Content uploaded by Bob Walrave
Author content
All content in this area was uploaded by Bob Walrave
Content may be subject to copyright.
Exploitation and exploration dynamics in
recessionary times
Bob Walrave
A catalogue record is available from the Eindhoven University of Technology
library
ISBN: 978-90-386-3173-8
Walrave, Bob
Exploitation and exploration dynamics in recessionary times
Eindhoven: Eindhoven University of Technology, 2012.
Keywords: exploitation-exploration, recession, recovery, management-board
interaction, success trap, suppression process, system dynamics.
Eindhoven University of Technology
School of Industrial Engineering
http://www.tue.nl
Beta Ph.D. Theses Series D152
Cover design: Jeroen Frissen & Bob Walrave
Printed by: Proefschriftmaken.nl | | BOXPress BV
© 2012, Bob Walrave
Exploitation and exploration dynamics in
recessionary times
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de
Technische Universiteit Eindhoven, op gezag van de
rector magnificus, prof.dr.ir. C.J. van Duijn, voor een
commissie aangewezen door het College voor
Promoties in het openbaar te verdedigen
op dinsdag 12 juni 2012 om 14.00 uur
door
Bob Walrave
geboren te Roosendaal en Nispen
Dit proefschrift is goedgekeurd door de promotor:
prof.dr A.G.L. Romme
Copromotor:
dr.ir. K.E. van Oorschot
“I should have learned to play the guitar –
I should have learned to play them drums”
Mark Knopfler – Money For Nothing, 1984.
VII
Acknowledgements
This doctoral thesis would have never been completed without the support,
ideas, and advice of a lot of people. That includes friends, family, and
colleagues. As such, I would like to take the opportunity to express my
appreciation to all of them.
Firstly, I want to sincerely thank my supervisors: My promoter Georges
Romme and co-promoter Kim van Oorschot. Georges, thank you for giving
me the opportunity to start pursuing a PhD. You gave me all the insights I
needed to develop myself and this work to the fullest. I truly appreciate your
enduring commitment to this project. Thank you. Kim, thank you for all the
(personal) advice and for guiding me through the world called ‘systems
thinking’. I have always enjoyed our valuable discussions on models,
settings, and other ‘Vensim’ related stuff (besides all the other topics that
were discussed of course – like thinking of catchy titles for our papers).
Georges, Kim, I believe that the three of us form a well-balanced
(ambidextrous) team that already has, and still will, achieve great things.
Moreover, I would also like to thank Joop van der Meij (Vlisco) for
starting this project and Michel Frequin (Gamma Holding) for providing the
required funding to actually execute the endeavor. Although our ‘Vitesse’
project was terminated prematurely, I did find this period immensely
interesting and joyful. Moreover, it was in this period that the very
foundation of this dissertation was shaped.
Further gratitude goes to Fred Langerak for being intensively involved
with the ‘bear-bull’ paper. Fred, your extensive knowledge of the ‘field’ and
exceptionally sharp judgment gives us an edge when it comes to publishing
this work in a top-journal in the near future. Furthermore, thank you for
trusting my academic talent to such a degree that I was allowed to join the
ITEM group as an assistant professor.
Furthermore, I feel that I owe a ‘big thank you’ to Victor Gilsing, the
man who introduced me to the scientific discipline (by mentoring me
though my master’s thesis project). Victor, thank you for planting the ‘seeds
of interest’ and to enable them to grow by convincing Georges that I was the
man for the job. (And, Georges, I hope I lived up to the high expectations
that were set by Victor?!)
VIII
I would also like to thank all my (ex-)colleagues and friends of the ITEM
group, but Marion van den Heuvel and Bianca van Broekhoven deserve a
special mention. Marion and Bianca, thank you for the countless
conversations that we had in the secretary’s office. But, truly, your combined
efforts concerning all kinds of administrative tasks were, and are, of
invaluable help to the whole group.
Of course, I would like to thank my friends and family too. Therefore, to
all those who have supported me over the last few years, I say thank you.
Notably, my roommate Lydie, thank you for the good times we had while we
were sharing an office in the infamous M-corridor. Having you as a friend
around made my PhD research so much more enjoyable.
A special note of appreciation goes to the ‘es gibt nicht zu schnell’ team
members: Jeroen Schepers, Jeroen Frissen, Sharon Dolmans, and Agnieszka
Krzyżaniak, who, besides endless discussions on work-related issues,
stimulated the evolution of my competences in the geography of Europe and
South America. Also, Frissen, thank you for designing the (in my humble
opinion excellent) cover. I truly feel blessed with such friendships.
Here, I want to especially thank my mom and dad for letting me explore
and exploit. Explore and exploit, on the journey that shaped me into the
person that I am today. Thank you for your unconditional trust, love and
care. Ruud and Angeline, you always allowed me to pursue my dreams, and
simply put, without this kind of support I would not have accomplished the
things I have done so far (being it climbing volcanoes in Mexico or writing
this doctoral thesis).
I would also like to thank my brother, Björn, for serving as a true
brother: My dearest friend and great source of inspiration. In those low-
motivation moments, but also when the blood glucose levels were down, you
were, and are, there to help. I also thank Anna, my sister-in-law for patiently
allowing never-ending discussions on research topics that I pursued with
Björn – during holidays, during my birthday parties, his birthday parties or
even your birthday parties. Thank you for your understanding and also for
taking such great care over my brother.
Most of all, however, I want to thank Agnieszka, my soon wife-to-be.
Thank you for your unconditional support. Thank you for sharing your life
with me. Thank you for saying ‘yes’ at the summit of the Bishorn. Thank you
for making my life so much more worthwhile. Words can simply not express
my gratitude.
Bob Walrave. Helmond, 2012.
