Escolar Documentos
Profissional Documentos
Cultura Documentos
European SMEs
Liane Voerman
Published by: LABYRINT PUBLICATION
P.O. Box 334
2950 AH Alblasserdam
The Netherlands
Tel. 0180-463962
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
Proefschrift
door
Beoordelingscommissie:
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 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
References 249
1
The Export Performance of European SMEs
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
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.
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
The Export Performance of European SMEs
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.
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
Considering the above, the export performance of European exporting SMEs and
the factors determining this performance are at the heart of this thesis.
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
The Export Performance of European SMEs
measures. By integrating the results of both the existing reviews and the empirical
studies an overall export performance framework results.
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).
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
7
The Export Performance of European SMEs
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.
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
The Export Performance of European SMEs
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.
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.”
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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.”
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
The Export Performance of European SMEs
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
Figure 2-1 A General Model for Assessing Export Performance and Variables (Aaby & Slater 1989, p.9.)
14
Conceptualizing Export Performance and its Determinants
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.”
15
The Export Performance of European SMEs
16
Conceptualizing Export Performance and its Determinants
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.
17
The Export Performance of European SMEs
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
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.”
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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).
22
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.
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
Conceptualizing Export Performance and its Determinants
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.
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
The Export Performance of European SMEs
& 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
Conceptualizing Export Performance and its Determinants
16
Meta-analysis is defined as a statistical analysis of results from individual studies
(Cooper 1998).
27
The Export Performance of European SMEs
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.
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.
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
Conceptualizing Export Performance and its Determinants
29
The Export Performance of European SMEs
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).
30
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
Variables
# Method Dependent variable Independent variables
Export Performance (EPF) EV FC MC EB
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
Conceptualizing Export Performance and its Determinants
Variables
# Method Dependent variable Independent variables
Export Performance (EPF) EV FC MC EB
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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.
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
The Export Performance of European SMEs
characteristics, and the export behavior. As this is a quite extensive piece, this part
will be treated in a separate section.
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
Conceptualizing Export Performance and its Determinants
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.
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 Export Performance of European SMEs
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.
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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.
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
The Export Performance of European SMEs
46
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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).
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
Conceptualizing Export Performance and its Determinants
Thirkell & Dau 1998), and to gain a better competitive market position
(Holzmüller & Kasper 1991; Holzmüller & Stöttinger 1996)36.
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 Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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.
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
The Export Performance of European SMEs
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.
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
Conceptualizing Export Performance and its Determinants
53
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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).
44
Both Evangelista (1994), and Stump, Athaide & Axinn (1998) distinguish between
attitudinal commitment (“an enduring positive disposition of management towards
56
Conceptualizing Export Performance and its Determinants
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.
57
The Export Performance of European SMEs
(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.
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.
60
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.
62
Conceptualizing Export Performance and its Determinants
ratio47. Besides, the market position strategy selected (i.e. market leader versus
follower) does not affect export sales or export ratio (Ito & Pucik 1993).
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
The Export Performance of European SMEs
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).
49
‘Commitment of resources’ is sometimes used as a dependent variable, distinguishing
between experimental, committed, and active exporters based on their degree of
64
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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.
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
Conceptualizing Export Performance and its Determinants
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.
69
The Export Performance of European SMEs
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.
‘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).
56
This factor encompasses control of channels, advertising, innovation in marketing
techniques, and inter-state agreements.
70
Conceptualizing Export Performance and its Determinants
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.
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
The Export Performance of European SMEs
72
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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.
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 Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
77
The Export Performance of European SMEs
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.
78
Conceptualizing Export Performance and its Determinants
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
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.
79
The Export Performance of European SMEs
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
Conceptualizing Export Performance and its Determinants
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
The Export Performance of European SMEs
82
Chapter 3
The Interstratos Data Set
3.1 Introduction
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
The Export Performance of European SMEs
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.
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 (‘hout- en meubelen’), 34/35 (‘metaalproducten/machine-
industrie’), and 36 (‘elektrotechnische industrie’), respectively.
85
The Export Performance of European SMEs
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.
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
The INTERSTRATOS Data Set
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
87
The Export Performance of European SMEs
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
The INTERSTRATOS Data Set
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.
