers. This
supports the notion that SMEs use export market intelligence relatively more often
than export market research (cf. Leonidou & Katsikeas 1997). Still, the importance
of formal research cannot be ignored, given the use of business consultants,
Chambers of Commerce, training institutions and credit agencies, especially in the
home market. The respondents use export assistance from public promotion fairs to
a limited extent: At home (about 16%), and abroad (about 3,5%). Even worse is the
use of international organizations (on average 4% at home against about 2.5% on
the foreign market).
182
Looking at the rather consistent pattern of the use of information providers over the
three years, a careful conclusion can be drawn on the (perceived) usefulness of the
various sources. Considering organizations as learning entities, a supplier of
information whose merits are disappointing will not be consulted in a later stage
(Nonaka 1994). Therefore, the resources, which are preferred consistently over the
years, might be the most useful in the eyes of the owner-manager. On the other
hand, the low percentages for some sources might also indicate a low level of
awareness of these sources (McAuley 1993). However, combined with the previous
argumentation, if a source proofs to be valuable, word gets around and more SMEs
might consult the provider two years later.
6.2.2 The Number of Information Sources

In addition to the specific information sources, and the extent to which the
respondents have consulted these suppliers, the total amount of information used is
examined. As explained in sub-section 6.1.3, the extent of information collection
bares a close relation with export performance.
In this study, I define the quantity of domestic (foreign) information gathered as the
total number of domestic (foreign) information sources that the company consulted
in the last year. Both variables range from zero (no information source consulted at
all) to twelve (all information sources consulted). Besides, these numbers are
investigated for export market research, export market assistance, versus export
intelligence sources, as well as for formal versus informal information. Table 6-3
shows the average number of information providers referred to over the three years
(number of domestic sources outside, number of foreign sources in parentheses).
183
Table 6-3 Number of Domestic (Foreign) Information Sources Consulted.

1991 1993 1995
EMR EMR EMR
1.64 1.29 1.42
Formal Formal Formal
(.30) (.19) (.26)
2.30 1.87 2.05
EMA Total EMA Total EMA Total
(.48) (.31) (.44)
.66 4.02 .58 3.37 .63 3.59
(.18) (1.47) (.12) (.98) (.17) (1.40)
EMI Informal EMI Informal EMI Informal
1.72 1.72 1.51 1.51 1.54 1.54
(.99) (.99) (.67) (.67) (.96) (.96)
EMR = export market research, EMA = export market assistance, EMI = export market intelligence.
Out of a possible twelve providers that the respondents could consult at home or
abroad, on average only four domestic providers are used, versus just over one
foreign provider on average. On the domestic market, mostly formal sources are
being referred to, while on the foreign market informal sources are more popular.
The percentage of SMEs that indicate not to have used any information supplier at
all is distressingly high, i.e. 13.7%, 24.7, and 20.6% for 1991, 1993, and 1995 on
the domestic market, versus 44.9%, 65.0%, and 49.9% on the foreign market.
Should I find evidence for a positive link between the collection of information and
export performance, there is ample room for improvement in the acquisition
behavior of our respondents.
6.3 Managerial Personality, Information Behavior,

and Export Performance
In this section, the estimations of both the static and longitudinal information
behavior model are displayed. Analogous to the procedure in the previous chapter,
first, the reflective attitudinal models are validated using confirmatory factor
analysis. Second, the conceptual (structural) static and longitudinal model are
estimated for all three imputed data sets. As explained before, these models are
estimated separately for both export sales and export ratio.
6.3.1 Methodology
In this chapter, the panel dataset 1991-1993-1995 is used (sample size is 1122), to
be able to measure both the static and the longitudinal effects of information and
184
attitude on export performance. The original dataset is imputed multiply using

NORM (see also chapter 4)147. As with the integral model (see Chapter 5), I use
structural equations modeling to estimate the direct and indirect effects between the
constructs and variables as hypothesized in section 6.1. More specifically, I again
apply a two-step MIMIC approach in LISREL after fixing the measurement models
(see section 5.3).
Similar to the one-year integral export performance model in Chapter 5, the first
imputed data set is used to estimate the measurement models and the structural
model. Next, the measurement and structural model structures are imposed upon
the other two imputed data sets. The results of these two imputed data sets are
represented in Appendix 4 (imputed panel data set 2) and Appendix 5 (imputed
panel data set 3). These are used to examine whether the solution given by the first
imputed data set is robust: Are the parameters (sign and significance) stable over
the three data sets, or are the imputed values surrounded with a high extent of
uncertainty (due to a high percentage of imputed missing values) causing the three
models to deviate from each other?
6.3.2 Measurement of Reflective Models

Table 6-4 shows the validation of the reflective models measuring the two risk
constructs. All model fit measures agree with the norms set (χ2 = non-significant,
GFI ≥ 0.90, CFI ≥ 0.90, RMSEA < 0.05). Although not all alpha’s, CR, and VEE
meet the criteria, all concepts are included, considering the rather exploratory set
up of the survey, and the wish to build a longitudinal model with consistent
concepts over the years. Convergent validity is proven by all lambda’s being
significantly related to the various concepts, and most of the standardized lambdas
being higher or close to 0.7. Concerning discriminant validity, of course there is a
considerable correlation between the same concepts over the years. It is common
147
As explained in Chapter 4, due to the insufficient computer capacity, first only the 1995
data set could be computed, which is used in Chapter 5. For the present chapter, a higher
capacity computer was available, and all three years could be imputed simultaneously.
Therefore, the 1995 estimates of the CFA’s can deviate somewhat from those in Chapter 5,
as these are build upon the one-year imputed data, which did not take into consideration the
longitudinal co-variances between the variables. Yet, the differences are only marginal.
185
knowledge that attitudes only change slowly. Yet, between the two different risk
measures, no correlations over .5 can be found. Furthermore, the condition indices,
variance inflation factors, and tolerance values all indicate no multicollinearity.
The results for the other two imputed data sets are displayed in Appendices 4 and
5. These results are very similar to the results displayed in Table 6-4. That is, all
CFAs show good fit and the reliabilities of both the set of measurement items and
the latent constructs are alike to the once in the first set. Besides, the lambdas
found in the second and third imputed set only deviate marginally from the once in
the first set (i.e, only .01 or .02). Therefore, the solution for the reflective attitude
models is stable over the three imputed data sets.
Table 6-4 Measurement of the Reflective Attitude Models 1991, 1993 & 1995.
Reliability Set of Reliability Measurement Model
Measurement Latent Goodness-of-Fit Statistics
Items Construct (N = 1122)
CFA Attitude 1991 χ2 = .21 (p = 0.65, df = 1)148
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.90)
Risk towards changes α = 0.63 CR = 0.63
(2 items, Risk1) VEE = 0.46
λav = 0.68
λlow = 0.67
Risk towards planning α = 0.57 CR = 0.60
λav = 0.72
λlow = 0.57
CFA Attitude 1993 χ2 = 1.87 (p = 0.17, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .04 (p = 0.52)
λav = 0.64
λlow = 0.56
λav = 0.58
148
Because of the size of the longitudinal dataset, the program was not able to estimate an
asymptotic covariance matrix, so the normal χ2 is used.
186

Items Construct (N = 1122)
λlow = 0.46
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.84)
λav = 0.71
λlow = 0.64
λav = 0.63
λlow = 0.48
Italic criteria are below the cut-off point.
6.3.3 Measurement Formative Variables

In operationalizing the formative information composites, a simple count of the
number of times a respondent indicates to have used an information source
suffices. This approach somewhat relates to existing scales by measuring the
amount of information collected or the breadth of search (cf. Marandu 1995;
Nijssen & Douglas 1999; Yeoh 2000). In this, the index encompasses part of the
statements used to measure the dimension Intelligence Creation in MARKOR149,
such as “We collect industry information by informal means” (Kohli, Jaworski &
Kumar 1993).
149
The INTERSTRATOS information sources encompass much of the content in statements 1,
2, 3, 5, 6, 7, and 10 in the scale for Intelligence Creation, i.e. “In this business unit, we meet
with customers at least once a year to find out what products or services they will need in
the future”, “Individuals from our manufacturing department interact directly with
customers to learn how to serve them better”, “In this business unit, we do a lot of in-home
market research”, “We poll end users at least once a year to assess the quality of our
products and services”, “We often talk with or survey those who can influence our end
users’ purchases (e.g., retailers, distributors)”, “We collect industry information by informal
means e.g., lunch with industry friends, talks with trade partners)”, and “We periodically
review the likely effect of changes in our business environment (e.g., regulation) on
customers”, respectively.
187
This can be done on three levels: (1) Total information at home and abroad, (2)
total amount of formal and informal information at home and abroad, and (3) the
total amount of export assistance, research, and intelligence information at home
and abroad. The first distinction is too broad, and leaves us with no possibilities to
say anything about the relative importance of various types of information. The
second categorization is used in Chapter 5, and is, to preserve consistency, applied
in this chapter as well. Next, the third, most specific categorization gives even
more information on the type of information. As one goal of this chapter is to zoom
in on information behavior, I also estimate the models with this categorization.
Export sales and export ratio will be measured as such. Table 6-5 shows the
various formative variables used.
Table 6-5 Formative Variables in the Model.
Variables Measurement
Total number of informal information sources used at home Sum of 7 possible sources
Total number of formal information sources at home Sum of 5 possible sources
Total number of informal information sources used abroad Sum of 7 possible sources
Total number of formal information sources abroad Sum of 5 possible sources
Total number of export research sources at home Sum of 4 possible sources
Total number of export assistance sources at home Sum of 3 possible sources
Total number of export intelligence sources at home Sum of 5 possible sources
Total number of export research sources abroad Sum of 4 possible sources
Total number of export assistance sources abroad Sum of 3 possible sources
Total number of export intelligence sources abroad Sum of 5 sources
Export sales Ratio scale
Export ratio Ratio scale
6.3.4 The Static Model

Figure 6.1 portrays the static model, with information behavior either consisting of
four or six separate variables and the risk avoiding propensity consisting of two
components. The hypothesized effects are portrayed as well. Export performance is
either export ratio or export sales.
188
H3 (-)
Risk Avoiding Export
Propensity Performance
H2 (-) Information H1 (+)

Behaviour
Figure 6-1 The Static Information Model.
This general model is the premise for the models to be estimated. Fixing the
lambdas of the reflective attitudinal variables, four one-year models are been
estimated150 (see results in Table 6-6, and Table 6-7):
(1) Export performance measured with export ratio, information categorized as
formal and informal information;
(2) Export performance measured with export sales, information categorized as
export market research, assistance, and intelligence information;
Both tables show the goodness-of-fit of the four models, the standardized loadings
and the t-values (in parentheses).
Appendix 4 and 5 display the results of the structural model for the other two
imputed data sets. Both the consistencies and the deviations over the data sets are
discussed in the results section, either to strengthen the proof for a hypothesis (if
proof is found consistently in all three data sets) or to refine the proof found in data
set one.
150
The modification indices clearly indicate that the information variables have correlated
errors. As these co-variances can be explained theoretically, these modifications are
accepted and included.
189
Table 6-6 Results Static Model; Categories Formal – Informal Information.

Model 1 (export ratio) Model 2 (export sales)
χ2 = 22.00 (.18, df = 17) χ2 = 30.43 (.023, df = 17)
RMSEA = .017 (1.00) RMSEA = .026 (1.00)
GFI = 1.00 GFI = .99
CFI = 1.00 CFI = .99
On ‘collecting informal information at home’ from
Risk towards changes -.15 (-2.54) -.15 (-2.54)
Risk towards planning .09 (1.47) .09 (1.47)
On ‘collecting formal information at home’ from
On ‘collecting informal information abroad’ from
Risk towards planning -.06 (-1.03) -.06 (-1.03)
On ‘collecting formal information abroad’
Risk towards planning -.02 (-.36) -.02 (-.36)
Risk towards planning -.09 (-1.33) -.07 (-.95)
Domestic informal information .02 (.58) .08 (2.83)
Domestic formal information -.01 (-.27) .03 (.95)
Standardized loadings, t-values in parentheses, significant parameters at 95% bold, at 90% italic.
Table 6-7 Results Static Model; Categories Intelligence - Research –

Assistance Information.
χ2 = 30.31 (.09, df = 21) χ2 = 38.80 (.01, df = 21)
RMSEA = .02 (1.00) RMSEA = .026 (1.00)
GFI = 1.00 GFI = .99
CFI = 1.00 CFI = .99
On ‘collecting export intelligence information at home’ from
On ‘collecting export research information at home’ from
On ‘collecting export assistance information at home’ from
190

On ‘collecting export intelligence information abroad’ from
Risk towards planning -.06 (-1.03) -.06 (-1.03)
On ‘collecting export research information abroad’ from
On ‘collecting export assistance information abroad’
Risk towards planning -.08 (-1.32) -.05 (-.95)
Dom. intelligence information .02 (.58) .08 (2.83)
Dom. research information -.02 (-.73) .01 (.43)
Dom. assistance information .02 (.65) .04 (1.48)
For. intelligence information .21 (7.17) .33 (12.01)
For. research information .12 (4.11) .27 (9.98)
For. assistance information .12 (4.22) .22 (7.98)
All four one-year models show a good fit, looking at the χ2 statistic, the RMSEA,
GFI and CFI151, although models 2 and 4 (using export sales) show a too large χ2
(in the second and third imputed data set, these statistics do show a good fit).
Looking at the hypotheses, the results show the following152. Concerning
hypothesis 1 (“The amount of information collected by an SME has a positive
effect on export performance”), the results show that the foreign information
sources indeed have a strong and significant positive impact on export
performance. Besides, the strongest relationship is between the informal or
intelligence information and performance. In addition, the relationships are
stronger when ‘export sales’ is used as a dependent variable. Furthermore, both the
research and assistance information from abroad is beneficial for performance. All
these results come forward in each of the three imputed data sets, and are,
therefore, considered stable. Looking at the domestic information used, only the
informal (or intelligence) information impacts export sales, but not export ratio. In
addition, the other two imputed sets show a strong positive effect from the
151
See section 5.3.2 for an argumentation for and explanation of these fit statistics.
152
Only the significant results are discussed.
191
assistance information gathered on the home market for both performance

measures. Therefore, there is support for hypothesis 1, but mostly for information
gathered from sources used abroad. Considering the low use of foreign information
sources (see Table 6-2), there is ample room for improving export performance by
increasing the collection of information form these sources.
When regarding hypothesis 2 (“SMEs with a risk-averse owner-manager collect

fewer export market information”), a clear distinction can be found between the
two risk variables used in this model. The first attitudinal variable touches upon the
feelings of the managers towards changes and has a significant negative impact on
the collection of informal (intelligence) information in all four models. Yet, the
second attitudinal variable, i.e. how managers reflect upon planning and proven
procedures, does not affect the amount of informal (intelligence) information
sought. Although the broader categorized models (1 and 2) show that both risk
variables impact the formal information acquired at home in all three imputed data
sets, the more specifically categorized models (3 and 4) refine these results. ‘Risk
towards changes’ influences the collection of formal information both at home and
abroad negatively (in all three sets). ‘Risk towards changes’ mainly affects the
research component of formal information negatively, at home and abroad, while
the only abroad-acquired export assistance is affected negatively, especially
according to the second and third data set153. The three data sets also give support
for a positive effect of ‘Risk towards planning’ on the amount of export research
and assistance information asked for (two out of three sets significant). The other
relationships between the risk variables and information variables are non-
significant. Therefore, the attitude of managers towards planning only influences
one component of information collection, i.e. export market assistance (part of the
formal information category), and positively, which is converse to the hypothesis
stated. By distinguishing between two different constructs of risk attitude (as
indicated by both EFA and CFA), a more specified picture emerges of the impact
153
On a 90% level of confidence, ‘Risk towards changes’ does affect export assistance
information negatively as well, while the relationship with export research information is
just not significant on 90%, but these results are not found in the other two imputed data
sets, and are, therefore, not included as stable.
192
of the risk attitude of the manager. Therefore, there is proof for hypothesis 2, but
only for the first attitudinal component (‘risk towards change’).
Lastly, the third hypothesis proposes that SMEs with a risk-averse owner-manager
display a lower export performance. Here, only the ‘risk towards changes’ has a
significant, negative impact on export ratio and, especially, on export sales. Again,
it is mainly the way managers think about sticking with the old ways that affects
especially export sales, and, to a lesser extent, export ratio. Therefore, there is
proof for hypothesis 3, but only for the attitude of owner-managers towards
changes.
6.3.5 The Longitudinal Model

The basis for the longitudinal model consists of three sets of static relationships as
stated in the three hypotheses above. Therefore, these hypotheses are tested within
each year, leading to nine static hypotheses (three for each year). In longitudinal
model specification, a common assumption is that measurement errors in the
manifest variables are correlated over the years (Jöreskog & Sörbom 1999; Byrne
1998)154. For example, Byrne (1998, pp. 359-360) states that “[G]iven the known
possibility of memory carryover effects associated with measuring instruments,
such correlated error parameters would appear to be perfectly reasonable.” These
correlations over time are added to the model. Next, the longitudinal relationships
are added to form the longitudinal model.
6.3.5.1 Longitudinal Hypotheses