IX
Table of Contents
Chapter 1. Introduction 1
1.1 Overview of the dissertation 2
1.2 On the methodology 8
Chapter 2. Fighting the bear and riding the bull 11
Exploitation and exploration during times of recession and recovery 11
2.1 Introduction 12
2.2 Hypotheses 20
2.3 Method 27
2.4 Results 36
2.5 Discussion and conclusions 41
2.6 Conclusion 47
Chapter 3. Getting trapped in the 49
suppression of exploration
A simulation model 49
3.1 Introduction 50
3.2 Theoretical background 52
3.3 Method 57
3.4 Model description 58
3.5 History replicating and diverging simulations 68
3.6 A case narrative of the suppression process 71
3.7 Discussion 76
3.8 Conclusion 81
Chapter 4. Counteracting the suppression process 83
A simulation model 83
4.1 Introduction 84
4.2 Theoretical background 86
4.3 Method 91
4.4 Experiments as strategic interventions 96
X
4.5 Results 102
4.6 Discussion and implications 111
4.7 Conclusion 117
Chapter 5. Conclusions 119
5.1 Summary of the findings and theoretical implications 120
5.2 Study 1 – Fighting the bear and riding the bull:
Exploitation and exploration during times of recession
and recovery 120
5.3 Study 2 – Getting trapped in the suppression of
exploration: A simulation model 122
5.4 Study 3 – Counteracting the success trap:
A simulation model 124
5.5 Synergy among – and reflection on – the findings 126
5.6 Practical discussion 129
5.7 Limitations and future research 131
5.8 Closing remarks 133
References 135
Appendix I 149
Appendix II 153
AII.1 Model overview 155
AII.2 Model description 155
AII.3 Model settings and sensitivity 162
AII.4 Deterministic versus stochastic 171
AII.5 Adjustments required for experimentation 172
AII.6 Sensitivity of the experiments 180
Summary 183
About the author 189
Chapter 1
Introduction
Schumpeter, often claimed to be the first author who wrote about the
importance of innovation and exploration, stated that: “The process of
Creative Destruction is the essential fact about capitalism. […] It is not [price]
competition which counts but the competition from the new commodity, the
new technology, the new source of supply, the new type of organization. […]
Competition which commands a decisive cost or quality advantage and
which strikes not at the margins of the profits and the outputs of existing
firms but at their foundations and their very lives” (Schumpeter, 1942, pp.
83–84). Although this sounds compelling, firms also have to make a living
by exploiting the offerings in their current portfolio. In this respect,
exploration is an expensive endeavor, characterized by long lead times,
which needs to be financed by the short-term returns generated by
exploitative investments. This implies that firms have to be able to
simultaneously pursuit exploitation and exploration. This specific capability
has proven to be profitable but difficult to develop, and is thus important for
firms.
Since March’s (1991) seminal work, the terms exploitation and
exploration have taken center stage in organization science. Exploitation
captures things like “refine, choice, production, efficiency, selection,
implementation, and execution”, while exploration is characterized by
“search, variation, risk taking, experimentation, play, flexibility, discovery,
and innovation” (March, 1991, p.71). Despite the simplicity of the idea, the
exploitation-exploration framework has developed into an important and
repeatedly applied lens for explaining organizational behavior and
performance. Gupta et al. (2006), and more recently Lavie et al. (2010),
provide exhaustive reviews of the exploitation-exploration literature. The
2
INTRODUCTION
large number of studies discussed in these two reviews underline that
considerable progress in this specific research area has been made since the
early nineties.
Many different hypotheses have been constructed and accepted, relating
exploitation and exploration with organizational performance in intricate
manners (e.g., Auh and Menguc, 2005; Jansen et al., 2006; Uotila et al.,
2009). For instance, Uotila et al. (2009) uncovers an inverted U-shaped
relationship between the exploitation-exploration ratio and firm
performance, which is positively moderated by R&D intensity. Moreover,
certain patterns, or archetypes, of organizational behavior are discussed in
the exploitation-exploration literature. For example, many company failures
can be explained by the self-reinforcing nature of the ‘success trap’ – the
process in which exploitative investments are increasingly preferred over
explorative investments, often due to early successes with exploitation
(Levinthal and March, 1993; Walrave et al., 2011). Think, for instance, about
Toys “R” Us, the American Locomotive Company, Polaroid, and many others
(e.g., Tripsas and Gavetti, 2000; Walrave et al., 2011; Wiersema, 2002). All
these firms got caught by the success trap and started suppressing essential
explorative investments (Auh and Menguc, 2005; Levinthal and March,
1993). In this respect, much has been written about the importance of the
exploitation-exploration combination for firm performance, but getting it
‘right’ seems to be particularly difficult for many firms. This observation is
the raison d'être for conducting this study.
1.1 Overview of the dissertation
This dissertation aims to investigate how firms should orchestrate their
exploitation-exploration activities in recessionary times. One can think here of
recessionary times caused by economic contractions, such as the financial
crisis that started around 2007 (focus of Chapter 2). But one could also think
of organizational crisis situations caused by shifts in the environmental
context (possibly, but not necessarily, an economic contraction), to which top
management did not (sufficiently) respond (focus of Chapters 3 and 4). The
time spent by firms in such contexts is at least as high as 35 per cent
CHAPTER 1
3
(Claessens et al., 2009; Terrones et al., 2009). Three studies are conducted
that shed light on the main research question.
1.1.1 Study 1 – Fighting the bear and riding the bull:
Exploitation and exploration in times of recession
and recovery
The literature provides compelling empirical evidence that keeping a balance
between exploitation and exploration enhances profitability (He and Wong,
2004; Jansen et al., 2006; Uotila et al., 2009). Moreover, research has
focused on environmental influences, like dynamism (reflecting the rate of
change and the instability of the external environment), as a moderator
between exploitation-exploration investments levels and firm performance.
For instance, Jansen et al. (2006) provide evidence for the moderating effect
of environmental dynamism on the relationship between exploitation-
exploration investment levels and firm performance. Moreover, Lin et al.
(2007) find that in an uncertain environment an ‘ambidextrous’ formation
of alliance partners enhances firm performance.
Although recessions and recoveries can be described in terms of, for
instance, dynamism and competitiveness, the (low) amount of
environmental munificence makes a recessionary context very different from
what has been studied till date (e.g., Jansen et al., 2006). As such, the best
course of action concerning the balance between exploitation and exploration
investments during times of crisis is not understood very well. In this
respect, it is not clear how recessionary times influence the most profitable
exploitation-exploration ratio. Nevertheless, past economic recessions and
recoveries have demonstrated that both periods can have a significant
influence on firm performance and that some firms are affected more than
others. This makes it paramount, for theory as well as for practice, that a
better understanding is developed concerning the relationship between firm
performance and the exploitation-exploration ratio in the context of
recessions and recoveries. As such, the first empirical study of this
dissertation, in Chapter 2, investigates what the relationship is between the
4
INTRODUCTION
exploitation-exploration ratio and firm performance in times of recession and
recovery (i.e., bear and bull).