67
So, the true net sample for 1994 is 3885 – 647 = 3238 enterprises.
89
The Export Performance of European SMEs
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
The INTERSTRATOS Data Set
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
The Export Performance of European SMEs
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.
93
The Export Performance of European SMEs
• 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.
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
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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.
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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.
77
.16, .25, and .23 for 1991, 1993, and 1995, respectively.
101
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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’.
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.
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
Operationalizing the Integral Export Performance Model
‘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).
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
The Export Performance of European SMEs
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).
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
Operationalizing the Integral Export Performance Model
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.
Year Valid Min Max Mean SE mean Std. Dev. Skewness
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
Valid = valid response, Min = minimum value entered, Max = maximum value entered, SE mean =
standard error mean, Std. Dev. = standard deviation
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
92
Some variables are dropped from further analysis for their lack of theoretical linkages
with the stimuli literature.
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
The Export Performance of European SMEs
Literature
Our model
Background variables
Background variables
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
112
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
attitude towards change’ (RTC), ‘Risk attitude towards planning’ (RTP), and
‘Government Restrictions’ (GR) (see Table 4-15)95.
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
Operationalizing the Integral Export Performance Model
115
The Export Performance of European SMEs
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
116
Operationalizing the Integral Export Performance Model
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
Valid = valid response, Min = minimum value entered, Max = maximum value entered, SE mean =
standard error mean, Std. Dev. = standard deviation
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.
117
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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.
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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
122
Operationalizing the Integral Export Performance Model
123
The Export Performance of European SMEs
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.
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.
106
This section is largely based on Rubin (1987).
124
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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.
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
The Export Performance of European SMEs
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.
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
Operationalizing the Integral Export Performance Model
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.
131
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
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
The Export Performance of European SMEs
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
Operationalizing the Integral Export Performance Model
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
The Export Performance of European SMEs
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).
137
The Export Performance of European SMEs
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
Subjective Managerial
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
138
The Integral Export Performance Model - Estimated
139
The Export Performance of European SMEs
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
The Integral Export Performance Model - Estimated
X = Λ xξ + δ (5.1)
Y = Λ yη + ε ( 5.2)
Where:
141
The Export Performance of European SMEs
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.
142
The Integral Export Performance Model - Estimated
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
The Export Performance of European SMEs
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.
121
Case number 847, 984, and 986, respectively.
144
The Integral Export Performance Model - Estimated
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
The Export Performance of European SMEs
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
The Integral Export Performance Model - Estimated
147
The Export Performance of European SMEs
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
The Integral Export Performance Model - Estimated
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.
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
The Export Performance of European SMEs
150
The Integral Export Performance Model - Estimated
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.
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
The Export Performance of European SMEs
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.
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
The Integral Export Performance Model - Estimated
On a whole, all formative models comply with the tests they have been subjected
to, and are included in the structural model.
153
The Export Performance of European SMEs
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).
154
The Integral Export Performance Model - Estimated
Firm Characteristics:
- 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
Subjective Managerial
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
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
The Export Performance of European SMEs
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 .
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
The Integral Export Performance Model - Estimated
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.
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
The Export Performance of European SMEs
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.
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
The Integral Export Performance Model - Estimated
159
The Export Performance of European SMEs
160
The Integral Export Performance Model - Estimated
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
The Export Performance of European SMEs
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.
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
The Integral Export Performance Model - Estimated
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).
163
The Export Performance of European SMEs
164
The Integral Export Performance Model - Estimated
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.
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
The Export Performance of European SMEs
performance136. Table 5-4 shows these total effects on export ratio and export sales
(t-values in parentheses).
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
The Integral Export Performance Model - Estimated
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
The Export Performance of European SMEs
five in all three sets for both export performance measures, but do have significant
and large total effects in all sets.
168
Chapter 6
The Information Behavior of
Exporting SMEs
169
The Export Performance of European SMEs
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
171
The Export Performance of European SMEs
172
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
& 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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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.
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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.
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
The Export Performance of European SMEs
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.
6.3.1 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
The Information Behavior of Exporting SMEs
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?