In the next step, longitudinal relationships are added, and tested. Although there is
just very little literature available on longitudinal relationships, most studies being
cross-sectional, various hypotheses can be stated for relationships between the
variables over the years. In this way, I can validate whether the inclusion of these
relationships has a significantly positive effect on the fit of the model. Below, the
longitudinal hypotheses are stated155.
154
Jöreskog & Sörbom (1993, p. 297) admonish that “[E]very correlation between error
terms must be justified and interpreted substantively.”
155
Although general strategic literature gives some evidence on the lagged associations
between the risk taking and corporation performance, the inclusion of these relationships is
193
First of all, there exists a possibility that SMEs do no use the information collected
in the same year as they collect it, leading to information having a lagged effect on
performance: Information is sometimes collected to be stored for future use
(Souchon & Diamantopoulos 1997).
H4: The amount of information collected has a lagged positive effect on export
performance.
Next, past performance might have an impact on information behavior. For

instance, Lages & Montgomery (2001) find that past performance plays a crucial
role in building SMEs’ commitment to exporting and to the determination of their
current marketing strategy”. In addition, if SMEs suffer from a resource problem,
an increase in export performance might induce exporting SMEs to heighten their
information collection. After all, information is costly, and if export performance
improves, the SME might have more resources or be more inclined to commit
resources to exporting.
H5: The export performance has a lagged positive effect on the amount of
information collection.
Adding these two longitudinal relationships, four models are estimated, analogous
to the one-year results, distinguishing between the two export performance
measures (export ratio or export sales) and the two information categorizations
not applicable in this context. Some authors hypothesize that taking risks has a negative
lagged effect on performance. For instance, Bromiley (1991) finds that corporate risk
taking appears to result in poor performance. Others state that a positive performance
induces managers to be even more risk averse than before, hypothesizing a negative lagged
effect (Singh 1986; Bromiley 1991; Hoskisson, Hitt & Hill 1991). Yet, the risk taking
measures used in these studies do not touch upon the risk attitude of a manager, but use the
uncertainty of the economic outcomes of the corporation. For example, Bromiley (1991)
measures risk taking as the variety in security analysts’ forecasts of the income of a
corporation. Therefore, I do not incorporate these lagged relationships from and to risk
attitude. A second reason is that I include the risk attitude in every year, and that literature
shows that attitudes are not very prone to change (Eagly & Chaiken 1993).
194
(formal or informal information versus export market intelligence, assistance or

research information):
export market research, assistance, and intelligence information.
6.3.5.2 Results Longitudinal Model

Table 6-8 represents the results of the models 5 through 8. First, the three static
hypotheses are examined within each of the three years (associations between risk
attitude, information behavior, and export performance within a year). Next, the
two longitudinal hypothesis are tested, by investigating the parameters of the
relationships between variables in subsequent years. For each model, the fit-
statistics, size and direction of the parameters, and the t-values are displayed.
Table 6-8 Results Longitudinal Model; Categories Formal - Informal

Information.
χ2 = 931.97 (.00, df = 239) χ2 = 1030.75 (.00, df = 239)
RMSEA = .053 (.10) RMSEA = .056 (.007)
GFI = .94 GFI = .94
CFI = .92 CFI = .92
Size and significance of effects within one year (hypotheses 1, 2 &3)
On ‘acquisition of informal information at home’ from
1991 1993 1995 1991 1993 1995
Risk towards changes .01 -.10 -.17 .02 -.06 -.10
(.18) (-1.91) (-2.97) (.33) (-1.20) (-1.82)
Risk towards planning .14 .25 .10 .14 .23 .07
(2.44) (4.98) (1.82) (2.37) (4.62) (1.25)
On ‘acquisition of formal information at home’ from
Risk towards changes -.07 -.02 -.20 -.05 .03 -.14
(-1.21) (-.37) (-3.58) (-.88) (.60) (-2.57)
(2.46) (4.02) (3.11) (2.18) (3.66) (2.45)
On ‘acquisition of informal information abroad’ from
Risk towards changes -.20 -.17 -.17 -.22 -.06 -.09
195
(-3.31) (-3.28) (-3.17) (-3.53) (-1.22) (-1.63)

Risk towards planning .14 .12 -.05 .16 .09 -.08
(2.30) (2.43) (-.86) (2.56) (1.90) (-1.46)
On ‘acquisition of formal information abroad’
(-3.67) (-2.90) (-2.24) (-3.67) (-.79) (-1.44)
Risk towards planning .03 .01 .00 .03 -.03 -.01
(.46) (.13) (.05) (.50) (-.56) (-.16)
(-.93) (-2.08) (-1.39) (-5.75) (-4.43) (-3.75)
Risk towards planning .01 -.01 -.11 .08 -.01 -.08
(.20) (-.16) (-2.12) (1.49) (-.26) (-1.89)
Acquisition of informal information .05 -.03 .04 .06 .02 .08
at home (1.80) (1.01) (1.37) (2.21) (.94) (3.83)
Acquisition of formal information at .17 .06 -.01 .14 .03 .03
home (6.46) (2.52) (-.25) (5.42) (1.29) (1.57)
Acquisition of informal information .13 .20 .13 .21 .27 .20
abroad (4.51) (4.40) (5.04) (8.09) (11.28) (9.58)
Acquisition of formal information .06 .19 .08 .17 .24 .17
abroad (2.24) (3.32) (2.99) (6.32) (9.79) (8.14)
Size and significance of longitudinal effects (hypotheses 4 & 5)
On ‘acquisition of informal information at home’ in 1993 from
Export performance 1991 .04 (1.55) .08 (2.51)
On ‘acquisition of formal information at home’ in 1993 from
On ‘acquisition of informal information abroad’ in 1993 from
On ‘acquisition of formal information abroad’ in 1993 from
Export performance 1991 .14 (4.72) .29(10.08)
On ‘export performance’ in 1993 from
Acquisition of informal information .01 (.24) .05 (2.28)
at home in 1991
Acquisition of formal information at .06 (2.52) .06 (2.78)
home in 1991
Acquisition of informal information .11 (4.40) .18 (8.32)
abroad in 1991
Acquisition of formal information .08 (3.32) .17 (7.91)
abroad in 1991
On ‘collecting informal information at home’ in 1995 from
Export performance 1993 -.01 (-.44) .09 (3.12)
On ‘collecting formal information at home’ in 1995 from
Export performance 1993 .02 (.89) .07 (2.35)
On ‘collecting informal information abroad’ in 1995 from
On ‘collecting formal information abroad’ in 1995 from
Acquisition of informal information .02 (.62) .03 (1.35)
at home in 1993
Acquisition of formal information at .03 (1.09) .05 (2.38)
home in 1993
196
Acquisition of informal information .18 (6.82) .24 (10.53)

abroad in 1993
Acquisition of formal information .18 (6.69) .21 (8.93)
abroad in 1993
Next, the models 7 and 8 with export ratio and export sales as dependent variables ,
respectively, and using export market research, - assistance, and – intelligence
information categories are tested (see Table 6-9).
Table 6-9 Results Longitudinal Model; Categories Intelligence - Research -

Assistance Information.
χ2 = 1233.29 (.00, df = 357) χ2 = 1322.13 (.00, df = 357)
RMSEA = .048 (.87) RMSEA = .05 (.57)
GFI = .94 GFI = .93
CFI = .91 CFI = .92
Size and significance of effects within one year (hypotheses 1,2 & 3)
On ‘Acquisition of intelligence information at home’ from
1991 1993 1995 1991 1993 1995
(.08) (-1.93) (-2.93) (.33) (-1.23) (-1.77)
(2.54) (4.99) (1.81) (2.43) (4.62) (1.26)
On ‘Acquisition of research information at home’ from
(-.61) (-.24) (-3.61) (-.26) (.61) (-2.53)
(2.45) (3.86) (2.85) (2.23) (3.54) (2.22)
On ‘Acquisition of assistance information at home’ from
(-2.36) (-.76) (-2.29) (-2.05) (.03) (-1.72)
(2.76) (3.44) (3.19) (2.54) (3.12) (2.82)
On ‘Acquisition of intelligence information abroad’ from
(-3.33) (-3.22) (-3.06) (-3.45) (-1.16) (-1.37)
Risk towards planning .14 .12 -.05 .15 .08 -.08
(2.32) (2.35) (-.94) (2.52) (1.81) (-1.57)
On ‘Acquisition of research information abroad’
(-3.07) (-2.13) (-1.88) (-3.03) (.14) (-.97)
Risk towards planning .02 -.04 .01 .03 -.07 -.01
(.41) (-.74) (.10) (.44) (-1.55) (-.20)
On ‘Acquisition of assistance information abroad’
(-3.21) (-2.72) (-1.74) (-3.06) (-1.32) (-1.00)
197
Risk towards planning .03 .05 .00 .03 .03 -.01

(.50) (1.00) (.06) (.46) (.54) (-.12)
(-.40) (-1.68) (-.98) (-5.07) (-3.83) (-3.22)
Risk towards planning -.02 -.02 -.12 .05 -.02 -.05
(-.29) (-.35) (-2.24) (.95) (-.42) (-2.07)
Acquisition of intelligence information at home .05 -.03 .04 .06 .02 .08
(1.85) (-1.13) (1.59) (2.39) (.80) (3.99)
Acquisition of research information at home .14 .08 .00 .11 .04 .03
(4.94) (2.93) (-.15) (4.32) (1.59) (1.62)
Acquisition of assistance information at home .18 -.01 .00 .14 .00 .03
(6.65) (-.36) (.18) (5.21) (-.15) (1.40)
Acquisition of intelligence information abroad .14 .20 .13 .23 .26 .20
(4.80) (7.29) (5.21) (8.71) (11.23) (9.82)
Acquisition of research information abroad .10 .18 .06 .19 .22 .15
(2.96) (6.61) (2.46) (7.25) (9.30) (7.48)
Acquisition of assistance information abroad .04 .13 .06 .11 .15 .11
(1.30) (4.87) (2.32) (4.03) (6.37) (5.36)
Size and significance of longitudinal effects (hypotheses 4 & 5)
On ‘Acquisition of intelligence information at home’ in 1993 from
On ‘Acquisition of research information at home’ in 1993 from
On ‘Acquisition of assistance information at home’ in 1993 from
On ‘Acquisition of intelligence information abroad’ in 1993 from
On ‘Acquisition of research information abroad’ in 1993 from
On ‘Acquisition of assistance information abroad’ in 1993 from
Acquisition intelligence information at home 1991 .01 (.41) .06 (2.59)
Acquisition research information at home 1991 .07 (3.00) .06 (2.44)
Acquisition assistance information at home 1991 .02 (.79) .05 (2.74)
Acquisition intelligence information abroad 1991 .12 (4.72) .19 (8.71)
Acquisition research information abroad 1991 .10 (4.11) .18 (4.71)
Acquisition assistance information abroad 1991 .03 (1.32) .10 (8.48)
On ‘Acquisition of intelligence information at home’ in 1995 from
On ‘Acquisition of research information at home’ in 1995 from
On ‘Acquisition of assistance information at home’ in 1995 from
On ‘Acquisition of intelligence information abroad’ in 1995 from
On ‘Acquisition of research information abroad’ in 1995 from
On ‘Acquisition of assistance information abroad’ in 1995 from
198
Acquisition intelligence information at home 1993 .02 (.61) .03 (1.32)

Acquisition research information at home in 1993 .04 (1.55) .06 (2.60)
Acquisition assistance information at home 1993 .00 (.04) .03 (1.27)
Acquisition intelligence information abroad 1993 .18 (6.75) .23 (10.42)
Acquisition of research information abroad 1993 .15 (5.75) .19 (8.34)
Acquisition of assistance information abroad 1993 .15 (5.76) .16 (7.01)
When examining the four longitudinal information behavior models, the first thing
to notice is that the fit of these models is lower than the one-year static models (see
sub-section 6.3.4)156. Nevertheless, the GFI and CFI are good for all four models,
and the RMSEA is around .05 for all models (and all imputed data sets). Only the
2
is significant and shows a bad fit. Here, we have to keep in mind the sensitivity
of this statistic for large samples, as is the case here. Overall, the results are
satisfactory, and will be examined for patterns, which can lead to acceptance or
rejection of the hypotheses.
Considering the nine static hypotheses (compare sub-section 6.3.4), a few things
stand out. First, the effect of information on export performance in the three years
is examined (hypothesis 1). Again, foreign informal (intelligence) and formal
(research and to a lesser extent assistance) positively affect both export ratio and
export sales consistently over the three years, again emphasizing the importance of
this information. Looking at the information collected on the home market, there
are some significant associations, but most are non-significant. The results show
that domestic informal (intelligence) information only impacts export sales, and not
export ratio, and only in 1991 and 1995 (the other two imputed sets also show
significant results for 1993, and in two years for export ratio, but as these only
come forward in one of the three imputed data sets, these are not stable). Domestic
formal information impacts export performance positively in some cases: export
ratio in 1991 and 1993, and export sales in 1991. When breaking this type of
information down into research and assistance it appears that the impact of export
assistance information at home on export ratio is negligible in 1993, although the
other two sets do show a positive significant relationship. Of course, the fact that
export performance in the later two years is explained by lagged independent
156
Although the program suggests modification indices, these have no theoretical
justification, and are, therefore, not included.
199
variables as well, might explain this leveling out of the impact of formal
information at home over the years. Again, hypothesis 1 is accepted partly.
Secondly, the results are inspected for evidence on hypothesis 2. The difference in
the direction of the sign between the relationships leading from ‘risk towards
change’ and ‘risk towards planning’ to the information collection still holds. That
is, in general the more risk averse a manager is towards changes, the less
information he/she collects, while a stronger preference for order and planning
induces a manager to collect more information. For all three years, a consistently
strong positive link exists between ‘risk towards planning’ and the collection of
informal (intelligence) information at home and abroad (with the exception of
1995, where the link between this attitudinal variable and the collection of informal
information abroad is non-significant), for both model 5 and 6. For the impact on
formal information, only the information collected at home is positively associated
with the tendency towards planning. This goes for both the research and the
assistance information acquired at home. Investigating the relationships between
‘risk towards change’ and the information collection, there is a consistent negative
association with the foreign formal (both research and assistance) and informal
information (consistent over the three data sets for model 5 and 7; for model 6 and
7 only in some of the imputed data sets significant negative parameters are found).
Again, the inclusion of lagged variables in 1993 and 1995 could explain this partly.
At home, the significant relationships between change attitude and information are
sparse and follow no specific pattern, although there is some indication for a
negative association. So, hypothesis 2 is accepted only partially, and only for the
risk attitude towards changes.
The last static proposition is on the assumed negative relationship between the risk
avoiding propensity and export performance (hypothesis 3). The results make clear
that this only accounts for the consistently negative and significant impact of ‘risk
towards changes’ on export sales. For export ratio, the signs are negative but not
significant (exception is 1993). The second and third imputed data set does show
some negative associations with export ratio, but these are non-consistent over the
three sets. Although ‘risk towards planning’ positively influences information
collection, the relationship with export performance is non-significant over the
200
three data sets. Again, hypothesis 3 is accepted partly, but only for ‘risk towards
changes’.
Moving towards the longitudinal hypotheses (hypotheses 4 and 5), the following
picture emerges. The information collected in 1991 does affect export performance
in 1993 positively, but this is especially true for export sales, and for information
collected abroad (almost twice as much as domestic information). In the other two
imputed data sets, the positive impact of domestic intelligence information 1991 on
export performance 1993 is even less, and sometimes non-significant. For export
ratio in 1993, it appears that the domestic informal (intelligence) and the assistance
information collected (both at home and abroad) in 1991 do not have any
significant relationship with the export intensity. The linkages between information
in 1993 and export performance 1995 are almost similar, although only the
domestically acquired formal information in 1993 improves export sales in 1995
(and, more specifically, only the research component of formal information). Yet,
the other two imputed data sets do show rather strong support for the positive
impact of the domestic information collected in 1993 on export ratio in 1995 (but
not export sales). Thus, hypothesis 4 is accepted as well, with the emphasis on the
positive effect of foreign information for both export ratio and export sales.
Lastly, the information collected in 1993 is consistently impacted positively by the

level of export performance (hypothesis 5). Yet, there are some differences when
using export ratio or export sales as the export performance variable. A higher
export sales in 1991 especially boosts the collection of information abroad in 1993
(almost twice as much than the information collected at home), while high export
ratio affects the information both at home and abroad almost equally. In addition,
the linkages between export sales and information collected abroad are much
stronger than those with export ratio. In 1995, almost the same pattern can be
found, although the impact of performance on information acquired at home is
mostly non-significant. Again, export sales boosts the use of foreign information
providers much more than export ratio. Although some difference can be found in
the other two imputed data sets, these are rather at random, and are not taken into
consideration. So, the last hypothesis is accepted as well.
201
6.3.6 Conclusions
Table 6-10 summarizes the proof found for the five hypotheses in the eight models,
with the peculiarities found in these results.
Table 6-10 Summary Results on Hypotheses.
Hypotheses Models 1-4 (static) Models 5-6 (longitudinal)
H1: export information Accepted Especially foreign Idem
affects export performance information sources, informal
positively stronger than formal, stronger
impact on export sales than export
ratio
H2: risk avoiding attitude Distinction between 2 components: Idem
affects export information 1 Risk attitude towards changes:
negatively negative => Accepted
2 Risk attitude towards planning:
positive => Rejected
H3: risk avoiding affects Accepted partly: Idem
export performance Only ‘risk towards changes’
negatively affects performance negatively
H4: lagged information Not included Accepted:
affects export performance Especially information acquired
positively abroad and formal information
acquired at home
H5: lagged performance Not included Accepted:
affects export information Stronger relationships for export
positively sales with foreign information
The results clearly show that all hypotheses at least receive partial support. Results
that stand out are the deviating impact of the attitude of managers towards changes
and their attitude towards the use of planning and proven procedures: The first
makes owner-managers collect less information and deteriorates the export
performance directly as well, while the second induces owner-managers to use
more information sources, although this attitude does not impact performance
directly, but only through the information collected. Secondly, the positive impact
of information collected abroad is much stronger than that stemming from
domestic information. Besides, informal information seems to boost performance
stronger than formal information. On the other hand, the formal information has a
stronger impact on the performance measure in the following year. Thirdly,
although the success gained in the previous year does heighten the number of
information sources collected in the subsequent year, the link is especially strong
between export sales and the amount of foreign information.
202
Chapter 7
Conclusions and Discussion
7.1 Introduction
The ever-increasing globalization of the business world, along with the importance
of small- and medium-sized enterprises, drives many researchers to seek for ways
how to improve the international performance of these firms. The objective of this
thesis is to examine the determinants of export performance of European SMEs. In
addition to determinants that directly influence the international performance,
indirect influences are essential, and by simultaneously modeling these, total
effects on export performance are found. An integral export performance
framework is build that encompasses all determining factors and relationships. To
test this model empirically, an existing data set is used (from the INTERSTRATOS
research project). The estimation of the theoretical model enables me to make
inferences on the significance, size and direction of all the effects found in
literature. The detection of these influential factors helps SMEs managers to
become aware of possibilities to improve their export performance.
Given the increasing attention for the internationalization of organizations, I start

with an overview of the conceptual and empirical literature on export performance
(Chapter 2). Chapter 3 introduces the INTERSTRATOS project and the panel data set,
which I use to test the integral model found in Chapter 2 empirically. Using this
data set and the variables herein, the theoretical model is operationalised in Chapter
4. Besides, the multiple imputation technique NORM (Schafer 1997) is explained
and applied to the panel data set, considering the nature of the data. Chapter 5 is the
empirical study on export performance. In Chapter 5, I use the resulting imputed
1995-data sets to empirically test a one-year integral model as operationalised in
Chapter 4. Focusing on a sup-part of this integral model, the relationships between
203
managerial characteristics, information behavior and export performance are

empirically analyzed in both a static and a longitudinal manner, in Chapter 6.
In this chapter, I provide a brief overview of this thesis and discuss prospects for
future research. The next section summarizes the findings of the literature review
(Chapter 2), the operationalization of the theoretical model (Chapter 4), and the
empirical tests of (part of) this integral model (Chapter 5 and 6). Section 7.3
provides the reader with the most important findings resulting from this thesis,
while the last section (7.4) identifies the limitations of this research and some
topics for future research.
7.2 Summary
In this section, I briefly summarize the content of the chapters in this thesis.
Chapter 2 Towards an Integral Framework