Firm data from the most recent global economic recession and recovery
are analyzed to explore exploitation-exploration performance implications.
By applying system GMM estimation on a panel dataset, consisting of 105
firms in the IT industry over the period 2007-2010, we aim to open up this
black box. The main theoretical contribution of this chapter lies in
identifying the change in the most profitable exploitation-exploration balance
given shifts in the macroeconomic conditions; that is, this chapter explores
whether the absolute and/or relative importance of exploitation-exploration
changes over time within the same industrial context. From a managerial
perspective, our findings provide important insights in how to effectively
‘fight the bear’ and ‘ride the bull’.
1.1.2 Study 2 – Getting trapped in the suppression
process: A simulation model
Investing more in exploration during times of (economic) decline is a
counter-intuitive strategy; at least one that is highly different from what
many firms actually do in crisis-like situations. Many firms intuitively
overemphasize exploitation efforts while facing environmental turbulence.
Often, a cost reduction strategy is adopted, with damage control as the main
goal (Helfat et al., 2007; Tushman et al., 2004; Wiersema, 2002). This
frequently reinforces the declining trend in performance, triggering a
further focus on exploitation (Levinthal and March, 1993). Think for instance
of Toys “R” Us where, as the result of environmental change, a pronounced
focus on exploitation became a catalyst for even more exploitative activities
(Wiersema, 2002). Although this behavior has been given a specific name
(i.e., the success trap), there is no real underlying rationale, or process theory
(cf. Van de Ven, 2007), explaining this trap. It is merely known that a
primary focus on exploitation in some cases works self-reinforcing, but it is
not known how firms get trapped in the success trap.
Previous studies attribute the success trap to managerial incompetence
and/or myopia. For instance, the study by Tripsas and Gavetti (2000)
CHAPTER 1
5
outlines the decline of Polaroid due to management’s misunderstanding of
the world. However, some management teams appear to adequately
recognize the exploration need as a result of environmental change, while
still not being able to bring about the required strategic (and organizational)
change (Helfat et al., 2007; Wiersema, 2002). As such, the second study of
this dissertation, reported in Chapter 3, investigates how it is possible that top
managers enhance their firm’s exploitation focus, when the need to explore in
response to environmental change is evident.
The main theoretical contribution is a process theory, underlying the
success trap, at the managerial level. This process theory is coined the
‘suppression process’. A case study of a firm that got stuck in the success
trap is conducted and a simulation model is developed that replicates the
firm’s behavior in terms of exploitation-exploration investments. More
specifically, we draw on system dynamics modeling to develop the
‘suppression process’ theory. The process theory developed in this chapter
describes and explains how the interplay between top managers, board
members, and exploitation–exploration activities can trap the firm in the
suppression of exploration.
1.1.3 Study 3 – Counteracting the success trap: A
simulation model
Not much is known about how to counteract the suppression process (or the
success trap) once initiated. Some suggestions can be distilled from the
literature. For instance, Levinthal and March (1993, p.106) indicate that “the
trap can be broken by rapid upward adjustment of aspirations or by false
feedback as to the high value of exploration”. Other studies merely suggest
that drastic turnarounds are required to escape the success trap (Helfat et al.,
2007; Walrave et al., 2011). Building on the formal model developed in
Chapter 3, the third study investigates several possible escape paths from the
suppression process. As such, Chapter 4 deals with the question how to
counteract the suppression process characterized by underinvestment in
exploration.
6
INTRODUCTION
The main contribution to the exploitation-exploration literature and
practitioners alike is the identification of mechanisms aimed at restoring a
profitable exploitation-exploration balance. The findings provide a first
insight into whether the suppression process can be counteracted. In this
respect, this chapter contributes to the emerging body of research on the
scenarios and implications of (in)correctly (re)balancing exploitation and
exploration activities. By means of experimentation, we identify several
critical conditions required to break up the self-reinforcing workings of the
suppression process and, as such, avoid the success trap.
1.1.4 Overall contribution to the literature
The overarching theoretical framework for the three studies included in this
dissertation is the resource-based view of the firm (Barney, 1991). The
resource-based view assumes that a firm achieves a competitive advantage by
owning and developing proprietary assets, while simultaneously possessing
a superior ability to make good use of those assets (Barney, 1991). In this
respect, the underlying mechanism to develop and maintain such valuable,
rare, imperfectly imitable, and non-substitutable resources is to utilize
available resources to conduct exploitative and explorative (learning)
activities. More specifically, exploitative activities might for instance be used
to enhance existing assets (be it products or processes), making it more
difficult for competitors to imitate them. Explorative activities, on the other
hand, could be employed to disrupt the value and rareness of the resources
of the competition. In this respect, a firm enjoying a sustained competitive
advantage is always susceptible to major shifts in the competitive structure
that can nullify their advantage (Barney, 1991). Developing and maintaining
a sustainable competitive advantage thus depends on how the portfolio of
exploitative and explorative activities is organized (March, 1991).
The concepts of exploitation and exploration have been studied in a
wide variety of literatures. For instance, from an organizational learning
perspective (e.g., Levinthal and March, 1993; March, 1991), an organizational
design perspective (e.g., Tushman and O’Reilly, 1996), and an
organizational adaptation perspective (e.g., Brown and Eisenhardt, 1997).
CHAPTER 1
7
Nevertheless, as outlined above, exploitation and exploration are in this
dissertation mainly conceptualized as organizational learning activities
(Levinthal and March, 1993). As such, the original definitions and
conceptualizations of March are utilized (March, 1991), in line with other
recent work in this area (e.g., Uotila et al., 2009).