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
The Export Performance of European SMEs
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
(2 items, Risk2) VEE = 0.43
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)
Risk towards changes α = 0.57 CR = 0.58
(2 items, Risk1) VEE = 0.41
λav = 0.64
λlow = 0.56
Risk towards planning α = 0.54 CR = 0.58
(2 items, Risk2) VEE = 0.43
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
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
188
The Information Behavior of Exporting SMEs
H3 (-)
Risk Avoiding Export
Propensity Performance
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
formal and informal information;
(3) Export performance measured with export ratio, information categorized as
export market research, assistance, and intelligence information;
(4) Export performance measured with export ratio, 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
The Export Performance of European SMEs
190
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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.
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
The Export Performance of European SMEs
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.
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
The Information Behavior of Exporting SMEs
195
The Export Performance of European SMEs
196
The Information Behavior of Exporting SMEs
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).
197
The Export Performance of European SMEs
198
The Information Behavior of Exporting SMEs
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
The Export Performance of European SMEs
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
The Information Behavior of Exporting SMEs
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.
201
The Export Performance of European SMEs
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.
203
The Export Performance of European SMEs
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.
204
Conclusions and Discussion
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.
205
The Export Performance of European SMEs
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).
206
Conclusions and Discussion
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.
207
The Export Performance of European SMEs
208
Conclusions and Discussion
209
The Export Performance of European SMEs
aware of the effect of his or her risk attitude on his or her behavior and on
the performance of a company.
210
Conclusions and Discussion
211
The Export Performance of European SMEs
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.
212
Conclusions and Discussion
framework and this conceptual model should be the base for the data collection.
Then, the problems of operationalization can be minimized.
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
The Export Performance of European SMEs
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
The Export Performance of European SMEs
216
Appendix 1
217
The Export Performance of European SMEs
218
Appendix 1
219
The Export Performance of European SMEs
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.
Reliability Set of Reliability Measurement Model
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
(2 items, Supply) VEE = 0.67
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
(2 items, Capital) VEE = 0.52
λav = 0.72
λlow = 0.71
3. CFA Competencies χ2SB = 63.36 (p = 0.00, df = 14)
GFI = 0.98
CFI = 0.97
RMSEA = 0.056 (p = 0.22)
3a. Marketing Capability α = 0.73 CR = 0.73
(2 items, Mrktng) VEE = 0.58
λav = 0.76
λlow = 0.69
3b. Product Capability α = 0.77 CR = 0.76
(2 items, Prodcap) VEE = 0.63
221
The Export Performance of European SMEs
222
Appendix 2
223
The Export Performance of European SMEs
224
Appendix 2
225
The Export Performance of European SMEs
226
Appendix 3
Results Integral Export
Performance Model on
Imputed Data Set 3
Table A3-4 Validity reflective models – set 3.
Reliability Set of Reliability Latent Measurement Model
Measurement Construct Goodness-of-Fit Statistics
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
(2 items, Supply) VEE = 0.59
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
(2 items, Capital) VEE = 0.53
λav = 0.73
λlow = 0.71
3. CFA Competencies χ2SB = 61.91 (p = 0.00, df = 14)
GFI = 0.98
CFI = 0.97
RMSEA = 0.055 (p = 0.25)
3a. Marketing Capability α = 0.73 CR = 0.73
(2 items, Mrktng) VEE = 0.58
λav = 0.76
λlow = 0.69
3b. Product Capability α = 0.78 CR = 0.78
(2 items, Prodcap) VEE = 0.65
227
The Export Performance of European SMEs
228
Appendix 3
229
The Export Performance of European SMEs
230
Appendix 3
231
The Export Performance of European SMEs
232
Appendix 4
Results Information Model on
Imputed Data Set 2
Table A4-1 Validity reflective models – set 2.
Reliability Set of Reliability Measurement Model
Measurement Latent Goodness-of-Fit Statistics
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)
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 = 0.59 CR = 0.61
(2 items, Risk2) VEE = 0.43
λav = 0.65
λlow = 0.54
CFA Attitude 1993 χ2 = 1.25 (p = 0.26, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .015 (p = 0.71)
Risk towards changes α = 0.56 CR = 0.58
(2 items, Risk1) VEE = 0.41
λav = 0.64
λlow = 0.55
Risk towards planning α = 0.55 CR = 0.60
(2 items, Risk2) VEE = 0.45
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
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.