The theoretical framework (Chapter 2) proved to be a good starting point for
answering the central question of this dissertation, which pertains to finding all
factors that explain the level of export performance of an SME either directly or
indirectly. For this, nine conceptual studies on export performance are examined,
followed by a review of 43 empirical studies that research export performance and
the determinants of this output variable. The conceptual studies enabled me to
classify determinants into several categories. These categories pertain to the
environment, the firm, the manager, and the export activities. As the focus of this
thesis is on SMEs, the manager is an important category. Therefore, I distinguish
between objective and subjective managerial characteristics. The nature of these
characteristics is inherently different from the characteristics of the management
team, used by most studies. In SMEs, often the manager him- or herself takes the
decisions, instead of a (more rational) management team. Besides, the literature
synthesis of the empirical studies leads to an examination of the measurement of
export performance. That is, (1) the most commonly used measures are identified
(i.e., export sales and export ratio), and (2) the development towards including
multi-dimensional measures, and subjective measures next to objective measures.
Finally, the review synthesizes into an integral export performance model,
containing not only all possible variables in the categories found in both the
204
conceptual and empirical studies, but all possible relationships between these
variables as well. This integral model served as the starting-point for the empirical
work done in this thesis.
Chapter 4 Operationalizing the integral export performance model

Using the INTERSTRATOS panel data set containing the years 1991, 1993, and 1995
(see Chapter 3), the goal of Chapter 4 is to operationalize the conceptual integral
export performance model as found in Chapter 2. Thus, the determinants
researched in the literature are confronted with the variables in the INTERSTRATOS
panel data set. This data set is an existing data set, bringing along all possible
problems researchers can have with secondary data. First, only a sub-set of all
possible indicators is covered by the panel data set. Secondly, the specific variables
questioned in the survey do not always fit the variables on a one-on-one basis.
Therefore, I had to make some transformations to proxy specific determinants.
Third, the operationalized variables in the panel data set contain missing values,
due to item non-response. This loss of information is highest in the performance
variables. To increase the amount of information in the data set, a multiple
imputation approach is applied, i.e. NORM (Schafer 1997). Using a Bayesian
approach (i.e. using the prior distribution of the original variables), NORM
simulates the missing part three times, creating three equally plausible versions of
the complete data for both the 1995-data used in Chapter 5, and the 1991-1993-
1995 panel data set used in Chapter 6. In both chapters, one of the complete data
sets is the input for model building and testing, while the other two data sets serve
to validate the robustness of the parameters found in the first imputed data set.
Chapter 5 Estimating the Integral Model

I empirically test the operationalized integral export performance model in Chapter
5 with one of the three imputed 1995 data sets, using the structural equations
modeling technique LISREL. Without including any dynamic relationships, this
chapter focuses on the static relationships between the various determinants and
export performance. In this, export performance is measured by the two available
performance measures in the INTERSTRATOS survey, i.e. ‘export ratio’ (export sales
divided by total sales), and ‘export sales’. Conform the discussion started by
Diamantopoulos (1999); these export performance measures are not considered as
reflective variables, but as two separate (and sometimes conflicting) aspects of
205
export performance. Thus, the model is estimated once with export sales as the
dependent variable, and once with export ratio as the dependent variable. First, the
various constructs are validated. This is the first step in the two-step procedure
(Anderson & Gerbing 1988). The second step consists of building two MIMIC
models to test the integral framework, fixing the reflective measurement models in
the first step. In this empirical application, the structural models show mediocre to
good fit. I find that export performance is specifically affected by environmental
factors, such as changes on the supply and (national and international) sales
market, by firm characteristics, such as the firm size, capabilities and experience
gained through exporting, and by export activities, such as the collection of foreign
(formal and informal) information, and the cooperation with foreign partners.
Differences between the influential factors for ‘export sales’ and ‘export ratio’ are
sparsely found (especially when taking into account all three imputed data sets).
Chapter 6 The Information Behavior of Exporting SMEs

As the one-year integral model proved, the information behavior of SMEs is an
important factor in improving export performance. Recent publications on the
international market orientation emphasize the collection, dispersion, and use of
international information as well (e.g., Cadogan & Diamantopoulos 1995). Besides,
in both the SME literature (e.g., Kuratko & Hodgetts 2001), and the international
marketing orientation literature (e.g., Avlonitis & Gounaris 1999), the role of the
owner-manager is imperative. Therefore, Chapter 6 specifically focuses on the
static and longitudinal relationships between the risk attitude of the manager, the
information behavior of the SME, and the export performance. Concerning
information behavior, two categorizations are used (i.e. formal versus informal
information collection, and export market research versus export assistance versus
export intelligence information collection). The respondents use only a small part
of all possible information providers, with an emphasis on formal information
sources on the domestic market and on intelligence sources in the foreign
environment. In addition, a large percentage of all respondents indicated not to
consult any information provider at all. The risk attitude is twofold (cf. Chapter 4):
(1) The attitude of the owner-manager towards changes, and (2) the attitude of the
owner-manager towards planning and proven procedures. Again, a two-step
approach leading to a MIMIC model, using LISREL, is applied to estimate the
static and longitudinal relationships between the risk attitude, information
206
collection (for both categorizations), and export performance (for both ‘export
ratio’ and ‘export sales’). All models show good fit,. The results show that (1)
export information does impact export performance positively (especially foreign
information sources), (2) the risk attitude does impact the information behavior, but
(3) export performance only partly, (4) the information collected in the previous
year has a lagged effect on export performance in the subsequent year, and (5) the
export performance in the previous year positively affects the amount of
information sought in the year hereafter.
7.3 Important Findings

The most important findings of this thesis are summarized below:
1) Notwithstanding the ever-increasing body of empirical evidence on export

performance, there is still no agreement on the measurement of export
performance. Most authors agree that the concept is multi-dimensional, but
should not be treated as a reflective variable, as some aspects of export
performance could be conflicting. Besides, there exists a move towards
including subjective measures and dynamic measures. The INTERSTRATOS
data set shows an especially large amount of missing values in the
objective performance questions, reinforcing the call for subjective
measures (Chapter 2, 3, 4, and 5).
2) The empirical evidence on export performance determinants is still very
dispersed, with a lack of agreement on the operationalization of the
determinants. Consequently, a synthesis of findings is impossible or very
difficult, as different operationalizations lead to different effects (Chapter
2).
3) There is no consensus on the direction of the relationships between the
determinants. This is partly due to the overemphasis on cross-sectional
research, as this is easier and faster than longitudinal research (Chapter 2).
4) Although over time researchers on export performance started to include
indirect effects, there is still too little evidence on indirect relationships
leading to export performance (Chapter 2).
207
5) In longitudinal research, the largest problem concerns the survey non-

response over the years, which makes it very complicated to draw up a
panel data set (Chapter 3).
6) Using a secondary data set for a theoretical model the data set was not
originally collected for, leads to problems when trying to operationalize the
theoretical model (Chapter 4).
7) Using a simple imputation method may lead to erroneous results, as these
procedures do not take into account the uncertainties surrounding the
correct value of the missing variable. Multiple-imputation techniques may
(partly) solve this problem by providing the researcher with various data
sets. By using one data set to build and estimate the model, and the other
data sets to re-estimate this model, the researcher can compare the
parameters from all models and examine patterns. The more scattered the
parameters, the less certain a researcher can be with regard to the true sign
and strength of the relationship (Chapter 4, 5, and 6).
8) The most important influential activities in improving export performance
(either export sales or export ratio) are the collection of foreign
information and the co-operation for the international activities. In
collecting information from information providers abroad, the informal
information (through day-to-day business) is even more beneficial than
formal information (systematically gathered research or assistance
information). Next, an exporting SME should specifically look for
cooperation with foreign partners (Chapter 5).
9) Other significant determinants in the static and integral export performance
models are much less under the control of the firm, i.e. changes in the
environment or firm characteristics. In case of a deteriorating supply or
national sales market, or an improving international sales market, export
sales and export ratio are enhanced. With regard to firm characteristics, a
larger and more export-experienced firm has better output (Chapter 5).
10) There are few differences between the influential factors for ‘export ratio’
and ‘export sales’. Although different aspects of export performance might
have conflicting relationships with determinants, both ‘export sales’ and
‘export ratio’ can be labeled as ‘economic’ (cf. Katsikeas, Leonidou &
Morgan 2000), or ‘sales-related’ (cf. Shoham 1996; 1998; 2000) aspects,
which might explain this similarity. The only evident deviation is the much
208
stronger positive effect of ‘firm size’ (in terms of number of employees) on

export sales versus export ratio (almost ten times as strong). Thus, larger
SMEs might accomplish much higher export sales, while the export ratio
differences between smaller and largest SMEs are smaller (Chapter 2, and
5).
11) The risk attitude of the owner-manager is not a one-dimensional concept,
but has many aspects. In this thesis, two dimensions have been
operationalized, i.e. the way owner-managers feel towards changes and
their feelings towards the use of planning and proven procedures.
Consequently, not all aspects relate in the same manner to information
behavior or export performance. That is, the more risk avoiding the owner-
manager is towards changes, the lower the performance, but the second
risk aspect has no direct impact on performance. With regards to the
information collection, the two risk aspects show opposite results: The
more change avoiding the less information is requested, while owner-
managers that like planning use more informal information to reduce risk
(Chapter 4, and 6).
12) There is a lagged positive effect from information behavior on export
performance. All information acquired abroad, and the formal information
acquired at home have a lagged positive effect on both export ratio and
export sales. So, the information collected is also useful on the long term
(Chapter 6).
13) The export performance in the previous year positively impacts the
information collected in the subsequent year. Especially the amount of
foreign information is raised. A possible explanation is the fact that
(foreign) information is often costly, and a better result in the previous year
may result in more resources for information. A second explanation could
be that higher export sales or ratio might increase the need for information
on the foreign markets, no matter the costs. After all, a higher proportion of
total sales stems from exporting, increasing the importance of the
international market for the SME (Chapter 6).
14) Overall, I advise an SME-manager to improve export performance by
collecting information, especially from foreign information providers, and
to cooperate with foreign partners. By doing this, the SME can gain higher
export sales and/or export ratio. Besides, an SME-manager should be
209
aware of the effect of his or her risk attitude on his or her behavior and on
the performance of a company.
7.4 Limitations and Future Research

Every research has limitations and invites the researcher to consider new directions
for future research. In that respect, the research described in this dissertation is not
different from any other research. Conceptual topics that need further investigation
are:
1) Measurement of export performance;
2) Establishing a comprehensive list of export performance determinants, and
the operationalization of these;
3) Use of conceptual models previous to the collection of data;
4) Place of the owner-manager in export performance modeling SMEs;
5) Further study of international market orientation;
6) Direction of relationships;
Technical topics that call for further examination are:
7) Set up of international research projects;
8) Set up of longitudinal research projects;
9) (Multiple) imputation of missing values.
Ad 1) Measurement of export performance.

Notwithstanding the ever-increasing attention for export performance, and the
measurement of this concept, the empirical applications still deviate largely on the
operationalization of this measure. Researchers should come to an agreement on
international performance, including a better understanding of what owner-
managers themselves consider important outcomes. The inclusion of subjective
measures can also be beneficial for decreasing the item non-response on the more
objective performance measures. On the other hand, when all aspects of export
performance are to be included, multiple imputation techniques such as NORM
(Schafer 1997) could help out.
210
Ad 2) Comprehensive list of export performance determinants &

operationalization.
The attention for export performance mentioned above is accompanied by the wish
to explain export performance and to find the determinants influencing export
performance. With this, both managerial and governmental implications can be
made. Over time, the empirical results contribute to an ever more comprehensive
list of possible determinants, but this list is still not conclusive, although Chapter 2
is good step in the right direction. Besides, for some variables or relationships the
evidence is quite scattered so that no synthesized directions can be given.
Therefore, more integral studies should be undertaken. On the other hand, integral
models are very complicated to estimate simultaneously, as many possible direct
and indirect effects can be assumed. Therefore, a balance should be found between
the wish to be integral and the problem of estimation (often these models lead to
few significant results). Also, when a comprehensive list of factors can be made
(even if a non-comprehensive list is used), researchers should come to an
agreement on how to operationalize the variables on the list. Only then, the
empirical results can be synthesized in a meaningful way.
Ad 3) Use of conceptual models previous to the collection of data.

I strongly emphasize the need for conceptual models before starting the data
collection. Preferably, the conceptual model synthesized in Chapter 2 should be the
basis for a survey collecting the data on the specific variables. In this thesis, the
data set used was already collected for other purposes. By first building a
conceptual model, the researcher focuses him- or herself and is able to enlarge the
academic knowledge on the specific topic. The data set will be more consistent,
and the amount of item non-response will probably drop as well (see also Leeflang
et al. 2000).
Ad 4) Place of the owner-manager in export performance modeling SMEs.

The special place of the owner-manager in SMEs should be emphasized in export
performance research with SME-respondents. This not only implies explicit
modeling of both the objective and subjective characteristics of the owner-
manager, but also incorporating subjective measures of export performance, as
these are especially imperative in SMEs where a single person is in charge of the
company.
211
Ad 5) International market orientation.

In this study, and in most empirical studies, only one aspect of information
behavior is examined, namely the collection of information. Yet, the increasing
attention for market orientation, and, more specifically, international market
orientation calls for a thorough investigation of all aspects of information behavior
leading to a better market orientation, and, as hypothesized, a better export
performance. These three aspects are the information collection, information
dissemination, and information use (cf. Kohli & Jaworski 1990, Cadogan &
Diamantopoulos 1995). By only focusing on information collection, researchers
forego the subsequent use of information. After all, if information is collected but
remains with only one person, or is stored and never used, this information has not
much value. In this thesis, the amount of information collected is used. There
should also be attention for the quality and relevancy of information (see also
Glazer, Steckel & Winer 1992).
Ad 6) Direction of relationships.
To be able to examine the direction of various relationships (i.e. does the
determinant impact export performance, is it the other way around, or both),
longitudinal research projects should be started. Also, lagged effects of influential
factors can be determined, conform the dynamic relationships found between
information behavior and export performance in Chapter 6.
Ad 7) International research projects.

In this thesis, the basis for the empirical applications is the international research
project INTERSTRATOS. Chapter 3 elaborates on the background of this project. The
collection of data in various countries brings along various problems, leading to the
need for a strong coordinating mechanism. Members should agree on the treatment
of survey non-response, item non-response, and data registration. Besides,
definitions of variables should be equivalent over all countries. A second problem
does not specifically touch upon the international character of a project, but upon
the fact that various researchers are brought together to start a joint research
project. This group of researchers should have a well-defined, and confined goal
with the research. In the case of INTERSTRATOS, the goal was too dispersed, and
individual interests prevail. The common goal should be placed in a theoretical
212
framework and this conceptual model should be the base for the data collection.
Then, the problems of operationalization can be minimized.
Ad 8) Longitudinal research projects.