Several studies started investigating the performance implications of the
so-called ‘ambidexterity-hypothesis’ (i.e., the organizational ability to
simultaneously explore and exploit). While some studies reported that
pursuing either exploitation or exploration results in improved performance
(e.g., Ebben and Johnson, 2005), other studies found that conducting both
activities simultaneously significantly improves performance (e.g., He and
Wong, 2004; Uotila et al., 2009). The first study (found in chapter 2)
contributes directly to this line of research on performance implications in
the exploitation-exploration literature by providing further empirical
evidence for the ambidexterity hypothesis. Moreover, recent research has
started to analyze potential moderating effects (like environmental
dynamism and competitiveness) on the relationship between exploitation-
exploration and firm performance (e.g., Auh and Menguc, 2005; Jansen et
al., 2006; Uotila et al., 2009). The first study also contributes to this line of
research by investigating the moderating effects of a recession and recovery
context on the exploitation-exploration firm performance link.
Whereas the first study investigates the exploitation-exploration
dilemma on the organizational level, the second and third studies focus at
the managerial level. In this respect, these two studies consider exploitation-
exploration from a ‘dynamic managerial capability’ point of view (Helfat et
al., 2007).
Numerous antecedents of successful (simultaneous) execution of
exploitation and exploration have been identified (e.g., Gibson and
Birkinshaw, 2004; Hoang and Rothaermel, 2010; Simsek et al., 2009); an
important antecedent arising from this previous work is top management
(e.g., Hambrick and Mason, 1984; Jansen et al., 2008; Uotila et al., 2009).
Top managers play a decisive role in establishing a supportive context for
managing the tension between exploitation and exploration (Gibson and
Birkinshaw, 2004; Jansen et al., 2008; Smith and Tushman, 2005). Yet,
8
INTRODUCTION
management often fails to develop a profitable exploitation-exploration
balance. The underlying cause for this failure is argued to be organizational
path-dependence, due to top managers’ myopic tendencies, which limit their
ability to adapt the strategic direction when required (e.g., Hannan and
Freeman, 1984; Tushman et al., 2004). This is likely to result in the often
described success trap (Levinthal and March, 1993; March, 1991). Study 3 and
4 (found in chapter 3 and 4) contribute to this specific stream within the
exploitation-exploration literature in two ways: first, by developing a more
fine-grained process theory underlying the success trap, called the
‘suppression mechanism’; second, by identifying possible mechanisms and
interventions that are instrumental in counteracting the suppression
mechanism.
1.2 On the methodology
This dissertation aims to advance the exploitation-exploration research field
in the context of recessionary times. The three studies in this doctoral thesis
share a longitudinal research approach. More specifically, the first study
analyzes a panel dataset to distill results; the last two studies draw on system
dynamics modeling (in combination with a case study) to infer conclusions.
As such, this dissertation contributes to the growing body of longitudinal
research within the exploitation-exploration research domain (e.g., Hoang
and Rothaermel, 2010; Lavie and Rosenkopf, 2006; Lin et al., 2007).
Different research techniques are utilized in order to provide answers to
the research questions previously introduced. That is, the first study aims to
answer a what question, that is: “what are the antecedents or consequences
of the issue?” (Van de Ven, 2007, p.145). The second and third study deal
with how questions: “How does the issue emerge, develop, grow, or
terminate over time?” (Van de Ven, 2007, p.145). These two basic question
types require different methodologies. What questions are generally
answered by developing variance models, utilizing statistical analyzes to
explain discrepancies in certain outcomes. As such, dictated by the dynamic
panel data set, the first study adopts a system GMM methodology (Arellano
and Bover, 1995). How questions, on the other hand, require a process
CHAPTER 1
9
model based on, for instance, a story or historical narrative. As such, in the
second study we adopt a so-called history-friendly simulation approach
(Malerba et al., 1999). History-friendly models “aim to capture, in stylized
form, qualitative and ‘appreciative’ theories about the mechanisms and
factors affecting […] change”, (Malerba et al., 1999, p.3). The actual formal
model is developed by means of system dynamics (Sterman, 2000), drawing
on the case narrative of a Dutch international firm. Subsequently, the third
study utilizes the dynamic model developed in the second study and further
exploits this by means of experimentation. This history-divergent modeling
approach provides the researcher with the means to systematically vary the
theoretically relevant variables, after which the impact on organizational
performance can be assessed (e.g., Malerba et al., 1999; Romme, 2004).
The ‘core’ chapters of this work (i.e., 2, 3, and 4) are presented as
separate research papers. This implies that these chapters can, in principle,
be read independently of each other. This also causes for some overlap to
exist between the three chapters (e.g., definitions and assumptions). The
remainder of this doctoral thesis is organized as follows. Chapter 2 analyzes
the moderating effect of recession and recovery on the link between
exploitation-exploration and firm performance. Subsequently, Chapter 3
investigates how it is possible that some top managers choose to enhance
their firm’s exploitative focus, even when the need to explore in response to
environmental change is evident. Chapter 4 subsequently explores
interventions necessary to restore the equilibrium between exploitative and
explorative activities and the firm’s environment, in order to prevent the
success trap from becoming a firm’s end state. Finally, in Chapter 5, the
results are summarized and final conclusions are drawn. The implications of
the three studies are also integrated in a practical discussion. Moreover,
limitations and suggestions for further research are given.
Chapter 2
Fighting the bear and riding the bull:
Exploitation and exploration during times
of recession and recovery
The benefits of balancing exploitation and exploration activities in non-
recessionary contexts are increasingly better understood. However, periods of
economic recession (and recovery) are a primary cause of organizational failure.
As such, there is a need to understand the moderating effects of times of extreme
economic turbulence (i.e., bear and bull) on the exploitation-exploration firm
performance relationship. We adopt a longitudinal research approach. By
applying system GMM estimation on a panel dataset of 105 firms in the IT
industry over the period 2007-2010, we find three results. An inverted U-shaped
relationship is established between the exploitation-exploration ratio and firm
performance, which is influenced, in terms of absolute outcomes, by the phase of
the business cycle (relatively more positive during the bull phase). Secondly, the
relative importance of balancing exploitation-exploration for firm performance
appears to change, when moving from the bear (more important) to bull phase
(less important). Finally, the optimal exploitation-exploration ratio for firm
performance changes, at large, toward more exploitation when moving from the
recession to the recovery phase. In this respect, the recession and recovery phases of
the business cycle provide significantly different contexts for managing the
exploitation-exploration ratio. This then constitutes our principal theoretical
contribution to the exploitation-exploration literature. Moreover, our findings
provide practical insights in how to ‘fight the bear’ and ‘ride the bull’.