233
The Export Performance of European SMEs
234
Appendix 4
235
The Export Performance of European SMEs
236
Appendix 4
237
The Export Performance of European SMEs
238
Appendix 4
239
The Export Performance of European SMEs
240
Appendix 5
Results Information Model on
Imputed Data Set 3
Table A5-1 Validity reflective models – set 3.
Reliability Set of Reliability Latent Measurement Model
Measurement Construct Goodness-of-Fit Statistics
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)
Risk towards changes α = 0.63 CR = 0.63
(2 items, Risk1) VEE = 0.46
λav = 0.68
λlow = 0.68
Risk towards planning α = 0.58 CR = 0.60
(2 items, Risk2) VEE = 0.44
λav = 0.66
λlow = 0.55
CFA Attitude 1993 χ2 = 3.27 (p = 0.07, df = 1)
GFI = 1.00
CFI = 1.00
RMSEA = .045 (p = 0.45)
Risk towards changes α = 0.57 CR = 0.58
(2 items, Risk1) VEE = 0.41
λav = 0.63
λlow = 0.53
Risk towards planning α = 0.57 CR = 0.62
(2 items, Risk2) VEE = 0.42
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
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.
241
The Export Performance of European SMEs
242
References
243
The Export Performance of European SMEs
244
References
245
The Export Performance of European SMEs
246
References
247
The Export Performance of European SMEs
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.
Albaum, G., Strandskov, J., & Duerr, E. 2002, International Marketing and Export
Management, 4 edn, Pearson Education Limited, Edinburgh Gate.
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.
Bagozzi, R. P. & Yi, Y. 1988, “On the Evaluation of Structural Equation Models”,
Academy of Marketing Science, vol. 16 , no. 1, pp. 74-94.
249
The Export Performance of European SMEs
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.
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.
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.
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.
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., 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.
251
The Export Performance of European SMEs
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.
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.
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.
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.
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.
253
The Export Performance of European SMEs
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.
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.
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.
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
The Export Performance of European SMEs
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.
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.
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.
Johanson, J. & Vahlne, J. E. 1977, “The internationalization process of the firm - A model
of knowledge development and increasing foreign market commitments”, Journal of
International Business Studies, vol. 8, no. 1, pp. 23-32.
256
References
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.
Kaiser, H. F. 1970, “A Second Generation Little Jiffy”, Psychometrika, vol. 35, no.
December, pp. 401-415.
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.
257
The Export Performance of European SMEs
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. & Armstrong, G. 2004, Principles of Marketing, 10 edn, Pearson/ Prentice Hall,
Upper Sadle River.
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.
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.
259
The Export Performance of European SMEs
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.
March, J. G. & Shapira, Z. 1987, “Managerial Perspectives on Risk and Risk Taking”,
Management Science, vol. 33, no. 11, pp. 1404-1418.
260
References
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.
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.
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.
261
The Export Performance of European SMEs
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.
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
International Business Studies, vol. 12, no. 3, pp. 101-112.
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.
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
Schafer, J. L. 1997, Analysis of Incomplete Multivariate Data, Chapman & Hall, London.
Slater, S. F. & Narver, J. C. 1994, “Market Orientation, Customer Value, and Superior
Performance”, Business Horizons, vol. 37, no. 2, pp. 22-28.
263
The Export Performance of European SMEs
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.
Stearns, P. N. & Langer, W. L. 2001, The Encyclopedia of World History, 6th edn, James
Clarke & Co, Cambridge.
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.
264
References
Styles, C. 1998, “Export Performance Measures in Australia and the United Kingdom”,
Journal of International Marketing, vol. 6, no. 3, pp. 12-36.
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.
Tracy, J. D. 1990, The Rise of Merchant Empires. Long Distance Trade in the Early
Modern World, 1350-1750, Cambridge University Press, New York.
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.
Wolf, F. M. 1986, Quantitative Methods for Research Synthesis, Sage, Beverly Hills CA.
265
The Export Performance of European SMEs
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.
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
The Export Performance of European SMEs
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.
269
The Export Performance of European SMEs
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.
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
The Export Performance of European SMEs
272