The INTERSTRATOS project is a longitudinal research project. The most important
issue to consider in building a panel data set is the mortality of respondents. Only if
respondents participate all years, the panel data set has potential. Although multiple
imputations might be a solution when respondents do not participate in all years,
this means that all information of the specific year is lacking and has to be imputed,
thereby decreasing the usefulness of the data set. Only if the imputation technique
specifically allows for longitudinal results from one respondent, this has potential.
Ad 9) Multiple imputation.
As this thesis shows, the rather unknown technique of multiply imputing data sets
with missing values enables researchers to enrich their data sets without relying on
chance when filling out missing values. This technique results in two or more
imputed data sets, of which one data set should be used for building and estimating
the model, while the other data sets should be used to validate the parameters found
in this model and to visualize the uncertainty surrounding this parameters due to
the amount of missing data. In contrast, researchers that impute missing values
with single imputation methods, use simple or more complex methods to find the
most logical value, but this value is never without uncertainty. The model is build
and estimated with only this one imputed data set, and the results of only this
model are presented. As Chapter 5 and 6 show, this can lead to erroneous
conclusions concerning significant relationships. Therefore, I urge researchers to
take this imputation technique into account, which might seem more complex, but
leads to more valid results and conclusions.
213
214
Appendix 1
The INTERSTRATOS Survey
DESCRIPTION CATEGORIES
CHARACTERISTICS OF THE ENTERPRISE
Country 100 Austria
104 Belgium
108 Netherlands
130 Switzerland & Liechtenstein
135 Norway
136 Sweden
140 Finland
Enterprise identification number Open
Branch number 1 Textiles en clothing
2 Electronics
3 Food
4 Furniture making
5 Mech. engineering &
manufacture metal products
How old are you? Open
How many full time employees do you have? Open
How many of these employees work abroad? Open
SUCCESS FACTORS
Quality of management 0 no answer
Reputation, local image and personal contacts 1 no importance
Workers’ skills 2 low importance
Technology 3 medium importance
Ability to solve technical problems 4 high importance
Product quality 5 very high importance
After sales service
Creativity
Distribution and selling staff
Product design and brand image
Pricing policy (lowest possible price)
Low cost level
Market share
Financial strength
Customer relations (public relations)
Flexibility (being able to respond quickly to customer or
changing circumstances)
Reliability on delivery
Administration
Ability to modify products
215
Most important of these factors 01…19

INFORMATION, CONSULTATION, TRAINING AND FURTHER EDUCATION
What external sources of information, if any, have you consulted about your activities at home
or abroad within the last twelve months?
In your country 0 no answer
Further education/ training institutions 1 yes
Consultants 2 no
Credit insurance agencies
Suppliers
Customers
Export clubs
Organizations of the camber of commerce
Research institutions or agencies
Public promotion institutions (local governments, federal
governments etc.)
International organizations (EC, UNIDO)
In foreign countries 0 no answer
Further education/ training institutions 1 yes
Consultants 2 no
Credit insurance agencies
Suppliers
Customers
Export clubs
Organizations of the camber of commerce
Research institutions or agencies
Public promotion institutions (local governments, federal
governments etc.)
International organizations (EC, UNIDO)
SPECIALISATION, PRODUCT AND MARKET STRATEGIES
Turnover net (in mio, excl. VAT) Open
Of this, export (in mio)
CHANGES IN DIFFERENT MARKETS
How do you see recent changes (last 12 months) from your firm’s viewpoint?
216
Appendix 1
On the labor market 0 no answer

Availability of qualified employees (managers and skilled 1 very positive
workers) 2 positive
Availability of unskilled workers 3 neutral
Availability of apprentices or youth employment schemes 4 negative
On the supply market 5 very negative
For raw materials, part and components
For products bought for resale
For technologies(machines, licenses)
On the sales market
Local market
National market
International markets
On the capital/credit market
Availability of finance
Availability and quality of financial advice
Interest and other capital costs
FOREIGN MARKET ORIENTATION AND INTERNATIONAL OPERATIONS (LAST 12
MONTHS)
Did you buy any foreign goods from a local supplier? 0 no answer
Did you buy any foreign goods from a supplier abroad? 1 yes
2 no
Which proportion of your supplies came from abroad? 0-100%
Did you make or sell anything under license from abroad? 0 no answer
Did you sell through an agent in your country? 1 yes
Did you sell as supplier to an exporting producer? 2 no
Did you sell to a local firm which eventually exports them?
Did you sell through agents abroad?
Did you sell directly to customers abroad?
Did you license any of your products abroad?
Did you have any subsidiaries abroad for distribution?
Did you have any subsidiaries abroad for manufacture?
MONTHS)
Please list the countries abroad from which you drew supplies Special coding
Please list the countries abroad to which you sold your products.
MONTHS)
Please tick any cases below in which you have cooperated with other firms in any of the
activities listed below
217
With domestic partners in foreign countries 0 no answer

Extension of product range 1 yes
Research and development 2 no
Raising funds
Sales
Market research
After sales service
Advertising/promotion
Purchase/supply
Transport/warehousing
Manufacture
Administration
Electronic data processing
With foreign partners in foreign countries 0 no answer
Extension of product range 1 yes
Research and development 2 no
Raising funds
Sales
Market research
After sales service
Advertising/promotion
Purchase/supply
Transport/warehousing
Manufacture
Administration
Electronic data processing
MONTHS)
Did you search actively for orders abroad or did you just fill any orders which you received?
Actively 0 no answer
Passively 1 yes
2 no
Which of the factors have encouraged or discouraged your present level of activities abroad?
218
Appendix 1
Workers’ skills ( domestic) 0 no answer

Technology (domestic) 1 encouraged
Price levels (domestic) 2 not relevant
Cost levels (domestic) 3 discouraged
Customer relations
Supplier relations
Larger market
Full use of capacity (domestic)
Spreading risks
Subsidies (domestic)
Tax on profits (domestic)
Raw material supplies
Political stability (domestic)
Favourable entrepreneurial climate (domestic)
Product range
Export know how
Finance
Frontier formalities
Prestige
Growth
Acquisition of information (domestic)
Following the market leader
PERSONAL CHARACTERSTICS OF ENTREPRENEURS
At what age did you leave full time education (including Open
apprenticeship, university etc.)?
How many years of experience do you have in your industry?
How many weeks have you stayed abroad during the last 3 years?
In how many languages (including your mother language) can you
or your senior managers carry on business?
What is your opinion about the following statements
219
The government should not restrict competition even through the 0 no answer
use of incentives. 1 I strongly disagree
Professional bodies and similar organizations should only provide 2 I disagree
assistance to their members 3 no opinion
Changes in a business should be avoided at all costs. 4 I agree
A firm should not leave the region where it is established 5 I strongly agree
Jobs should be clearly described and defined in detail
Managers should plan rather than follow their intuition
Firms should only introduce proven office procedures and
production techniques
In family owned businesses the management should stay in the
hands of the family
Small firms should not hesitate to do business with large firms
Small business managers should take personal responsibility for the
recruitment of all employees
A manager should consider ethical principles in his behaviour
Business should take precedence over family life
Total number of years of education (including primary) open
220
Appendix 2
Results Integral Export
Performance Model on
Imputed Data Set 2
Table A2-1 Validity reflective models – set 2.
Measurement Items Latent Goodness-of-Fit Statistics
Construct (n = 1122)
1. CFA Environment χ2SB = 8.51 (p = .74, df = 2)
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 1.00)
1a. Labour α = 0.53 CR = 0.57
(2 items, Labour) VEE = 0.41
λav = 0.63
λlow = 0.47
1b. Supply α = 0.78 CR = 0.80
λav = 0.82
λlow = 0.76
1c. National Sales NA NA
(1 item, Natsal)
1d. International Sales NA NA
(1 item, Intsal)
1e. Capital α = 0.68 CR = 0.69
λav = 0.72
λlow = 0.71
GFI = 0.98
CFI = 0.97
RMSEA = 0.056 (p = 0.22)
λav = 0.76
λlow = 0.69
221

Measurement Items Latent Goodness-of-Fit Statistics
Construct (n = 1122)
λav = 0.80
λlow = 0.75
λav = 0.70
λlow = 0.68
λav = 0.70
λlow = 0.69
4. CFA Attitude χ2SB = .66 (p = 0.42, df = 2)
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.81)
4a. Risk attitude towards α = 0.66 CR = 0.67
changes VEE = 0.51
(2 items, RTC) λav = 0.71
λlow = 0.63
4b. Risk attitude towards α = 0.56 CR = 0.59
planning VEE = 0.43
(2 items, RTP) λav = 0.64
λlow = 0.49
5. CFA Internal Stimuli χ2SB = 21.41 (p = 0.09, df = 2)
GFI = .99
CFI = 1.00
RMSEA = 0.022 (p = 1.00)
5a. Firm Proactive α = 0.80 CR = 0.81
Technology VEE = 0.67
(2 items, Ftechn) λav = 0.82
λlow = 0.81
5b. Firm Proactive Financial α = 0.86 CR = 0.86
(2 items, Ffinanc) VEE = 0.76
λav = 0.87
λlow = 0.79
222
Appendix 2
Table A2-2 Results structural models integral export performance – set 2.

Model A (export ratio) Model B (export sales)
χ2 = 1810.07 (.00, df = 528) χ2 = 1799.01 (.00, df = 528)
RMSEA = .047 (.99) RMSEA = .049 (.99)
GFI = .92 GFI = .92
CFI = .88 CFI = .89
On ‘risk attitude towards change’ from
Age .14 (4.22) .14 (4.22)
Education -.08 (-2.38) -.08 (-2.38)
Firm size -.17 (-4.44) -.17 (-4.44)
Age .12 (3.48) .12 (3.48)
Education .03 (.91) .03 (.91)
Firm size -.07 (-1.72) -.07 (-1.72)
Supply market -.07 (-2.09) -.07 (-2.09)
Labour market .08 (2.31) .08 (2.31)
National sales market .03 (.91) .03 (.91)
International sales market .00 (-.14) .00 (-.14)
Capital market -.02 (-.52) -.02 (-.52)
Education .07 (.91) .07 (.91)
Firm size .10 (2.96) .10 (2.96)
Risk attitude towards change -.10 (-2.34) -.10 (-2.34)
Supply market -.10 (-2.86) -.10 (-2.86)
Labour market .12 (3.36) .12 (3.36)
Capital market -.05 (-1.32) -.05 (-1.32)
Education .02 (.81) .02 (.81)
Firm size .15 (4.29) .15 (4.29)
Export experience .00 (-.07) .00 (-.07)
Import experience -.03 (-.90) -.03 (-.90)
Linguistic skills -.06 (-1.76) -.06 (-1.76)
223

Supply market .00 (.11) .00 (.11)
Labour market .01 (.26) .01 (.26)
International sales market -.03 (-.99) -.03 (-.99)
Capital market -.07 (-1.91) -.07 (-1.91)
Education -.02 (-.82) -.02 (-.82)
Firm size .12 (3.92) .12 (3.92)
Linguistic skills .01 (.18) .01 (.18)
Supply market -.04 (-1.24) -.04 (-1.24)
Labour market .00 (.11) .00 (.11)
International sales market .02 (.61) .02 (.61)
Capital market .02 (.48) .02 (.48)
Education .00 (-.12) .00 (-.12)
Firm size .21 (6.47) .21 (6.47)
Supply market -.03 (-.75) -.03 (-.75)
Labour market .00 (-.12) .00 (-.12)
National sales market -.01 (-.17) -.01 (-.17)
Capital market .02 (.54) .02 (.54)
Education .00 (.12) .00 (.12)
Firm size -.09 (-2.49) -.09 (-2.49)
Import experience .00 (-.11) .00 (-.11)
Risk attitude towards change .04 (1.07) .04 (1.07)
Domestic informal information .10 (3.19) .10 (3.19)
Domestic formal information .12 (3.87) .12 (3.87)
Supply market -.08 (-2.49) -.08 (-2.49)
224
Appendix 2
Labour market -.08 (-2.55) -.08 (-2.55)

Capital market .05 (1.33) .05 (1.33)
Education .06 (2.29) .06 (2.29)
Firm size .03 (.85) .03 (.85)
Supply market .11 (3.17) .00 (.05)
Labour market .01 (.41) .00 (-.20)
Capital market .05 (1.37) .03 (1.33)
Firm size .04 (1.18) .56 (24.43)
Internal financial stimuli .02 (.53) -.02 (-.83)
Marketing capability -.06 (-1.47) .00 (.12)
Product superiority -.01 (-.22) -.04 (-1.37)
Education .07 (2.44) .04 (1.91)
Age .04 (1.55) .00 (.10)
Domestic informal information -.06 (-1.94) -.06 (-2.78)
Domestic formal information -.01 (-.17) -.03 (-1.31)
Domestic cooperation .00 (.14) -.01 (-.31)
225
Table A2-3 Total effects of variables on export performance – set 2.

Total (direct + indirect) effects of variable: On export ratio On export sales
Environment:
Supply market .13 (3.04) .00 (.08)
Labour market .01 (.20) -.03 (-.60)
Capital market .06 (1.23) .07 (1.24)
Firm size .09 (2.78) 1.04 (27.55)
Internal financial stimuli .02 (.53) -.04 (-.83)
Marketing capability -.07 (-1.47) .01 (.12)
Product superiority -.01 (-.22) -.10 (-1.37)
Linguistic Skills .17 (2.10) .30 (2.99)
Import experience -.04 (-1.84) -.01 (-.46)
Education .49 (2.66) .46 (2.02)
Age .15 (.97) -.10 (-.51)
Export Activities:
Domestic informal information -.05 (1.79) -.10 (-2.72)
Domestic formal information .00 (.18) -.03 (-1.10)
Domestic cooperation .00 (.14) -.01 (-.31)
226
Appendix 3
Results Integral Export
Performance Model on
Imputed Data Set 3
Items (n = 1122)
1. CFA Environment χ2SB = 13.92 (p = 0.31, df = 12)
GFI = 1.00
CFI = 1.00
RMSEA = 0.012 (p = 1.00)
1a. Labour α = 0.54 CR = 0.59
(2 items, Labour) VEE = 0.43
λav = 0.64
λlow = 0.48
1b. Supply α = 0.78 CR = 0.81
λav = 0.83
λlow = 0.75
1c. National Sales NA NA
(1 item, Natsal)
1d. International Sales NA NA
(1 item, Intsal)
1e. Capital α = 0.68 CR = 0.69
λav = 0.73
λlow = 0.71
GFI = 0.98
CFI = 0.97
RMSEA = 0.055 (p = 0.25)
λav = 0.76
λlow = 0.69
227

Items (n = 1122)
λav = 0.81
λlow = 0.76
λav = 0.71
λlow = 0.69
λav = 0.70
λlow = 0.69
4. CFA Attitude χ2SB = 1.19 (p = 0.27, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .013 (p = 0.72)
4a. Risk attitude towards α = 0.65 CR = 0.66
changes VEE = 0.50
(2 items, RTC) λav = 0.72
λlow = 0.67
4b. Risk attitude towards α = 0.56 CR = 0.60
planning VEE = 0.44
(2 items, RTP) λav = 0.68
λlow = 0.53
5. CFA Internal Stimuli χ2SB = 16.41 (p = 0.25, df = 14)
GFI = 1.00
CFI = 1.00
RMSEA = 0.012 (p = 1.00)
5a. Firm Proactive Technology α = 0.78 CR = 0.76
(2 items, Ftechn) VEE = 0.63
λav = 0.80
λlow = 0.79
5b. Firm Proactive Financial α = 0.86 CR = 0.86
(2 items, Ffinanc) VEE = 0.76
λav = 0.87
λlow = 0.81
228
Appendix 3
Table A3-5 Results structural models integral export performance – set 3.

Model A (export ratio) Model B (export sales)
χ2 = 1811.36 (.00, df = 528) χ2 = 1803,80 (.00, df = 528)
RMSEA = .046 (1.00) RMSEA = .046 (1.00)
GFI = .92 GFI = .92
CFI = .88 CFI = .89
On ‘risk attitude towards change’ from
Age .13 (4.03) .13 (4.03)
Education -.07 (-2.13) -.07 (-2.13)
Firm size -.18 (-4.86) -.18 (-4.86)
Age .12 (3.50) .12 (3.50)
Education .04 (1.03) .04 (1.03)
Firm size -.06 (-1.51) -.06 (-1.51)
Supply market -.08 (-2.36) -.08 (-2.36)
Labour market .11 (2.95) .11 (2.95)
International sales market .02 (.70) .02 (.70)
Capital market -.01 (-.38) -.01 (-.38)
Education .00 (.09) .00 (.09)
Firm size .10 (2.85) .10 (2.85)
Risk attitude towards change -.02 (-.46) -.02 (-.46)
Supply market -.10 (-2.78) -.10 (-2.78)
Labour market .16 (4.26) .16 (4.26)
Capital market -.03 (-.91) -.03 (-.91)
Education .00 (.04) .00 (.04)
Firm size .10 (3.02) .10 (3.02)
Linguistic skills .00 (.07) .00 (.07)
229

Risk attitude towards planning .05 (1.32) .05 (1.32)
Supply market -.03 (-.92) -.03 (-.92)
Labour market .04 (1.00) .04 (1.00)
Capital market .02 (.50) .02 (.50)
Education -.01 (-.40) -.01 (-.40)
Firm size .13 (4.12) .13 (4.12)
Risk attitude towards change -.04 (-.96) -.04 (-.96)
Supply market -.08 (-2.38) -.08 (-2.38)
Labour market -.03 (-.72) -.03 (-.72)
Capital market .05 (1.45) .05 (1.45)
Education .00 (-.13) .00 (-.13)
Firm size .23 (6.98) .23 (6.98)
Linguistic skills .00 (-.06) .00 (-.06)
Supply market -.02 (-.45) -.02 (-.45)
Labour market .01 (.33) .01 (.33)
International sales market .00 (-.07) .00 (-.07)
Capital market .03 (.78) .03 (.78)
Education -.02 (-.55) -.02 (-.55)
Firm size -.06 (-1.67) -.06 (-1.67)
Import experience -.01 (-.24) -.01 (-.24)
Risk attitude towards change .01 (.31) .01 (.31)
Supply market -.08 (-2.40) -.08 (-2.40)
230
Appendix 3
Labour market -.10 (-2.92) -.10 (-2.92)

Capital market .01 (.19) .01 (.19)
Education .05 (1.99) .05 (1.99)
Firm size .03 (.90) .03 (.90)
Supply market .16 (5.08) .07 (2.74)
Labour market .03 (.82) -.04 (-1.44)
Capital market .04 (1.38) .03 (1.29)
Firm size .09 (2.94) .52 (21.05)
Internal technological stimuli .12 (3.50) .05 (1.76)
Internal financial stimuli .00 (-.13) .02 (.99)
Product superiority -.07 (-1.47) -.09 (-2.73)
Competitive pricing -.07 (-1.83) .01 (.25)
Production capability -.03 (-.96) .03 (1.19)
Education .02 (.91) .05 (2.45)
Age .05 (2.05) .02 (1.05)
Linguistic skills -.06 (-1.99) .00 (-.08)
Risk attitude towards change .01 (.20) .04 (1.26)
Domestic informal information -.08 (-2.88) -.02 (-1.08)
Domestic formal information .00 (.07) .01 (.24)
Domestic cooperation -.01 (-.49) -.02 (-1.15)
231
Table A3-6 Total effects of variables on export performance – set 3.