12
FIGHTING THE BEAR AND RIDING THE BULL
2.1 Introduction
The recent global economic recession, which started in 2007 and lasted for
18 months, resulted in the collapse of large financial institutions (Hall et al.,
2010) and caused a significant yet unexpected contraction in demand,
employment levels, cash flows, and profits (Srinivasan et al., 2011;
Steenkamp and Fang, 2011). Such a state of affairs is also known as a ‘bear
market’ (Barsky and Long, 1990). However, from 2009 till (at least) the end
of 2010, many markets were recovering, investor confidence was being
restored, and the financial situation of the surviving firms was readily
improving. Such an upward market trend is often referred to as a ‘bull
market’ (Barsky and Long, 1990). In this respect, the terminology of bear
and bull markets is derived from the manner in which each animal attacks
its opponent: a bear will swipe downwards, while a bull will thrust its horns
upwards.
The aftermath of the recent economic recession and recovery makes
clear that some firms are affected more than others. For instance, Apple saw
only little downfall during the most recent global economic recession and
achieved a tremendous recovery afterwards. In this respect, Srinivasan et al.
(2011) observed that during the 2001 recession, 20 per cent of the firms that
were initially in the bottom quartile of performance statistics rose to the top
quartile. As such, these parts of the business cycle seem to have a profound
effect on (relative) firm performance, making it critical for management to
understand how to best oppose these strong exogenous forces (Deleersnyder
et al., 2004; Grewal and Tansuhaj, 2001). Moreover, the past decade has
seen several periods of economic upheaval and the proportion of time spent
by firms in such contexts is as high as 35 per cent (Claessens et al., 2009;
Grewal and Tansuhaj, 2001; Terrones et al., 2009). As such, there is a need
to understand the factors that lead to superior or inferior performance, in
both bear and bull markets of extreme economic upheaval (e.g., Rosenblatt
et al., 1993; Schmitt, 2010).
In general, firm performance largely depends on the ability to adapt to,
and exploit, changes in the business environment (Helfat et al., 2007;
CHAPTER 2
13
Hoang and Rothaermel, 2010; Teece et al., 1997). That is, firms should
maintain ecological fitness by reconfiguring their resource base to cope with
emerging threats and explore new opportunities, while simultaneously
exploiting existing resources (O’Reilly and Tushman, 2008; Simsek, 2009).
As such, companies possessing the ability to simultaneously build
exploitative and explorative knowledge may be more resilient to situations of
economic turmoil (cf. Raisch et al., 2009; Walrave et al., 2011). In this
respect, several empirical studies suggest a positive link between the
strategic division of exploitation-exploration and firm performance (e.g., He
and Wong, 2004; Jansen et al., 2006; Uotila et al., 2009). Moreover,
environmental influences, like competitiveness, dynamism, and R&D
intensity, are also known to affect the most profitable exploitation-
exploration distribution (Auh and Menguc, 2005; Jansen et al., 2006; Uotila
et al., 2009).
Although these findings provide a rough handhold (i.e., the need to
balance/emphasize exploitation and/or exploration given specific
environmental conditions), it is less well understood how bear and bull
contexts influence the effectiveness of these two types of organizational
learning. In other words, the moderating effect of the business cycle on the,
for firm performance, optimal exploitation-exploration balance has never
been investigated. Yet, such knowledge will enhance both our theoretical
understanding (e.g., does the absolute and/or relative importance of
exploitation-exploration change over time within the same industrial context)
and managerial practice in this area (e.g., how to effectively handle bear and
bull markets by means of exploitation and/or exploration).
In this paper we consider the bear and bull market of the most recent
business cycle to investigate their effects on the relationship between the
exploitation-exploration ratio and firm performance. A longitudinal research
approach is adopted, involving system generalized methods of moments
estimation on a panel data set of 105 firms in the information technology
industry over the period 2007-2010. Overall, our results indicate that the
implications – and management requirements – of the exploitation-
exploration ratio within the same industrial and competitive context strongly
depend on the phase of the business cycle. In this respect, we uncover that
14
FIGHTING THE BEAR AND RIDING THE BULL
periods of economic recession and recovery have a significantly different
impact on the relationship between the exploitation-exploration ratio and
firm performance. This constitutes our main theoretical contribution and
extends previous (cross-sectional) studies in this field (e.g., He and Wong,
2004; Jansen et al., 2006; Uotila et al., 2009).
In the next sections, we review the literature and develop hypotheses.
Then, the research method is described and the empirical findings are
presented. Finally, we discuss the theoretical contributions and managerial
implications of our findings, next to issues left for future research.
2.1.1 Theoretical background
2.1.1.1 On exploitation and exploration
Ever since March’s (1991) seminal article, the terms exploitation and
exploration have taken center stage in organization studies (e.g., Gupta et al.,
2006; Lavie et al., 2010). Exploitation helps a firm to reduce its knowledge
variety, increase its efficiency, enhance the fit with the current
environmental context, and therefore generate profits on the short run
(March, 1991). Exploitation, in a broad sense, captures things like “refine,
choice, production, efficiency, selection, implementation, and execution”
(March, 1991, p.71). As such, exploitation draws on learning processes that
aim to incrementally improve the existing knowledge base of the firm
(Levinthal and March, 1993). By contrast, exploration serves to gather and
develop knowledge that is different from the current knowledge base (Lavie
et al., 2010). Thus, exploration involves “a pursuit of new knowledge”
(Levinthal and March, 1993, p.105) and is therefore characterized by “search,
variation, risk taking, experimentation, play, flexibility, discovery, and
innovation” (March, 1991, p.71). Exploration enhances a firm’s future
adaptability by development of new knowledge and, thus, allows for
adjustment alongside changing environmental contexts (March, 1991).
Exploitation and exploration have been conceptualized in two distinct
manners (cf. Gupta et al., 2006; Lavie et al., 2010). That is, the literature has
treated the exploitation-exploration relation either as a zero sum game (thus
CHAPTER 2
15
as two ends of one continuum) (e.g., March, 1991; Uotila et al., 2009;
Walrave et al., 2011) or as two fundamentally different orthogonal aspects
(e.g., He and Wong, 2004; Katila and Ahuja, 2002; Rothaermel, 2001).
Although both exploration and exploitation are essential for survival and
prosperity, the ‘balancing act’ typically needs to be conducted with a limited
set of available resources. As such, an increase in exploration activities will
decrease the resources available for exploitation, and vice versa. As such, and
in line with March’s (1991) original characterization, we consider
exploitation-exploration as two ends of the same continuum, constrained by
a shared set of resources. That we consider exploitation and exploration as
activities aimed at organizational learning supports this choice (Auh and
Menguc, 2005; Gupta et al., 2006; Levinthal and March, 1993).
Despite the apparent differences between the two modes of
organizational learning they need to be conducted simultaneously. Since
exploitation and exploration require fundamentally different and often
competing learning acts, creating and maintaining a strategically sound
balance between the two is difficult (Jansen et al., 2008; Levinthal and
March, 1993; Walrave et al., 2011). Nevertheless, the organizational failure to
achieve a sound balance can have destructive consequences. On the one
hand, excessive exploration (at the cost of exploitation) can be extremely
costly as the outcomes will likely be realized in the distant future and the
short-term opportunities of exploitation are overlooked. Moreover, such an
organizational emphasis can result in the perilous ‘failure trap’ (cf. Levinthal
and March, 1993; March, 1991). On the other hand, a mere focus on
exploitation (at the cost of exploration) potentially results in short-term
profits but discourages long-term learning investments (thereby inhibiting
the development of a sustainable competitive advantage). This peculiar
situation is expected to result in the ‘success trap’ (cf. Levinthal and March,
1993), whether or not initiated through the ‘suppression process’ (cf.
Walrave et al., 2011).
Therefore, it should come as no surprise that recent empirical research
findings illustrate that a carefully orchestrated combination of exploitation
and exploration has a significant positive effect on firm performance (i.e.,
under the ‘normal’ course of events) (e.g., Auh and Menguc, 2005; He and
16
FIGHTING THE BEAR AND RIDING THE BULL
Wong, 2004; Jansen et al., 2006; Uotila et al., 2009). For instance, He and
Wong (2004) demonstrate that equal levels of exploitation and exploration
are required for a superior sales growth rate. Auh and Menguc (2005) show
that the costs associated with neglecting either exploitation or exploration
can negatively influence firm performance. Subsequent research further
developed the ‘ambidexterity hypothesis’, by abandoning the idea that equal
levels of exploitation and exploration are needed for superior performance.
For example, Jansen et al. (2006) find that the level of environmental
dynamism and competitiveness, which strongly varies between different
industries and markets, dictates the most profitable mix of exploitation-
exploration. Most recently, Uotila et al. (2009) show that the relationship
between exploitation-exploration and firm performance is characterized by
an inverted U-shaped relationship; and they demonstrate this relationship to
be moderated by the R&D intensity of the industry.
Furthermore, a large array of antecedents concerning the successful
(simultaneous) execution of exploitation and exploration have been
described (e.g., Gibson and Birkinshaw, 2004; Hoang and Rothaermel,
2010; Simsek et al., 2009). Nevertheless, scholars have long emphasized
that top management is crucial to firm outcomes (Hambrick and Mason,
1984). These actors play a decisive role in establishing a supportive context
for managing the tension between exploitation and exploration (Gibson and
Birkinshaw, 2004; Jansen et al., 2008; Smith and Tushman, 2005). Sidhu et
al. (2004), for instance, provide empirical evidence that managerial
intentions significantly influence an organization’s explorative orientation.
In this respect, top managers decide upon the processes, such that their firm
can both exploit and explore. For instance, top managers have the power to
prevent short-term performance pressures, salient to lower-level managers,
from taking over the need for more explorative knowledge development
(Adler et al., 1999). Company success, as such, resides to a large extent in
the capability of top management to sense external treats and opportunities
and subsequently strategically divide resources to both exploitative and
explorative learning (Christensen and Bower, 1996; Walrave et al., 2011;
Zollo and Winter, 2002).
CHAPTER 2
17
2.1.1.2 On exploitation and exploration and economic recessions and
recoveries
Economic recessions and recoveries are recurring events in the major world
economies (Srinivasan et al., 2011). Although there is little consensus as to
the reasons why economic recessions and recoveries arise, they are both
characterized by the co-movement of many macroeconomic indicators
(Grewal and Tansuhaj, 2001; Smart and Vertinsky, 1984). In this respect,
both bear and bull markets represent a ‘low probability, high impact’
situation that is likely to threaten organizational survival (Lee and Makhija,
2009; Smart and Vertinsky, 1984). Economic recessions and recoveries,
therefore present top management with a unique challenge. That is, top
managers need to strategically reconfigure their resources to address
emerging threats by exploring new opportunities and exploiting existing
openings (O’Reilly and Tushman, 2008; Simsek, 2009).
Economic recessions and recoveries are inherently linked to business
cycles (Steenkamp and Fang, 2011). On average, advanced economies went
through six complete business cycles of economic recession, recovery, and
expansion since the 1960’s (Claessens et al., 2009; Terrones et al., 2009).
Figure 2.1 illustrates the differences between these three phases. The
recession phase – or bear market – is the period between ‘Peak’ and ‘Through’
(Claessens et al., 2009; Terrones et al., 2009). We draw on the definition of
an economic recession provided by the National Bureau of Economic
Research (Hall et al., 2010, p.1), also adopted in other recent work (e.g.,
Srinivasan et al., 2011): “A period of falling economic activity spread across
the economy, lasting more than [six] months, normally visible in real GDP,
real income, employment, industrial production, and wholesale-retail sales.”