Total (direct + indirect) effects of variable: On export ratio On export sales
Environment:
Supply market .19 (4.40) .10 (1.97)
Labour market .02 (.41) -.09 (-1.63)
Capital market .07 (1.63) .09 (1.58)
Firm size .14 (4.63) .91 (24.15)
Internal technological stimuli .15 (3.50) .10 (1.76)
Internal financial stimuli -.01 (-.13) .05 (.99)
Product superiority -.09 (-1.47) -.20 (-2.73)
Competitive pricing -.10 (-1.83) .02 (.25)
Production capability -.05 (-.96) .07 (1.19)
Linguistic skills -.11 (-1.43) .03 (.28)
Education .20 (1.09) .55 (2.45)
Age .35 (2.34) .29 (1.57)
Risk attitude towards change .00 (.07) .06 (.92)
Export Activities:
Domestic informal information -.07 (-2.46) -.03 (-.84)
Domestic formal information .02 (.68) .02 (.56)
Domestic cooperation -.02 (-.49) -.05 (-1.15)
232
Appendix 4
Results Information Model on
Imputed Data Set 2
Items Construct (n = 1122)
CFA Attitude 1991 χ2 = .011 (p = 0.91, df = 1)157
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.98)
λav = 0.68
λlow = 0.67
λav = 0.65
λlow = 0.54
GFI = 1.00
CFI = 1.00
RMSEA = .015 (p = 0.71)
λav = 0.64
λlow = 0.55
λav = 0.65
λlow = 0.50
CFA Attitude 1995 χ2 = .29 (p = 0.59, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.89)
157
233

Items Construct (n = 1122)
λav = 0.71
λlow = 0.63
λav = 0.65
λlow = 0.48
Table A4-2 Results Structural model – Information Behavior (Formal versus

Informal Information) – set 2.
χ2 = 26.89 (.06, df = 17) χ2 = 26.49 (.066, df = 17)
RMSEA = .022 (1.00) RMSEA = .022 (1.00)
GFI = .99 GFI = .99
CFI = .99 CFI = .99
Risk towards planning .03 (.44) .03 (.44)
Risk towards planning .02 (.27) -.01 (-.10)
Domestic informal information .00 (-.17) .09 (2.98)
Domestic formal information .01 (.48) .06 (2.06)
Standardised loadings, t-values in parentheses, significant parameters at 95% bold, at 90% italic.
Loadings deviating from the results stemming from the first imputed data set in significance are
highlighted.
234
Appendix 4
Table A4-3 Results Structural model – Information Behavior (Export Market

Research versus assistance versus Intelligence Information) – set 2.
χ2 = 32.29 (.06, df = 21) χ2 = 31.91 (.06, df = 21)
RMSEA = .022 (1.00) RMSEA = .021 (1.00)
GFI = .99 GFI = .99
CFI = .99 CFI = .99
Risk towards planning .01 (.19) -.01 (-.27)
Dom. intelligence information .00 (-.17) .09 (2.98)
Dom. research information -.02 (-.52) .05 (1.56)
Dom. assistance information .06 (1.98) .06 (2.02)
highlighted.
235
Table A4-4 Results Structural Model – Information Behavior (Formal versus

Informal Information) – set 2 – three years.
χ2 = 902.14 (.00, df = 239) χ2 = 1009.65 (.00, df = 239)
RMSEA = .051 (.32) RMSEA = .054 (.035)
GFI = .94 GFI = .94
CFI = .92 CFI = .92
Size and significance of effects within one year
1991 1993 1995 1991 1993 1995
(.10) (-.41) (-1.66) (.08) (-.30) (-1.39)
(2.67) (3.33) (1.97) (2.65) (3.03) (1.94)
(-1.08) (-.63) (-3.03) (-.91) (.10) (-2.86)
(2.94) (4.46) (3.08) (2.74) (3.90) (2.99)
(-5.55) (-3.07) (-2.86) (-5.84) (-.98) (-2.29)
(4.18) (2.28) (-.80) (4.44) (1.27) (1.13)
(-3.70) (-2.42) (-3.15) (-3.74) (-.44) (-2.46)
Risk towards planning .02 .01 .03 .02 -.04 .03
(.32) (.26) (.51) (.36) (-.84) (.65)
(-2.71) (-2.84) (-4.10) (-5.68) (-4.73) (-5.22)
(1.71) (1.19) (.95) (1.79) (-1.35) (1.11)
Domestic informal information .03 .02 .00 .11 .06 .08
(1.10) (.84) (-.10) (4.07) (2.14) (3.42)
Domestic formal information .11 .10 -.01 .09 .04 .04
(4.09) (3.44) (-.27) (3.58) (1.53) (1.85)
Foreign informal information .11 .15 .15 .16 .27 .21
(3.99) (5.34) (6.02) (5.72) (10.70) (9.67)
Foreign formal information .10 .14 .03 .15 .23 .19
(3.50) (4.96) (1.35) (5.59) (9.05) (8.58)
Size and significance of longitudinal effects
Export performance 1991 .05 (1.69) .00 (.13)
236
Appendix 4

Export performance 1991 .13 (4.34) .22(7.51)
Domestic informal information 1991 -.04 (-1.44) .01 (.55)
Domestic formal information 1991 .11 (4.56) .03 (1.23)
Foreign informal information 1991 .08 (3.32) .09 (4.32)
Foreign formal information in 1991 .14 (5.51) .16 (8.01)
Domestic informal information 1993 .05 (1.85) .02 (.76)
Domestic formal information 1993 .08 (3.00) .05 (1.90)
Foreign informal information 1993 .21 (7.75) .23 (9.52)
Loadings deviating from the results stemming from the first imputed data set in significance are highlighted.
237
Table A4-5 Results Structural Model – Information Behavior (Export Market

Research versus Assistance versus Intelligence Information) – set 2 – three
years.
χ = 1183.17 (.00, df = 357) χ2 = 1295.15 (.00, df = 357)
2
RMSEA = .046 (.98) RMSEA = .048 (.82)

GFI = .94 GFI = .94
CFI = .92 CFI = .92
On ‘collecting intelligence information at home’ from
1991 1993 1995 1991 1993 1995
(.08) (-.38) (-1.67) (.12) (-.19) (-1.31)
(2.72) (3.26) (2.00) (2.66) (2.93) (1.94)
On ‘collecting research information at home’ from
Risk towards changes .00 -.01 -.16 .00 .03 -.14
(-.08) (-.13) (-2.87) (-.02) (.43) (-2.45)
(2.50) (3.40) (2.89) (2.40) (2.92) (2.70)
On ‘collecting assistance information at home’ from
(-2.58) (-.82) (-1.81) (-2.26) (.02) (-1.87)
(3.27) (4.30) (2.52) (2.98) (3.75) (2.56)
On ‘collecting intelligence information abroad’ from
(-5.55) (-2.83) (-2.76) (-5.77) (-.63) (-1.99)
(4.17) (1.99) (.74) (4.35) (.91) (1.01)
On ‘collecting research information abroad’
(-3.47) (-1.63) (-1.74) (-3.49) (.52) (-.96)
Risk towards planning .04 -.05 -.04 .04 -.11 -.04
(.64) (-.96) (-.69) (.64) (-2.21) (-.67)
On ‘collecting assistance information abroad’
(-2.64) (-1.99) (-3.50) (-5.03) (-.57) (-2.78)
Risk towards planning -.02 .05 .10 -.03 .04 -.10
(-.26) (1.04) (1.83) (1.27) (.24) (1.92)
(-2.37) (-2.49) (-3.83) (-5.07) (-4.26) (-4.69)
Risk towards planning .08 -.04 -.04 .07 -.06 .04
(1.48) (.90) (-.76) (.95) (-1.48) (.87)
Domestic intelligence information .03 .02 .00 .11 .06 .08
(1.13) (.73) (-.14) (4.32) (2.33) (3.51)
Domestic research information .08 .07 -.01 .08 .04 .06
(2.79) (2.50) (-.58) (2.98) (1.40) (2.87)
238
Appendix 4
Domestic assistance information .12 .08 .03 .09 .05 .01

(4.39) (3.04) (1.02) (3.40) (1.86) (.28)
Foreign intelligence information .12 .15 .15 .17 .26 .21
(4.17) (5.39) (6.22) (6.25) (10.97) (10.13)
Foreign research information .10 .12 .03 .18 .22 .16
(3.77) (4.25) (1.14) (6.74) (8.82) (7.55)
Foreign assistance information .06 .10 .04 .09 .15 .15
(2.26) (3.82) (1.52) (3.58) (5.92) (6.91)
On ‘collecting intelligence information at home’ in 1993 from
On ‘collecting research information at home’ in 1993 from
On ‘collecting assistance information at home’ in 1993 from
On ‘collecting intelligence information abroad’ in 1993 from
On ‘collecting research information abroad’ in 1993 from
On ‘collecting assistance information abroad’ in 1993 from
Domestic intelligence information 1991 -.03 (-1.37) .02 (.80)
Domestic research information 1991 .10 (3.84) .01 (.71)
Domestic assistance information 1991 .09 (3.66) .05 (2.37)
Foreign intelligence information 1991 .09 (3.44) .10 (4.65)
Foreign research information 1991 .15 (5.89) .16 (7.88)
Foreign assistance information 1991 .07 (2.87) .12 (5.56)
Domestic intelligence information 1993 .05 (1.81) .02 (.97)
Domestic research information in 1993 .08 (2.56) .06 (2.41)
Standardised loadings, t-values in parentheses, significant parameters at 95% bold, at 90% italic. Loadings
deviating from the results stemming from the first imputed data set in significance are highlighted.
239
240
Appendix 5
Results Information Model on
Imputed Data Set 3
Items (n = 1122)
CFA Attitude 1991 χ2 = .oo27 (p = 0.96, df = 1)158
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 1.00)
λav = 0.68
λlow = 0.68
λav = 0.66
λlow = 0.55
GFI = 1.00
CFI = 1.00
RMSEA = .045 (p = 0.45)
λav = 0.63
λlow = 0.53
λav = 0.63
λlow = 0.48
CFA Attitude 1995 χ2 = .77 (p = 0.38, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .00 (p = 0.79)
158
241

Items (n = 1122)
λav = 0.71
λlow = 0.63
λav = 0.63
λlow = 0.46
Table A5-2 Results Structural model – Information Behavior (Formal versus

Informal Information) – set 3.
χ2 = 24.91 (.097, df = 17) χ2 = 21.03 (.23, df = 17)
RMSEA = .021 (1.00) RMSEA = .015 (1.00)
GFI = 1.00 GFI = 1.00
CFI = 1.00 CFI = 1.00
Risk towards planning -17 (2.97) .02 (.05)
Domestic informal information .01 (.29) .13 (4.50)
Domestic formal information .02 (.62) .10 (3.32)
highlighted.
242
References
Table A5-3 Results Structural model – Information Behavior (Export Market

Research versus Assistance versus Intelligence Information) – set 3.
χ2 = 28.11 (.14, df = 21) χ2 = 23.68 (.31, df = 21)
RMSEA = .018 (1.00) RMSEA = .011 (1.00)
GFI = 1.00 GFI = 1.00
CFI = 1.00 CFI = 1.00
Risk towards planning .17 (2.91) .02 (.32)
Dom. intelligence information .01 (.29) .13 (4.50)
Dom. research information -.01 (-.58) .06 (2.07)
Dom. assistance information .07 (2.37) .12 (4.00)
highlighted.
243
Table A5-4 Results Structural Model – Information Behavior (Formal versus

Informal Information) – set 3 – three years.
χ2 = 857.56 (.00, df = 239) χ2 = 921.62 (.00, df = 239)
RMSEA = .051 (.36) RMSEA = .051 (.34)
GFI = .94 GFI = .94
CFI = .92 CFI = .93
1991 1993 1995 1991 1993 1995
Risk towards changes .00 -.06 -.10 -.01 -.01 -.04
(-.01) (-1.05) (-1.94) (-.15) (-.21) (-.82)
(2.79) (3.79) (1.63) (2.90) (3.57) (1.17)
(-1.52) (-2.71) (-3.04) (-1.43) (-1.85) (-2.45)
(3.45) (6.09) (3.32) (3.36) (5.90) (3.03)
(-5.04) (-4.11) (-4.33) (-5.06) (-2.06) (-2.98)
(3.72) (3.19) (-.05) (3.72) (2.86) (-.65)
(-4.23) (-4.18) (-3.57) (-4.31) (-2.58) (-2.70)
(1.34) (1.45) (.90) (1.44) (1.08) (.55)
(-.47) (-1.02) (-6.02) (-4.81) (-3.89) (-4.76)
Risk towards planning .06 -.08 -.19 .10 -.04 .02
(1.08) (-1.52) (3.82) (1.73) (-.85) (.41)
Domestic informal information .07 .05 -.02 .08 .05 .07
(2.57) (1.88) (-.74) (3.04) (1.97) (3.50)
Domestic formal information .07 .12 -.06 .10 .05 .01
(2.51) (4.21) (-2.21) (3.87) (2.06) (.47)
Foreign informal information .11 .19 .11 .17 .26 .21
(3.76) (6.57) (4.42) (6.24) (10.51) (9.75)
Foreign formal information .11 .16 .03 .16 .24 .08
(4.15) (5.75) (1.29) (6.20) (9.63) (3.58)
244
References

Domestic informal information in 1991 -.03 (-1.12) .05 (1.97)
Domestic formal information in 1991 .00 (.07) .05 (2.06)
Foreign informal information in 1991 .11 (4.21) .26 (10.51)
Domestic informal information in 1993 .07 (2.48) .12 (5.10)
Domestic formal information in 1993 .04 (1.51) .11 (4.52)
Foreign informal information in 1993 .18 (6.66) .26 (11.15)
Loadings deviating from the results stemming from the first imputed data set in significance are highlighted.
245
Table A5-5 Results Structural Model – Information Behavior (Export Market

Research versus Assistance versus Intelligence Information) – set 3 – three
years.
χ = 1196.75 (.00, df = 357)
2
χ2 = 1228.16 (.00, df = 357)
RMSEA = .047 (.93) RMSEA = .047 (.96)
GFI = .94 GFI = .93
CFI = .92 CFI = .92
On ‘collecting intelligence information at home’ from
1991 1993 1995 1991 1993 1995
(-.13) (-1.12) (-1.99) (-.15) (-.15) (-.68)
(2.91) (3.81) (1.73) (2.91) (3.49) (1.14)
On ‘collecting research information at home’ from
(-1.08) (-2.39) (-3.06) (-.82) (-1.31) (-2.25)
(3.49) (5.79) (3.25) (3.25) (5.51) (2.85)
On ‘collecting assistance information at home’ from
(-2.39) (-2.29) (-2.05) (-2.27) (-1.62) (-1.57)
(3.40) (4.53) (3.04) (3.23) (4.36) (2.79)
On ‘collecting intelligence information abroad’ from
(-5.13) (-4.12) (-4.27) (-5.00) (-1.89) (-2.67)
(3.79) (3.16) (-.06) (3.62) (2.67) (-.81)
On ‘collecting research information abroad’
(-3.59) (-3.23) (-2.54) (-3.57) (-1.36) (-1.68)
Risk towards planning .08 .01 .02 .08 -.02 .00
(1.33) (.19) (.46) (1.29) (-.42) (.07)
On ‘collecting assistance information abroad’
(-3.84) (-3.76) (-3.39) (-4.18) (-2.62) (-2.36)
(1.08) (2.42) (1.30) (1.20) (2.10) (.82)
(-.17) (-.39) (-5.63) (-5.07) (-3.21) (-4.17)
(.84) (-1.90) (3.58) (.95) (-1.29) (-.16)
Domestic intelligence information .07 .06 -.02 .08 .05 .08
(2.51) (1.96) (-.62) (3.17) (2.10) (3.73)
Domestic research information .03 .09 -.06 .10 .07 .01
(1.13) (4.23) (-2.37) (3.73) (2.77) (.30)
246
References
Domestic assistance information .11 .12 -.01 .09 .05 .02

(4.10) (3.02) (-.26) (3.51) (1.82) (1.11)
Foreign intelligence information .11 .19 .12 .18 .25 .21
(3.82) (6.72) (4.49) (6.75) (10.60) (9.93)
Foreign research information .12 .13 .04 .20 .21 .07
(4.55) (4.71) (1.43) (7.65) (8.65) (3.45)
Foreign assistance information .07 .14 .02 .10 .17 .06
(2.61) (5.05) (.98) (3.65) (7.12) (2.65)
Domestic intelligence information 1991 -.03 (-1.22) .05 (2.46)
Domestic research information 1991 -.03 (-1.09) .04 (1.87)
Domestic assistance information 1991 .03 (1.36) .00 (.20)
Export performance 1993 .00 (-.12) .11 (4.20)
Domestic intelligence information 1993 .07 (2.49) .13 (5.40)
Domestic research information in 1993 .05 (1.75) .10 (4.45)
Standardised loadings, t-values in parentheses, significant parameters at 95% bold, at 90% italic. Loadings
deviating from the results stemming from the first imputed data set in significance are highlighted.
247
248
References
Aaby, N. E. & Slater, S. F. 1989, "Management Influences on Export Performance: A
Review of the Empirical Literature 1978-88”, International Marketing Review, vol. 6, no.
4, pp. 7-25.
Ahmed, S. U. 1985, “nAch, Risk-Taking Propensity, Locus of Control and

Entrepreneurship”, Personal Individual Differences, vol. 6, no. 6, pp. 781-782.
Aish, A.-M. & Jöreskog, K. G. 1990, “Structural Modeling by Example: Applications in

Educational, Sociological, and Behavioral Research”, Contemporary Psychology, vol. 35,
no. 7, p. 687.
Albaum, G., Strandskov, J., & Duerr, E. 2002, International Marketing and Export
Management, 4 edn, Pearson Education Limited, Edinburgh Gate.
Alsem, K.J. 2001, Strategische Marketingplanning Wolters-Noordhoff, Groningen/Houten.
Anderson, E. & Coughlan, A. T. 1987, “International Market Entry and Expansion via
Independent or Integrated Channels of Distribution”, Journal of Marketing, vol. 51, no.
January, pp. 71-82.
Anderson, J. C. & Gerbing, D. W. 1988, “Structural Equation Modeling in Practice: A

Review and Recommended Two-Step Approach”, Psychological Bulletin, vol. 103, no. 3,
pp. 411-423.
Avlonitis, G. J. & Gounaris, S. P. 1997, “Marketing Orientation and Company

Performance”, Industrial Marketing Management, vol. 26, no. 5, pp. 385-402.
Avlonitis, G. J. & Gounaris, S. P. 1999, “Marketing Orientation and its Determinants: An

Empirical Analysis”, European Journal of Marketing, vol. 33, no. 11/12, pp. 1003-1036.
Axinn, C. N. 1988, “Export Performance: Do Managerial Perceptions Make a Difference?”,

International Marketing Review, vol. 5, no. 2, pp. 61-71.
Bagozzi, R. P. & Yi, Y. 1988, “On the Evaluation of Structural Equation Models”,
Academy of Marketing Science, vol. 16 , no. 1, pp. 74-94.
Balabanis, G. I. & Katsikea, E. S. 2003, “Being an Entrepreneurial Exporter: Does it