The recovery phase – or bull market – is the period between ‘Through’ and
‘Recovered’. The recovery phase ends when the output returns to the peak
level achieved just before the recession phase started. In this respect, an
economic recovery is defined as the inverse of the recession phase: “A period
of [rising] economic activity spread across the economy, lasting more than
[six] months, normally visible in real GDP, real income, employment,
industrial production, and wholesale-retail sales” (Hall et al., 2010, p.1).
18
FIGHTING THE BEAR AND RIDING THE BULL
Although bear and bull markets can occur at any period within a
business cycle (e.g., outside an economic recession and recovery context), the
terms ‘bear’ and ‘bull’ in this paper explicitly refer to periods of recession
and recovery associated with a context of substantial economic turmoil. The
end of the recovery phase denotes the start of the expansion phase, which is
the period from ‘Through’ till (a new) ‘Peak’.
Figure 2.1: Economic recession, Recovery, and Expansion.
Business cycles have received ample scholarly attention, for instance,
from a marketing, sales, organizational capability, and strategy perspective
(Deleersnyder et al., 2004; e.g., Grewal and Tansuhaj, 2001; Lamey et al.,
2007; Lee and Makhija, 2009; Steenkamp and Fang, 2011). Recent research
contributions in the field of exploitation-exploration provide some – albeit
indirect – insights into how to effectively manage a context of recession and
recovery (Jansen et al., 2006; Walrave et al., 2011). Given the unpredictability
of the occurrence of a bear market, demand typically falls faster than the
supply, causing an increase in the level of environmental competitiveness
(Steenkamp and Fang, 2011). Environmental competitiveness is defined as
“the extent to which external environments are characterized by intense
Peak
Through
Output levels
t0 t1 t2
Time
Recession
(bear market)
Recovery
(bull market)
Expansion
Recovered
CHAPTER 2
19
competition” (Jansen et al., 2006, p.1664). Jansen et al. (2006) found that in
highly competitive environments, those firms that are directed toward
exploitation outperform firms that are steered toward exploration.
Nevertheless, a bear market also increases the level of dynamism in a
market. Environmental dynamism is about the unpredictability of change as
well as the amount of change present in the environmental context (cf. Dess
and Beard, 1984). It is defined as “the rate of change and the degree of
instability of the environment” (Jansen et al., 2006, p.1664). During an
economic recession the environment can be considered highly volatile (Dess
and Beard, 1984; Jansen et al., 2006; Walrave et al., 2011) and, therefore,
dynamic. High levels of environmental dynamism are best fought with
exploration rather than exploitation (Auh and Menguc, 2005; Jansen et al.,
2006).
On the other hand, during the bull phase of the business cycle, demand
typically grows faster than supply, which is likely to cause a decrease in the
level of competitiveness. Lower levels of environmental competitiveness are
expected to require more exploration for optimal firm performance (Jansen
et al., 2006; Walrave et al., 2011). Nevertheless, a bull market also likely
decreases the level of dynamism in a market. That is, after the extremely
volatile period of the economic recession, the market is expected to grow
relatively stable in terms of, for instance, customer preferences and product
demand. This change in the environmental dynamism is best counteracted
with exploitation (Jansen et al., 2006; Walrave et al., 2011).
Concluding, in both bear and bull markets, firms need to focus on
exploitative as well as explorative knowledge building. More specifically,
firms with a more balanced exploitation-exploration ratio are likely to
outperform their ‘non-balanced’ counterparts, in both economic recessions
and recoveries. This suggestion is also in line with recent observations from
the corporate turnaround literature who acknowledge that swift
organizational decline (e.g., due to a recession) should be fought with
retrenchment in combination with repositioning (e.g., Schmitt, 2010).
Furthermore, this idea aligns with the cross-sectional findings by Uotila et
al. (2009), which suggest an inverted U-shaped relationship between
exploitation-exploration and firm performance.
20
FIGHTING THE BEAR AND RIDING THE BULL
Nonetheless, although former research informs us with a rather clear
indication of the nature of the relationship between exploitation-exploration
and firm performance (i.e., given our assumption that exploitation-
exploration are two ends on a continuum: an inverted U-shape), it is not
known if/how the different phases of the business cycle specifically
moderate this link. For example, is there a difference in the absolute
performance outcomes – given a certain exploitation-exploration ratio –
during the bear and bull phase? Does the relative importance of exploitation
and exploration evolve over time (cf. Raisch et al., 2009)? Does the phase of
the business cycle have an effect on the optimal balancing point (i.e. is there
a change in the most profitable vertex)? The next section serves to develop
hypotheses related to these questions.
2.2 Hypotheses
2.2.1 Bear vs bull market affects the absolute
performance outcome of exploitation-exploration
ratio
The first moderating effect of the business cycle concerns a likely difference
in absolute performance outcomes due to a given exploitation-exploration
ratio. More specifically, business cycles have a profound effect on supply and
demand. The bear and bull markets themselves are therefore also likely to
affect firm performance. More specifically, economic recessions increase
unemployment levels and, as such, decrease the purchasing power of
customers (e.g., Deleersnyder et al., 2004; Lamey et al., 2007; Srinivasan et
al., 2011; Steenkamp and Fang, 2011). Therefore, in a bear market,
customers are more price sensitive and risk-adverse than in a bull market
(Claessens et al., 2009; D’Aveni and MacMillan, 1990; Lamey et al., 2007;
Steenkamp and Fang, 2011). Customers delay purchasing decisions at this
point in time, in view of increasing uncertainty about future purchasing
power (D’Aveni and MacMillan, 1990; Srinivasan et al., 2011; Steenkamp
and Fang, 2011). As a consequence, the market demand for the firm’s output
CHAPTER 2
21
is lower (Block, 1979; Deleersnyder et al., 2004; Steenkamp and Fang, 2011).
This results in a (for a bear market typical) industry-wide contraction that
strongly reduces opportunities for firm growth and development (Srinivasan
et al., 2011; Steenkamp and Fang, 2011).
The end of a recession signals the start of economic recovery. As such,
employment levels increase and the purchasing power of the customers
gradually returns (Deleersnyder et al., 2004; Lamey et al., 2007). This
triggers a general rise in the economic conditions and it can be anticipated
that customers will return to the market. As such, this period in time is
characterized by an increasing amount of growth opportunities
(Deleersnyder et al., 2004; Lamey et al., 2007). This then results in a (for a
bull market typical) industry-wide expansion that strongly increases potential
for firm growth and development (Srinivasan et al., 2011; Steenkamp and
Fang, 2011).