Pay?”, International Business Review, vol. 12, pp. 233-252.
249
Baldauf, A., Cravens, D. W., & Wagner, U. 2000, “Examining Determinants of Export
Performance in Small Open Economies”, Journal of World Business, vol. 35, no. 1, pp. 61-
79.
Barney, J. B. 1991, “Firm Resources and Sustained Competitive Advantage”, Journal of

Management, vol. 17, no. 1, pp. 99-120.
Beamish, P. W., Craig, R., & McLellan, K. 1993, “The Performance Characteristics of
Canadian versus U.K. Exporters in Small and Medium Sized Firms”, Management
International Review, vol. 33, no. 2, pp. 121-137.
Beamish, P. W., Karavis, L., Goerzen, A., & Lane, C. 1999, “The Relationship Between
Organizational Structure and Export Performance”, Management International Review, vol.
39, no. 1, pp. 37-54.
Belich, T. J. & Dubinsky, A. J. 1995, “Factors Related to Information Acquisition in

Exporting Organizations”, Journal of Business Research, vol. 33, pp. 1-11.
Belich, T. J. & Dubinsky, A. J. 1999, “Information Processing among Exporters: An

Empirical Examination of Small Firms”, Journal of Marketing Theory & Practice, vol. 7,
no. 4, pp. 45-58.
Bell, J., Murray, M., & Madden, K. 1991, “Developing Exportise: An Irish Perspective”,
Journal of Small Business Management, vol. 10, no. 2, pp. 37-53.
Bentler, P. M. 1992, “On the Fit of Models to Covariances and Methodology to the
Bulletin”, Psychological Bulletin, vol. 112, no. 3, p. 400.
Bernard, A. B. & Jensen, J. B. 1997, Exceptional Exporter Performance: Cause, Effect, or

Both?, National Bureau of Economic Research, Cambridge, MA.
Bernard, A. B. & Wagner, J. 1997, “Exports and Success in German Manufacturing”,

Weltwirtschaftliches Archiv, vol. 133 , no. 1, pp. 134-157.
Bijmolt, T. H. A. & Zwart, P. S. 1994, “The Impact of Internal Factors on the Export
Success of Dutch Small and Medium-Sized Firms”, Journal of Small Business
Management, vol. 32, no. 2, pp. 69-83.
Bilkey, W. J. 1978, “An Attempted Integration of the Literature on the Export Behavior of
Firms”, Journal of International Business Studies, vol. 9, pp. 33-46.
250
References
Boelmans Kranenburg, H. A. H. 1977, “De Export van Gezouten Haring uit Nederland naar
de Sowjet-Unie,” in Ondernemende Geschiedenis, Martinus Nijhoff, Den Haag, pp. 311-
320.
Bollen, K. A. 1994, Structural Equations with Latent Variables, 6 edn, Wiley, New York.
Bonaccorsi, A. 1992, “On the Relationship Between Firm Size and Export Intensity”,
Journal of International Business Studies, vol. 23, no. 4, pp. 605-635.
Boomsma, A. 1983, On the Robustness of LISREL (Maximum Likelihood Estimation)

Against Small Sample Size and Non-Normality, Rijksuniversiteit Groningen.
Bourantas, D. & Halikias, J. 1990, Patterns of export competitive strategy, Paper Presented
at EMAC 1990.
Bromiley, P. 1991, “Testing a Causal Model of Corporate Risk Taking and Performance”,
Academy of Management Journal, vol. 34, no. 1, pp. 37-59.
Byrne, B. M. 1998, Structural Equation Modeling with LISREL, PRELIS, and SIMPLIS:
Basic Concepts, Applications, and Programming, Lawrence Erlbaum Associates,
Publishers , Mahwah, N.J.
Cadogan, J. W. & Diamantopoulos, A. 1995, “Narver and Slater, Kohli and Jaworski and
the Market Orientation Construct: Integration and Internationalization”, Journal of
Strategic Marketing, vol. 3, pp. 41-60.
Cadogan, J. W., Diamantopoulos, A. , & De Mortanges, C. P. 1999, “A Measure of Export

Market Orientation: Scale Development and Cross-cultural Validation”, Journal of
International Business Studies, vol. 30, no. 4, pp. 689-707.
Cadogan, J. W., Paul, N. J., Salminen, R. T., Puurmalainen, K., & Sunqvist, S. 2001, “Key
Antecedents to ‘Export’ Market-Oriented Behaviors: A Cross-National Empirical
Examination”, International Journal of research in Marketing , vol. 18, pp. 261-282.
Cafferata, R. & Mensi, R. 1995, “The Role of Information in the Internationalisation of

SMEs: A Typological Approach”, International Small Business Journal, vol. 13, no. 3, pp.
35-46.
Cateora, P. R. & Graham, J. L. 2002, International Marketing, 11 edn, McGraw-Hill.
251
Cattell, R. B. & Vogelmann, S. 1977, “A Comprehensive Trial of the Scree and KG

Criteria for Determining the Number of Factors”, Multivariate Behavioral Research, vol.
12, no. 3, pp. 289-325.
Cavusgil, S. T. & Naor, J. 1987, “Firm and Management Characteristics as Discriminators

of Export Marketing Activity”, Journal of Business Research, vol. 15, pp. 221-235.
Cavusgil, S. T. & Kirpalani, M. V. H. 1993, “Introducing Products into Export Markets:

Success Factors”, Journal of Business Research, vol. 27, pp. 1-15.
Cavusgil, S. T. & Zou, S. 1994, “Marketing Strategy-Performance Relationship: An

Investigation of the Empirical Link in Export Market Ventures”, Journal of Marketing, vol.
58, no. 1, pp. 1-21.
Chetty, S. K. & Hamilton, R. T. 1993, “Firm-Level Determinants of Export Performance: A

Meta-Analysis”, International Marketing Review, vol. 10, no. 3, pp. 26-34.
Chetty, S. K. & Blankenburg Holm, D. 2000, “Internationalisation of Small to Medium-

Sized Manufacturing Firms: A Network Approach”, International Business Review, vol. 9,
no. 1, pp. 77-93.
Child, J. 1972, “Organizational Structure, Environment and Performance: The Role of

Strategic Choice”, Sociology, vol. 6, no. January, pp. 1-22.
Christensen, C. H., Da Rocha, A., & Kerbel Gertner, R. 1987, “An Empirical Investigation
of the Factors Influencing Export Success of Brazilian Firms”, Journal of International
Business Studies, vol. 18, no. Fall, pp. 61-77.
Chryssochoidis, G. M. 1996, “Successful Exporting: Exploring the Transformation of

Export Product Portfolios”, Journal of Global Marketing, vol. 10, no. 1, pp. 7-31.
Churchill, G. A., Jr. 1999, Marketing Research: Methodological Foundations, 7 edn,

Dryden Press, Fort Worth.
Churchill, N. C. & Lewis, V. L. 1983, “The Five Stages of Small Business Growth”,
Harvard Business Review, vol. 61, no. 3.
Cooper, H. 1998, Synthesizing Research, A Guide for Literature Reviews, 3 edn, Sage,
California.
Culpan, R. 1989, “Export Behavior of Firms: Relevance of Firm Size”, Journal of Business
Research, vol. 18, pp. 207-218.
252
References
Cyert, R. M. & March, J. G. 1963, A Behavioral Theory of the Firm, Prentice-Hall, Inc.,
Englewood Cliffs, NJ.
Czinkota, M. R. 2000, “International Information Cross-Fertilization in Marketing: An

Empirical Assessment”, European Journal of Marketing, vol. 34, no. 11/12, pp. 1305-1314.
Das, M. 1994, “Successful and Unsuccessful Exporters from Developing Countries: Some
Preliminary Findings”, European Journal of Marketing, vol. 28, no. 12, pp. 19-33.
Davis, E. & Keys, B. 1996, “Success Factors in Going International: Experiences from
Small Firms in Southeastern USA”, European Business Review, vol. 96, no. 2, pp. 3-10.
De Luz, M. 1993, “Relationship Between Export Strategy Variables and Export

Performance for Brazil-Based Manufacturers”, Journal of Global Marketing, vol. 7, no. 1,
pp. 87-110.
De Mortanges, C. P. 1990, Expert Systems as a Potential Tool for Export Marketing

Decision-Making: A Preliminary Investigation, Rijksuniversiteit Groningen.
Dean, D. L., Mengüç, B., & Myers, C. P. 2000, “Revisiting Firm Characteristics, Strategy,
and Export Performance Relationship: A Survey of the Literature and an Investigation of
New Zealand Small Manufacturing Firms”, Industrial Marketing Management, vol. 29, no.
5, pp. 461-477.
Denis, J. E. & Depelteau, D. 1985, “Market Knowledge, Diversification and Export

Expansion “, Journal of International Business Studies, vol. 16, no. Fall, pp. 77-89.
Diamantopoulos, A. & Inglis, K. 1988, “Identifying Differences Between High- and Low-
Involvement Exporters”, International Marketing Review, vol. 5, no. 2, pp. 52-60.
Diamantopoulos, A., Schlegelmilch, B. B., & Allpress, C. 1990, “Export Market Research
in Practice: A Comparison of Users and Non-users”, Journal of Marketing Management,
vol. 6, no. 3, pp. 257-273.
Diamantopoulos, A. 1999, “Viewpoint- Export Performance Measurement: Reflective

versus Formative Indicators”, International Marketing Review, vol. 16, no. 6, pp. 444-457.
Diamantopoulos, A. & Winklhofer, H. M. 2001, “Index Construction with Formative

Indicators: An Alternative to Scale Developments”, Journal of the Academy of
Management Science, vol. 38, no. 2, pp. 269-277.
253
Dichtl, E., Köglmayr, H. G., & Müller, S. 1990, “International Orientation as a

Precondition for Export Success”, Journal of International Business Studies, vol. 21, no. 1,
pp. 23-40.
Does, R. M. T. & De Mortanges, C. F. W. P. 1998, “Marktgerichtheid en Exportresultaat”,

Maandblad voor Accountancy en Bedrijfseconomie, vol. 72, no. 6, pp. 337-343.
Dominguez, L. V. & Sequeira, C. G. 1993, “Determinants of LDC Exporters' Performance:

A Cross-National Study”, Journal of International Business Studies, vol. 24, no. 1, pp. 19-
40.
Donthu, N. & Kim, S. H. 1993, “Implications of Firm Controllable Factors on Export

Growth”, Journal of Global Marketing, vol. 7, no. 1, pp. 47-63.
Douglas, S. P. & Craig, C. S. 1989, “Evolution of global marketing strategy: Scale, scope
and synergy”, Columbia Journal of World Business, vol. 24, no. 3, pp. 47-59.
Eagly, A. H. & Chaiken, S. 1993, The Psychology of Attitudes, Harcourt Brace Jovanovich,
Inc., Orlando.
Edwards, J. R. & Bagozzi, R. P. 2000, “On the Nature and Direction of Relationships
Between Constructs and Measures”, Psychological Methods, vol. 5, no. 2, pp. 155-174.
European Commission 2002, Highlights from the 2002 Survey, Enterprise Publications, 8.
Evangelista, F. U. 1994, “Export Performance and its Determinants: Some Empirical

Evidence from Australian Manufacturing Firms”, Advances in International Marketing,
vol. 6, pp. 207-229.
Evangelista, F. U. 1996, “Linking Business Relationships to Marketing Strategy and Export

Performance: A Proposed Conceptual Framework”, Advances in International Marketing,
vol. 8, pp. 59-83.
Finkel, S. E. 1995, Causal Analysis with Panel Data, Sage, Thousand Oaks CA.
Fornell, C. & Bookstein, F. L. 1982, “Two Structural Equation Models: LISREL and PLS
Applied to Consumer Exit-Voice Theory”, Journal of Marketing Research, vol. 19, no.
November, pp. 440-452.
Francis, J. & Collins-Dodd, C. 2000, “The Impact of Firms' Expert Orientation on the
Export Performance of High-Tech Small and Medium-Sized Enterprises”, Journal of
International Marketing, vol. 8, no. 3, pp. 84-103.
254
References
Gankema, H. G. J., Snuif, H. R., & Zwart, P. S. 2000, “The Internationalization Process of
Small and Medium-Sized Enterprises: An Evolution of Stage Theory”, Journal of Small
Business Management, vol. 38, no. 4, pp. 15-27.
Gelman, A., Carlson, S., Stern, H. S., & Rubin, D. B. 1995, Bayesian Data Analysis
Chapman and Hall, New York.
Gemünden, H. G. 1991, “Success Factors of Export Marketing,” in New Perspectives on

International Marketing, S. J. Paliwoda, ed., Routledge, London, pp. 33-62.
Glass, G., McGraw, B., & Smith, M. L. 1981, Meta-Analysis in Social Research, Sage,
Beverly Hills CA.
Glazer, R., Steckel, J. H., & Winer, R. S. 1992, “A Locally Rational Decision Making: The
Distracting Effect of Information on Managerial Performance”, Management Science, vol.
38, no. 2, pp. 212-226.
Gómes-Mejia, L. R. & Palich, L. E. 1997, “Cultural Diversity and the Performance of

Multinational Firms”, Journal of International Business Studies, vol. 28, no. 2, pp. 309-
335.
Grant, R. M. 1991, “The Resource-Based Theory of Competitive Advantage: Implications

for Strategy Formulation”, California Management Review, vol. 33, no. 3, pp. 114-135.
Gray, B. J. 1997, “Profiling Managers to Improve Export Promotion Targeting”, Journal of

Green, P. E., Tull, D. S., & Albaum, G. S. 1988, Research for Marketing Decisions, 5 edn,
Prentice Hall, Englewood Cliffs.
Haahti, A., Hall, G., & Donckels, R. 1998, The Internationalization of SMEs, Routledge,
London.
Hair, J. F. jr., Anderson, R. E., Tatham, R. L., & Black, W. C. 1998, Multivariate Data
Analysis, 5 edn, Prentice-Hall International, London.
Hart, S. & Tzokas, N. 1999, “The Impact of Marketing Research Activity on SME Export
Performance: Evidence from the UK”, Journal of Small Business Management, vol. 37, no.
2, pp. 63-75.
255
Hart, S. J., Webb, J. R., & Jones, M. V. 1994, “Export Marketing Research and the Effect
of Export Experience in Industrial SMEs”, International Marketing Review, vol. 11, no. 6,
pp. 4-22.
Havnes, P.-A. 1998, Dynamics of Small Business Internationalisation. A European Panel

Study, University of Stirling.
Hayduk, L. A. 1987, Structural Equations Modeling with LISREL: Essentials and

Advances, the John Hopkins University Press, Baltimore.
Hayduk, L. A. 1996, LISREL Issues, Debates, and Strategies John Hopkins University
Press , London.
Holzmüller, H. H. & Kasper, H. 1990, “The Decision-Maker and Export Activity: A Cross-
National Comparison of Austrian Managers”, Management International Review, vol. 30,
pp. 217-230.
Holzmüller, H. H. & Kasper, H. 1991, “On a Theory of Export Performance: Personal and
Organizational Determinants of Export Trade Activities Observed in Small and Medium-
Sized Firms”, Management International Review, vol. 31, no. special issue, pp. 45-70.
Holzmüller, H. H. & Stöttinger, B. 1996, “Structural Modeling of Success Factors in

Exporting: Cross-Validation and Further Development of and Export Performance Model”,
Journal of International Marketing, vol. 4, no. 2, pp. 29-55.
Hoskisson, R. E., Hitt, M. A., & Hill, C. W. L. 1991, “Managerial Risk Taking in
Diversified Firms: An Evolutionary Perspective”, Organization Science, vol. 2, no. 3, pp.
296-314.
Israel.J.I. 1989, Dutch Primacy in World Trade, 1585-1740, Uitgeverij Van Wijnen,
Franeker.
Ito, K. & Pucik, V. 1993, “R&D Spending, Domestic Competition, and Export
Performance of Japanese Manufacturing Firms”, Strategic Management Journal, vol. 14,
pp. 61-75.
Jaworski, B. J. & Kohli, A. K. 1993, “Market Orientation: Antecedents and Consequences”,

Journal of Marketing, vol. 57, no. 3, pp. 53-70.
Johanson, J. & Vahlne, J. E. 1977, “The internationalization process of the firm - A model
of knowledge development and increasing foreign market commitments”, Journal of
256
References
Johanson, J. & Mattsson, L. G. 1988, “Internationalisation in Industrial Systems- A

Network Approach,” in Strategies in Global Competition, N. Hood & J. E. Vahlne, eds.,
Croom Helm, London, pp. 287-314.
Johnson, J. L. & Arunthanes, W. 1995, “Ideal and Actual Product Adaptation in US

Exporting Firms”, International Marketing Review, vol. 12, no. 3, pp. 31-46.
Jöreskog, K. G. & Sörbom, D. 1999, LISREL 8: Structural Equation Modeling with the
SIMPLIS Command Language, 4 edn, Scientific Software International, Chicago.
Julien, P.-A. 1998, The State of the Art in Small Business and Entrepreneurship, Ashgate
Publishing Company, Aldershot, England.
Julien, P.-A. & Ramangalahy, C. 2003, “Competitive Strategy and Performance of

Exporting SMEs: An Empirical Investigation of the Impact of Their Export Information
Search and Competencies”, Entrepreneurship: Theory & Practice, vol. 27, no. 3, pp. 227-
245.
Kaiser, H. F. 1970, “A Second Generation Little Jiffy”, Psychometrika, vol. 35, no.
December, pp. 401-415.
Kastelein, T. J. 1984, Groei Naar een Industriële Samenleving, 3 edn, Wolters-Noordhoff.
Katsikeas, C. S. 1994, “Export competitive advantages: the relevance of firm

characteristics”, International Marketing Review, vol. 11, no. 3, pp. 33-53.
Katsikeas, C. S., Piercy, N. F., & Ioannidis, C. 1996, “Determinants of Export Performance
in a European Context”, European Journal of Marketing, vol. 30, no. 6, pp. 6-35.
Katsikeas, C. S., Deng, S. L., & Wortzel, L. H. 1997, “Perceived Export Success Factors of
Small and Medium-Sized Canadian Firms”, Journal of International Marketing, vol. 5, no.
4, pp. 53-72.
Katsikeas, C. S., Leonidou, L. C., & Morgan, R. E. 2000, “Firm-Level Export Performance
Assessment: Review, Evaluation, and Development”, Journal of the Academy of Marketing
Science, vol. 28, no. 4, pp. 493-511.
Kaynak, E. & Kuan, W. K. 1993, “Environment, Strategy, Structure, and Performance in

the Context of Export Activity: An Empirical Study of Taiwanese Manufacturing Firms”,
Journal of Business Research, vol. 27, pp. 33-49.
257
Koh, A. C. & Robicheax, R. A. 1988, “Variations in Export Performance Due to

Differences in Export Marketing Strategy: Implications for Industrial Marketers”, Journal
of Business Research, vol. 17 , pp. 249-258.
Koh, A. C. 1991, “Relationships Among Organisational Characteristics, Marketing Strategy

and Export Performance”, International Marketing Review, vol. 8, no. 3, pp. 46-60.
Kohli, A. K. & Jaworski, B. J. 1990, “Market Orientation: The Construct, Research

Propositions, and Managerial Implications”, Journal of Marketing, vol. 54, no. April, pp. 1-
18.
Kohli, A. K., Jaworski, B. J., & Kumar, A. 1993, “MARKOR - A Measure of Market
Orientation”, Journal of Marketing Research, vol. 30, pp. 467-477.
Kotler, P. 2003, Marketing Management, 11 edn, Prentice Hall, New Jersey.
Kotler, P. & Armstrong, G. 2004, Principles of Marketing, 10 edn, Pearson/ Prentice Hall,
Upper Sadle River.
Kuratko, D. F. & Hodgetts, R. M. 2001, Entrepreneurship: A Contemporary Approach, 5

edn, The Dryden Press, Fort Worth.
Lages, L. F. 2000, “A Conceptual Framework of the Determinants of Export Performance:

Reorganizing Key Variables and Shifting Contingencies in Export Marketing”, Journal of
Global Marketing, vol. 13, no. 3, pp. 29-51.
Lages, L. F. & Montgomery, C. A. 2001, The Effects of Prior Export Performance On

Firms' Commitment to Exporting and Marketing Strategy Adaptation to the Foreign
Market: Evidence from Small and Medium-Sized Exporters, Graduate School of Business
Stanford University, Stanford.
Lang, J. R. & Calantone, R. J. 1997, “Small Firm Information Seeking as a Response to

Environmental Threats and Opportunities”, Journal of Small Business Management, vol.
35, no. 1, pp. 11-23.
Lautenschlager, G. J., Lance, C. E., & Flaherty, V. L. 1989, “Parallel Analysis Criteria:
Revised Equations for Estimating the Latent Roots of Random Data Correlation Matrices”,
Educational and Psychological Measurement, vol. 49, no. 2, p. 339.
Lee, C. S. & Yang, Y. S. 1990, “Impact of Export Market Expansion Strategy on Export
Performance”, International Marketing Review, vol. 7, no. 4, pp. 41-51.
258
References
Leeflang, P. S. H. & Mortanges de, C. P. 1993, “The Internal European Market and
Strategic Marketing Planning: Implications and Expectations”, Journal of International
Consumer Marketing, vol. 6, no. 2, pp. 7-23.
Leeflang, P. S. H., Wittink, D. R. , Wedel, M., & Naert, P. A. 2000, Building Models for
Marketing Decisions Kluwer Academic Publishers, Dordrecht.
Leeflang, P. S. H. 2003, Marketing Stenfert Kroese, Groningen .
Leonidou, L. C. 1995, “Empirical Research on Export Barriers Review, Assessment, and

Synthesis”, Journal of International Marketing, vol. 3, no. 1, pp. 29-43.
Leonidou, L. C. & Katsikeas, C. S. 1997, “Export Information Sources: The Role of

Organizational and Internationalization Influences”, Journal of Strategic Marketing, vol. 5,
pp. 65-87.
Leonidou, L. C. 1998, “Factors Stimulating Export Business: An Empirical Investigation”,

Journal of Applied Business Research, vol. 14, no. 2, pp. 43-68.
Leonidou, L. C. & Kaleka, A. A. 1998, “Behavioural Aspects of International Buyer-Seller

Relationships: Their Association with Export Involvement”, International Marketing
Review, vol. 15, no. 5, pp. 373-397.
Leonidou, L. C., Katsikeas, C. S., & Samiee, S. 2002, “Marketing Strategy Determinants of
Export Performance: A Meta-Analysis”, Journal of Business Research, vol. 55, no. 1, pp.
51-67.
Leonidou, L. C., Katsikeas, C. S., & Piercy, N. F. 1998, “Identifying Managerial Influences
on Exporting: Past Research and Future Directions”, Journal of International Marketing,
vol. 6, no. 2, pp. 74-102.
Leonidou, L. C., Katsikeas, C. S., & Samiee, S. 2002, “Marketing Stratgey Determinants of
Export Performance: a Meta-Analysis”, Journal of Business Research, vol. 55, no. 1, pp.
51-67.
Levitt, T. 1983, “The globalization of markets”, Harvard Business Review, vol. 61, no. 3,
pp. 92-102.
Ling-Lee, L. & Ogunmokun, G. O. 2001, “The Influence of Interfirm Relational

Capabilities on Export Advantage and Performance: An Empirical Analysis”, International
Business Review, vol. 10, no. 4, pp. 399-420.
259
Louter, P. J., Ouwekerk, C., & Bakker, B. A. 1991, “An Inquiry into Successful
Exporting”, European Journal of Marketing , vol. 25, no. 6, pp. 7-23.
Lumpkin, G. T. & Dess, G. G. 1996, “Clarifying the Entrepreneurial Orientation Construct

and Linking it to Performance”, Academy of Management Review, vol. 21, no. 1, pp. 135-
172.
MacCallum, R. C. & Browne, M. W. 1993, “The Use of Causal Indicators in Covariance

Structure Models: Some Practical Issues”, Psychological Bulletin, vol. 114, no. 3, pp. 533-
541.
Madsen, T. K. 1987, “Empirical Export Performance Studies: A Review of

Conceptualizations and Findings”, Advances in International Marketing, vol. 2, pp. 177-
198.
Madsen, T. K. 1989, “Successful Export Marketing Management: Some Empirical

Evidence”, International Marketing Review, vol. 6, no. 4, pp. 41-57.
Madsen, T. K. 1998, “Executive Insights: Managerial Judgments of Export Performance”,

Marandu, E. E. 1995, “Impact of Export Promotion on Export Performance: A Tanzanian

Study”, Journal of Global Marketing, vol. 9, no. 1/2, pp. 9-39.
March, J. G. & Shapira, Z. 1987, “Managerial Perspectives on Risk and Risk Taking”,
Management Science, vol. 33, no. 11, pp. 1404-1418.
McAuley, A. 1993, “The Perceived Usefulness of Export Information sources”, European

Journal of Marketing, vol. 27, no. 10, pp. 52-64.
Menard, S. W. 2002, Longitudinal Research, 2 edn, Sage, Newbury Park CA.
Menon, A. & Varadarajan, P. R. 1992, “A Model of Marketing Knowledge Use Within

Firms”, Journal of Marketing, vol. 56, no. 4, p. 53.
Miesenböck, K. J. 1987, “Small Business and Exporting: A Literature Review”,

International Small Business Journal, vol. 6, no. 2, pp. 42-61.
Milliken, F. J. 1987, “3 Types of Perceived Uncertainty About the Environment - State,

Effect, and Response Uncertainty”, Academy of Management Review, vol. 12, no. 1, pp.
133-143.
260
References
Mintzberg, H. 1989, Mintzberg on Management: Inside Our Strange World of

Organizations, Collier Macmillan, London.
Mohan-Neill, S. I. 1995, “The Influence of Firm's Age and Size on its Environmental
Scanning Activities”, Journal of Small Business Management, vol. 33, no. 4, pp. 10-21.
Moini, A. H. 1995, “An Inquiry into Successful Exporting: An Empirical Investigation

Using a Three-Stage Model”, Journal of Small Business Management, vol. 33, no. 3, pp. 9-
25.
Moini, A. H. 1997, “Barriers Inhibiting Export Performance of Small and Medium-Sized

Manufacturing Firms”, Journal of Global Marketing, vol. 10, no. 4, pp. 67-93.
Naidu, G. M. & Prasad, K. V. 1994, “Predictors of Export Strategy and Performance of

Small- and Medium-Sized Firms”, Journal of Business Research, vol. 31, pp. 107-115.
Nakos, G., Brouthers, L. E., & Brouthers, K. D. 1998, “The Impact of Firm and Managerial
Characteristics on Small and Medium-Sized Greek Firms' Export Performance”, Journal of
Global Marketing, vol. 11, no. 4, pp. 23-47.
Naman, J. L. & Slevin, D. P. 1993, “Entrepreneurship and the Concept of Fit: A Model and
Empirical Tests”, Strategic Management Journal, vol. 14, pp. 137-153.
Narver, J. C. & Slater, S. F. 1990, “The Effect of a Market Orientation on Business

Profitability”, Journal of Marketing, vol. 54, no. October, pp. 20-35.
Neal, L. 1990, “The Dutch and English East India Companies Compared: Evidence From
the Stock and Foreign Exchange Markets,” in The Rise of Merchant Empires. Long
Distance Trade in the Early Moderne World, 1350-1750, J. D. Tracy, ed., Cambridge
University Press, New York, pp. 195-223.
Netemeyer, R. G., Boles, J. S., McKee, D. O., & McMurrian, R. 1997, “An Investigation
Into the Antecedents of Organizational Citizenship Behaviors in a Personal Selling
Context”, Journal of Marketing, vol. 61, no. July, pp. 85-98.
Nijssen, E. J. & Douglas, S. P. 1999, “Gathering and Using Information for the Selection of
Trading Partners”, European Journal of Marketing, vol. 33, no. 1/2, pp. 143-162.
Nonaka, I. 1994, “A Dynamic Theory of Organizational Knowledge Creation”,

Organization Science, vol. 5, no. 1, pp. 14-37.
Nunnally, J. C. 1978, Psychometric Theory, 2 edn, McGraw-Hill, New York.
261
Pattillo, C. & Soderbom, M. 2000, Managerial Risk Attitudes and Firm Performance in
Ghanaian Manufacturing: An Empirical Analysis Based on Experimental Data, Centre for
the Study of African Economies, Oxford, WPS/00/17.
Pfeffer, J. & Salancik, G. R. 1978, The External Control of Organizations : A Resource

Dependence Perspective, Harper & Row, New York.
Postma, Th. J. B. M. & Zwart, P. S. 1998, “Een Strategieontwikkelingsmodel voor Kleinere

Bedrijven”, Maandblad voor Accountancy en Bedrijfseconomie, vol. 72, no. 6, pp. 306-316.
Powell, T. 1992, “Organizational Alignment as Competitive Advantage”, Strategic

Management Journal, vol. 13, no. 2, pp. 119-134.
Prahalad, C. K. & Hamel, G. 1990, “The Core Competence of the Corporation”, Harvard
Business Review, vol. 68, no. 3, pp. 79-91.
Reid, S. D. 1981, “The Decision-Maker and Export Entry and Expansion”, Journal of
Reid, S. D. 1984, “Information Acquisition and Export Entry Decisions in Small Firms”,
Journal of Business Research, vol. 12, pp. 141-157.
Rigdon, E. E. 1996, “CFI Versus RMSEA: A Comparison of Two Fit Indexes for Structural
Equation Modeling”, Structural Equation Modeling, vol. 3, no. 4, pp. 369-379.
Risseeuw, P. A. & Thurik, A. R. 2003, Handboek Ondernemers & Adviseurs: Management

en Economie van Midden- en Kleinbedrijf, 1st edn, Kluwer, Deventer.
Robertson, C. & Chetty, S. K. 2000, “A Contingency-Based Approach to Understanding

Export Performance”, International Business Review, vol. 9, no. 2, pp. 211-235.
Rose, G. M. & Shoham, A. 2002, “Export Performance and Market Orientation

Establishing an Empirical Link”, Journal of Business Research, vol. 55, no. 3, pp. 217-225.
Roux, E. 1987, “Managers' Attitudes Toward Risk Among Determinants of Export Entry of
Small- and Medium-Sized Firms,” in Managing export entry and expansion, P. J. Rosson,
ed., Praeger, New York, pp. 95-110.
Rubin, D. B. 1987, Multiple Imputations for Non-Response in Surveys, Wiley, New York.
Samiee, S. & Walters, P. G. P. 1990, “Influence of Firm Size on Export Planning and
Performance”, Journal of Business Research, vol. 20, pp. 235-248.
262
References
Samiee, S. & Walters, P. G. P. 1991, “Segmenting corporate exporting activities: sporadic

versus regular exporters”, Journal of the Academy of Management Science, vol. 19, no. 2,
pp. 93-104.
Schafer, J. L. 1997, Analysis of Incomplete Multivariate Data, Chapman & Hall, London.
Schafer, W. D. 1999, “An Overview of Meta-Analysis”, Measurement & Evaluation in

Counseling & Development, vol. 32, no. 1, pp. 43-61.
Seringhaus, F. H. R. 1986, “The Role of Information Assistance in Small Firms' Export

Involvement”, International Small Business Journal, vol. 5 , no. 2, pp. 26-35.
Seringhaus, F. H. R. 1993, “Comparative Marketing Behaviour of Canadian and Austrian

High-Tech Exporters”, Management International Review, vol. 33, no. 3, pp. 247-269.
Sharma, S. C. 1996, Applied Multivariate Techniques, John Wiley, New York.
Shoham, A. 1996, “Marketing-Mix Standardization: Determinants of Export Performance”,

Journal of Global Marketing, vol. 10, no. 2, pp. 53-73.
Shoham, A. 1998, “Export Performance: A Conceptualization and Empirical Assessment”,

Shoham, A. & Kropp, F. 1998, “Explaining International Performance: Marketing Mix,

Planning, and Their Interaction”, Marketing Intelligence and Planning, vol. 16, no. 2, pp.
114-123.
Shoham, A. 1999, “Bounded Rationality, Planning, Standardization of International

Strategy, and Export Performance: A Structural Model Examination”, Journal of
Shoham, A. 2000, “Firm Orientations: Do the Five Orientations Affect Export

Performance?”, Journal of Global Marketing, vol. 14, no. 3, pp. 31-47.
Singh, J. V. 1986, “Performance, Slack, and Risk Taking in Organizational Decision

Making”, Academy of Management Journal, vol. 29, no. 3, pp. 562-585.
Slater, S. F. & Narver, J. C. 1994, “Market Orientation, Customer Value, and Superior
Performance”, Business Horizons, vol. 37, no. 2, pp. 22-28.
263
Slater, S. F. & Narver, J. C. 2000, “The Positive Effect of a Market Orientation on Business
Profitability: A Balanced Replication”, Journal of Business Research, vol. 48, no. 1, pp. 69-
73.
Snow, C. C. & Hrebiniak, L. G. 1980, “Strategy, Distinctive Competence, and

Organizational Performance”, Administrative Science Quarterly, vol. 25, no. 2, p. 317.
Souchon, A. L. & Diamantopoulos, A. 1996, “A Conceptual Framework of Export

Marketing Information Use: Key Issues and Research Propositions”, Journal of
Souchon, A. L. & Diamantopoulos, A. 1997, “Use and Non-Use of Export Information:

Some Preliminary Insights into Antecedents and Impact on Export Performance”, Journal
of Marketing Management, vol. 13, pp. 135-151.
Souchon, A. L. & Diamantopoulos, A. 1999, “Export Information Acquisition Modes:

Measure Development and Validation”, International Marketing Review, vol. 16, no. 2, pp.
143-168.
Stearns, P. N. & Langer, W. L. 2001, The Encyclopedia of World History, 6th edn, James
Clarke & Co, Cambridge.
Steengaard, N. 1990, “The Growth and Composition of the Long-Distance Trade of

England and the Dutch Republic Before 1750,” in The rise of Merchant Empires. Long
Distance Trade in the Early Modern World, 1350-1750, J. D. Tracy, ed., Cambridge
University Press, New York, pp. 102-152.
Steenkamp, J. B. E. M. & Trijp, v. H. C. M. 1991, “The Use of LISREL in Validating

Marketing Constructs”, International Journal of research in Marketing, vol. 8, no. 4, pp.
283-299.
Stewart, D. B. & McAuley, A. 2000, “Congruence of Domestic and Export Marketing

Strategies- An Empirical Investigation of its Performance Implications”, International
Marketing Review, vol. 17, no. 6, pp. 563-586.
Stump, R. L., Athaide, G. A., & Axinn, C. N. 1998, “The Contingent Effect of the
Dimensions of Export Commitment on Exporting Financial Performance: An Empirical
Examination”, Journal of Global Marketing, vol. 12, no. 1, pp. 7-25.
Styles, C. & Amber, T. 1994, “Successful Export Practice: The UK Experience”,

International Marketing Review, vol. 11, no. 6, pp. 23-47.
264
References
Styles, C. 1998, “Export Performance Measures in Australia and the United Kingdom”,
Thirkell, P. C. & Dau, R. 1998, “Export Performance: Success Determinants for New
Zealand Manufacturing Exporters”, European Journal of Marketing, vol. 32, no. 9/10, pp.
813-829.
Toften, K. & Olsen, S. O. 2003, “Export Market Information Use, Organizational

Knowledge, and Firm Performance”, International Marketing Review, vol. 20, no. 1, pp.
95-110.
Tracy, J. D. 1990, The Rise of Merchant Empires. Long Distance Trade in the Early
Modern World, 1350-1750, Cambridge University Press, New York.
Trimpop, R. M. 1994, The Psychology of Risk Taking Behavior, North-Holland,

Amsterdam.
Usunier, J.-C. 2000, Marketing Across Cultures, Financial Times/ Prentice Hall, Harlow.
Wagner, J. 1995, “Export, Firm Size, and Firm Dynamics”, Small Business Economics, vol.
7, pp. 29-39.
Wallerstein, I. M. 1974, The Modern World-System, 1, Capitalist Agriculture and the

Origins of the European World-Economy in the Sixteenth Century edn, Academic Press,
New York.
Wallerstein, I. M. 1980, The Modern World-System, 2, Mercantilism and the Consolidation

of the European World-Economy, 1600-1750 edn, Academic Press, New York.
Walters, P. G. P. 1993, “Patterns of Formal Planning and Performance in U.S. Exporting

Firms”, Management International Review, vol. 33, no. 1, pp. 43-63.
Walters, P. G. P. 1996, “International Market Information Infusion: Data Acquisition

Behaviour in Norwegian Exporters”, International Journal of Information Management,
vol. 16, no. 6, pp. 437-444.
Willenborg, G. B. W. 2001, An Integrated Conceptual Model of Cooperative Consumer

Relationships in Services: Development and Test, Labyrint, Capelle a/d IJssel.
Wolf, F. M. 1986, Quantitative Methods for Research Synthesis, Sage, Beverly Hills CA.
265
Wolff, J. A. & Pett, T. L. 2000, “Internationalization of Small Firms: An Examination of

Export Competitive Patterns, Firm Size, and Export Performance”, Journal of Small
Business Management, vol. 38, no. 2, pp. 34-47.
Wood, V. R. & Robertson, K. R. 1997, “Strategic Orientation and Export Success: An

Empirical Study”, International Marketing Review, vol. 14, no. 6, pp. 424-444.
Wright, M. & Ashill, N. 1998, “A Contingency Model of Marketing Information”,

European Journal of Marketing, vol. 32, no. 1/2, pp. 125-144.
Yeoh, P. L. & Jeong, I. 1995, “Contingency Relationships Between Entrepreneurship,

Export Channel Structure and Environment”, European Journal of Marketing, vol. 29, no.
8, pp. 95-115.
Yeoh, P. L. 2000, “Information Acquisition Activities: A study of Global Start-Up

Exporting Companies”, Journal of International Marketing, vol. 8, no. 3, pp. 36-60.
Yip, G. S. 1989, “Global Strategy - In a World of Nations”, Sloan Management Review,

vol. 31, no. 1, pp. 29-41.
Zou, S., Taylor, C. R., & Osland, G. E. 1998, “The EXPERF Scale: A Cross-National
Generalized Export Performance Measure”, Journal of International Marketing, vol. 6, no.
3, pp. 37-58.
Zou, S. & Stan, S. 1998, “The Determinants of Export Performance: A Review of the
Empirical Literature between 1987 and 1997”, International Marketing Review, vol. 15, no.
5, pp. 333-356.
Zwart, P. S. 1991, Het Ondernemingsplan in het MKB: Een Hoeksteen van de

Bedrijfsvoering, Stenfert Kroese, Leiden.
Zwick, W. R. & Velicer, W. F. 1986, “Comparison of Five Rules for Determining the
Number of Components to Retain”, Psychological Bulletin, vol. 99, no. 3, p. 432.
266
Samenvatting
(Summary in Dutch)
Het exporterende MKB in Europa
Al eeuwenlang worden in Europa goederen (en tegenwoordig ook steeds meer
diensten) internationaal verhandeld. Eén van de oudste manieren om over de
grenzen van landen heen zaken te doen is exporteren. Al in de vijftiende eeuw
ontstond een open wereldeconomie, die vele Europese landen achtereenvolgens
roem en rijkdom bracht. In het bijzonder Nederland en Engeland profiteerden van
dit internationale netwerk van markten (denk aan de Engelse en Hollandse handel
met de (verre) Oost). Hoewel er in de loop van de jaren meer manieren van
internationalisering zijn ontstaan is het belang van export voor Europese bedrijven
nog steeds niet te onderschatten. Deze manier om de buitenlandse markt te
betreden is vooral van belang voor het Midden- en Kleinbedrijf (MKB) gezien de
lage kosten en het relatief lage risico dat exporteren met zich mee brengt,
vergeleken met bijvoorbeeld het opstarten van een joint venture of het opstarten
van een eigen vestiging in het buitenland. Het is dan ook interessant te
onderzoeken welke factoren het exportsucces van deze ondernemingen positief dan
wel negatief beïnvloeden. Hieruit volgt de doelstelling van dit proefschrift: het
verklaren van de exportprestatie van Europese exporterende MKB-bedrijven.
Theoretische achtergrond
In hoofdstuk 2, het theoretische hoofdstuk van deze studie, wordt een conceptueel
raamwerk ontwikkeld dat als basis dient voor de empirische hoofdstukken.
Teneinde de factoren die de exportprestatie beïnvloeden te kunnen benoemen en in
relatie tot elkaar te kunnen plaatsen is de bestaande literatuur over exportsucces
geanalyseerd. Als eerste zijn reeds gepubliceerde reviews geanalyseerd om tot een
classificatie van de determinanten van exportprestatie te komen. Deze exercitie
heeft tot de categorieën ‘omgeving’, ‘onderneming’, ‘manager’ en
‘exportactiviteiten’ geleid. Een verbijzondering is aangebracht in de categorie
‘manager’, aangezien het MKB de focus is van deze studie. Immers, de manager is
267
het centrale punt van de MKB-onderneming en verschilt wezenlijk van het

managementteam zoals we dat tegenkomen bij grote bedrijven. Daarom is een
onderscheid gemaakt tussen zowel objectieve en subjectieve karakteristieken van
de manager.
Als tweede zijn, gebruik makend van een gestructureerde meta-analyse, 43

empirische onderzoeken op het gebied van exportprestatie, gepubliceerd in de
periode 1989-2003, samengebracht. Over de operationalisering van exportprestatie
valt te concluderen dat de meest gebruikte maatstaven exportverkopen en
exportintensiteit zijn. Wel is er een ontwikkeling gaande naar het gebruik van ook
subjectieve maatstaven van succes en het gebruik van meerdimensionale
maatstaven. Deze meta-analyse resulteerde uiteindelijk in een integraal raamwerk
met alle specifieke variabelen (ondergebracht in de respectievelijke categorieën),
die volgens de literatuur exportprestatie kunnen verklaren. Daarnaast zijn alle
onderlinge relaties tussen deze determinanten bepaald, evenals hun directe en /of
indirecte invloed op exportsucces. Belangrijk is ook de conclusie dat slechts weinig
empirische studies de determinanten van exportsucces integraal benaderen; vaak
wordt slechts een deel van alle mogelijke variabelen opgenomen of wordt slechts
één (of enkele) categorie(ën) met exportsucces in relatie gebracht. Bovendien
wordt in veel gevallen alleen de directe invloed van de variabele op exportprestatie
onderzocht; de mogelijke indirecte relaties blijven buiten beschouwing. Het
integrale exportsuccesmodel is het startpunt voor de empirische verklaring van de
exportprestatie van de Europese MKB-ondernemingen in deze studie.
Methodologie
Om het theoretische raamwerk op zijn verklarende waarde te beoordelen, zijn
panelgegevens gebruikt die verzameld zijn door de Interstratos (International
Strategic Orientation of Small- and Mediumsized Enterprises) onderzoeksgroep.
Deze dataset is ontstaan uit een jaarlijks uitgestuurde schriftelijke enquête die
ingevuld is door industriele MKB-ondernemingen met minder dan 500
personeelsleden in zeven Europese landen; te weten België, Nederland, Oostenrijk,
Zwitserland, Zweden, Noorwegen en Finland. De gebruikte paneldata bestaat uit
gegevens van 1125 ondernemingen, die in de jaren 1991, 1993 en 1995
geparticipeerd hebben in de enquête (zie hoofdstuk 3).
268
Samenvatting (Summary in Dutch)
Aangezien vooral de prestatievariabelen een hoge mate van non-respons ten toon
spreidden, zijn in hoofdstuk 4 met behulp van het programma NORM drie
volledige datasets geïmputeerd, waarbij NORM een Bayesiaanse aanpak hanteert
gebaseerd op de a-priori verdeling van de variabelen. Deze meervoudige imputatie
van missende waarden heeft als voordeel boven een enkelvoudige imputatie, dat de
onzekerheid rondom de geïmputeerde missende waarden wordt meegenomen.
Hiertoe wordt één dataset gebruikt om het model vast te stellen en vervolgens te
schatten. De andere twee datasets worden ook gebruikt om het model te schatten.
Door de significantie, richting en grootte van de parameters van de drie modellen
met elkaar te vergelijken wordt de robuustheid van het gevonden model
vastgesteld. De gegevens zijn verder gevalideerd en geanalyseerd met behulp van
SPSS en Structural Equations Modeling (Lisrel 8.30), waarbij een tweestaps-
aanpak. In de eerste stap worden de meetmodellen van de latente variabelen
geschat en vastgezet, waarna in stap twee het structurele model wordt geschat.
Empirische toets integrale exportsuccesmodel

In hoofdstuk 5 wordt het geoperationaliseerde integrale exportsucces model
empirisch getoetst voor één jaar, te weten 1995. Exportprestatie wordt hierbij
gemeten als ‘exportverkopen’ en als ‘exportintensiteit’. Voor beide maatstaven
wordt het model afzonderlijk geschat, gehoor gevend aan de discussie in de
literatuur dat beide prestatiemaatstaven niet een zelfde onderliggende schaal
reflecteren, maar twee verschillende (en soms conflicterende) maatstaven van
succes zijn. Ondanks deze legitieme redenatie, geven de empirische resultaten van
het integrale model aan dat er weinig verschillen zijn tussen het model met
exportverkopen of exportintensiteit als afhankelijke variabele. De resultaten maken
duidelijk dat exportprestatie met name wordt beïnvloed door omgevingsvariabelen
(veranderingen op de aanbodsmarkt en de nationale en internationale
verkoopmarkt), ondernemingsvariabelen (ondernemingsgrootte, de bekwaamheden
van de onderneming en de exportervaring die een onderneming heeft) en
exportactiviteiten (het verzamelen van formele en informele informatie op de
buitenlandse markt en het samenwerken met buitenlandse partners).
Empirische toets informatieverzamelingsmodel

In de resultaten van het statische integrale model werd een significante invloed van
het verzamelen van informatie geconstateerd. Recent zijn meerdere publicaties
269
verschenen die het belang van informatie in een internationale setting benadrukken,
daarbij voortbouwend op het welbekende marktoriëntatieconcept. De literatuur
benadrukt daarbij ook de rol van de attitude van de manager in dit
informatiegedrag; een rol die nog wordt versterkt in een MKB-bedrijf. In hoofdstuk
6 zijn daarom de relaties tussen de risicoattitudes van de manager (ten aanzien van
veranderingen en planning), het informatieverzamelingsgedrag van de MKB-
onderneming en de exportprestatie zowel statisch (gebruikmakend van de 1995
data) en longitudinaal (gebruikmakend van de 1991, 1993 en 1995 data)
geanalyseerd. Het informatieverzamelingsgedrag is gespecificeerd naar een tweetal
dimensies:
1 Waar de informatie wordt verzameld, dat wil zeggen op de binnen- of
buitenlandse markt, en
2 Het type informatie dat wordt verzameld, dat wil zeggen (1) formele versus
informele informatie, of (2) export research, assistance of intelligence
informatie.
Een eerste opvallend resultaat is dat de MKB-ondernemingen in het panel slechts

een klein deel van alle mogelijke informatiebronnen gebruikt, waarbij men meer
vertrouwt op formele (of research) informatie van binnenlandse
informatiebronnen, terwijl op de buitenlandse markt de intelligence bronnen meer
worden geconsulteerd. Daarnaast negeert een groot gedeelte van het panel alle
informatiebronnen. Uit de resultaten blijkt vervolgens dat:
(1) Het verzamelen van exportinformatie de exportprestatie bevordert, in het
bijzonder informatie van buitenlandse informatiebronnen;
(2) De risicoaversie van de manager ten aanzien van veranderingen de
hoeveelheid informatie die wordt verzameld reduceert, terwijl een voorkeur
voor planning een manager aanspoort meer informatie te verzamelen;
(3) De risicoaversie ten aanzien van veranderingen heeft een negatief effect op
exportprestatie, maar alleen maar op exportverkopen, terwijl de houding
van de manager ten aanzien van planning geen relatie heeft met
exportprestatie;
(4) De informatie die in het voorgaande jaar wordt verzameld de
exportprestatie in het daaropvolgende jaar stimuleert;
(5) De exportprestatie in het voorgaande jaar de hoeveelheid informatie die
gezocht wordt in het daaropvolgende jaar stimuleert.
270
Samenvatting (Summary in Dutch)
Conclusies
De belangrijkste conclusies uit dit proefschrift kunnen als volgt worden
samengevat:
(1) In de exportliteratuur bestaat nog geen consensus over de operationalisering
van exportprestatie, noch over de determinanten van exportprestatie, noch
over de soort verbanden tussen de determinanten en exportprestatie (direct
en/ of indirect), noch over de causaliteit van de relaties tussen de
determinanten. Dit proefschrift heeft een aanzet gegeven de verschillende
conceptuele en empirische literatuur op het gebied van exportprestatie
samen te brengen en zo tot een overzicht te komen dat kan dienen als basis
voor verder empirisch onderzoek.
(2) Het gebruik van meervoudige imputatiemethoden bij het omgaan van
missende waarden heeft voordelen boven het gebruik van een enkelvoudige
imputatie, omdat de hiermee de robuustheid van de oplossing op basis van
één geïmputeerde dataset gevalideerd kan worden met de andere
geïmputeerde datasets. Bij het opzetten van een longitudinaal onderzoek
moet met name rekening worden gehouden met missende waarden. Ook
blijken veel respondenten moeite te hebben met het vrijgeven van
objectieve prestatieindicatoren, wat op te lossen is door tevens gebruik te
maken van meer subjectieve maatstaven.
(3) Indien een manager de exportprestatie van zijn of haar MKB-onderneming
wil verbeteren, kan hij of zij ertoe over gaan meer informatie te verzamelen
bij buitenlandse informatieverschaffers, en/of meer samen te werken met
buitenlandse partners. Een internationale marktoriëntatie is dus voordelig.
Ook kan hij bepaalde ondernemingskarakteristieken op het gebied van
bekwaanmheden (inclujsief exportervaring) proberen te verbeteren. De
andere significante omgevingsdeterminanten liggen veel meer buiten de
controle van de manager en met hij/zij als onbeheersbaar beschouwen. Dat
geldt ook voor de ondernemingskarakteristiek ‘ondernemingsgrootte’. Het
is goed dat de manager er zich van bewust is dat zijn/haar risicohouding
invloed heeft op zowel het informatieverzamelingsgedrag als de
exportprestatie van de onderneming. Hiermee wordt meteen de speciale
plaats van de eigenaar-manager in het geval van een MKB-onderneming
benadrukt. Ook is gebleken dat een hogere exportprestatie een manager
271
ertoe aanzet meer informatie te verzamelen in de toekomst, terwijl ook de

verzamelde informatie nog een uitgesteld, positief effect heeft op de
toekomstige exportprestatie.
272

Thesis

Enviado por

Dados do documento

Descrição original:

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Thesis

Enviado por

Direitos autorais:

Formatos disponíveis

The Export Performance of

Printed by: Offsetdrukkerij Ridderprint B.V., Ridderkerk

© 2003, J.A. Voerman

The Export Performance of European SMEs

ter verkrijging van het doctoraat in de

Jeannet Alina Voerman

geboren op 11 juli 1972

Prof. dr. P.S. Zwart

Prof. dr. T.H.A. Bijmolt

I am thankful to all my colleagues, for their support and interest in both my

I am also indebted to my student-assistants Sietske Vlutters, and Karla Visser for

Chapter 2 Conceptualizing Export Performance and its Determinants 9

4 Operationalizing the Integral Export Performance Model 93

Chapter 6 The Information Behavior of Exporting SMEs 169

Appendix 1 The INTERSTRATOS Survey 215

Samenvatting (Summary in Dutch) 267

1.1 European Export

regulations, the preferential treatment of own merchants over foreign merchants,

1.3 Success Factors

1.4 European Exporting SMEs: The Present

Moreover, I concentrate on small- and medium-sized enterprises (SMEs). SMEs

countries (Observatory of European SMEs 2002). For SMEs, ‘exporting’ is an

1.5 Export Performance of European SMEs

In chapter 2, this theoretical base is developed by examining both existing reviews

To test this theoretical framework empirically, I use data collected by the

Subsequently, I validate the constructs operationalized in Chapter 4, and

probably is the greatest single determinant of a firm’s export success”. In addition,

Finally, in chapter 7 the conclusions and managerial implications are stated.

2.1 Conceptual Work

In the eighties and especially the nineties, internationalization increasingly became

2.1.1 Miesenbock (1987)

2.1.2 Madsen (1987)

2.1.3 Aaby & Slater (1989)

Firm characteristics • Market selection

In addition, some research recommendations are made. Concerning the research

2.1.3.1 Chetty & Hamilton (1993)

2.1.3.2 Styles & Amber (1994)

2.1.4 Gemünden (1991)

Characteristics of the Firm

Home Country Managers Activities Export Success

Export Market Characteristics

Figure 2-2 Theoretical Framework (Gemünden 1991, p.38).

Gemünden (1991) also develops a conceptual framework, synthesizing all variables

2.1.5 Zou & Stan (1998)

Price determination (0/0/11)

Management Attitudes and Perceptions

Management Characteristics Industry Characteristics

Firm’s Characteristics and Competencies Foreign Market Characteristics

2.1.6 Leonidou, Katsikeas & Piercy (1998)

should include the organizational parameters, the external environment, and

2.1.7 Leonidou, Katsikeas & Samiee (2002)

2.1.8 Extending the Existing Reviews

determinants, while ‘management characteristics’ are not categorized as such11.

and ‘managerial characteristics’. In this, the competencies are divided according to

As the concept of environment is still researched very fragmentarily, I use an

2.2 Literature Synthesis: Method & Descriptives

2.2.1.1 Literature Search

2.2.1.2 Data Evaluation

2.2.1.3 Data Analysis

A more appropriate method is the vote-counting method, as this implies measuring

2.2.1.4 Interpretation: descriptives

Table 2-5 Studies Reviewed and Descriptives.

Concerning the geographical focus, the United States is over-represented with 16