In this respect, the bear and bull market differ in terms of their
environmental munificence (Dess and Beard, 1984). Environmental
munificence refers to “the extent to which the environment can support
sustained [organizational] growth” (Dess and Beard, 1984, p.55). Sales
growth, for instance, is a primary variable determining the amount of
environmental munificence (Dess and Beard, 1984; Hofer, 1975). Whereas a
bear market is characterized by a decrease in the level of environmental
munificence, a bull market is associated with an increase in the amount of
environmental munificence.
As such, we argue that the anticipated inverted U-shaped
relationship between the exploitation-exploration ratio and firm performance
is more positive in a bull market than in a bear market. Figure 2.2 depicted
the expected effect graphically. The figure implies that the entire graph (incl.
its optimum) for a recession is likely to be positioned higher than the graph
for the recovery (as in Figure 2.2).
Hypothesis 1 The exploitation-exploration ratio has a more positive effect on
firm performance in a bull market than in a bear market.
22
FIGHTING THE BEAR AND RIDING THE BULL
An exploitation-exploration ratio of 0 implies a complete focus on exploitation, while
a ratio of 1 implies an exclusive focus on exploration
Figure 2.2: Illustration of hypothesis 1.
2.2.2 Bear vs bull market affects the relative importance
of exploitation-exploration ratio
Bear and bull markets have different characteristics, as previously argued. As
such, besides the expected difference in absolute performance (i.e.,
hypothesis 1), the relative importance of the exploitation-exploration ratio is
likely to be different over the course of the two market types (cf. Raisch et al.,
2009). This then constitutes the second anticipated moderating effect of the
business cycle on the exploitation-exploration firm performance relationship.
Consider the 2001 bear market: 20 per cent of the firms that were
initially in the bottom quartile of performance statistics rose to the top
quartile in their respective markets, and more than 20 per cent in the top
quartile fell to the bottom quartile (Srinivasan et al., 2011). Interestingly, 70
per cent of the firms that increased performance in the bear market
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Exploitation-Exploration ratio
Effect on firm performance
Bear phase
Bull phase
CHAPTER 2
23
sustained those gains in the ensuing economic recovery, while fewer than 30
per cent of the firms that lost ground regained their positions (cf. Srinivasan
et al., 2011; Steenkamp and Fang, 2011). In this respect, especially recessions
appear to provide opportunities for accelerated firm growth; or for decline if
one fails to explore these opportunities (Srinivasan et al., 2011; Steenkamp
and Fang, 2011). This observation resonates with findings originating from
the organizational decline literature (cf. Porter and Harrigan, 1983;
Rosenblatt et al., 1993) and suggest that the relative importance of correctly
handing the exploitation-exploration ratio is larger in a bear market than in a
bull market.
These patterns of firm growth and decline can be explained by the fact
that customers delay purchasing decisions in recessions (Block, 1979;
Srinivasan et al., 2011). As argued earlier, in the bear market, this results in
an industry-wide contraction that causes a reduction of firm growth
prospects (Grewal and Tansuhaj, 2001). This decreased level of
environmental munificence makes for a severe environment selection
regime. Therefore, the organizational ability to satisfy consumer needs in a
bear market becomes relatively more important (Grewal and Tansuhaj,
2001). In this respect, marketing scholars have long maintained that
contractions, compared to expansions, provide companies with the rare
opportunity to boost market share and long-term profitability as competitors
might be forced to cut back expenditures (e.g., Steenkamp and Fang, 2011).
By contrast, the general rise in output levels in the bull market provides
ample opportunities for profitable growth due to the increasing level of
environmental munificence. Therefore, mismanagement of the exploitation-
exploration ratio in a bear market is likely to have a, compared to a bull
market, larger (negative) impact on firm performance.
Moreover, failure to take advantage of the reduced amount of
opportunities in a bear market, in combination with the general decline in
output levels, will cause the firm to face rapidly decreasing financial
performance (Walrave et al., 2011). This can give rise to a vicious feedback
loop (i.e., success or failure trap) in which swiftly decreasing performance,
caused by the drop in output levels and significant deviation from the
optimal exploitation-exploration ratio in the bear phase, further distorts the
24
FIGHTING THE BEAR AND RIDING THE BULL
development of a profitable exploitation-exploration ratio, which in turn
accelerates organizational decline (Leonard-Barton, 1992; Levinthal and
March, 1993). In the context of reactions to competitive and environmental
threats, such as a bear market, it appears to be fairly common for managers
to let their firms slip into such a vicious process (Walrave et al., 2011). By
contrast, this vicious feedback loop is less likely to develop in a bull market,
because of the general rise in output levels (Deleersnyder et al., 2004).
Together, these arguments suggest that a deviation from the optimal
exploitation-exploration ratio for firm performance in a bear market is likely
to have, relatively, greater (i.e., negative) consequences than such deviation
has in a bull market. We thus expect that the relative importance of the
exploitation-exploration ratio is larger during an economic recession than
during a recovery:
Hypothesis 2 The relative importance of the exploitation-exploration ratio
for firm performance is greater in a bear market than in a bull
market.
Figure 2.3 illustrates the expected moderating effect of a bear or bull
market on the relationship between the exploitation-exploration ratio and
firm performance (and builds on Figure 2.2). The difference with Figure 2.2
is that in this figure the steepness of the inverted U-shaped relationship
between the exploitation-exploration ratio and firm performance is changed.
The steeper curve of the bear market denotes its higher relative importance
as there is more to be lost by deviating from the vertex. Vice versa, there is
more to be gained by getting as close as possible to the optimum (hence in
Figure 2.3, for any given c: a < b).
2.2.3 Bear vs bull affects the optimal exploitation-
exploration ratio
The third expected moderating effect of the business cycle on the inverted U-
shaped link between exploitation-exploration and firm performance
concerns a shift in the optimal exploitation-exploration ratio. In this respect,
a recession is likely to change the business environment in terms of
CHAPTER 2
25
expectations and behaviors of customers, competitors and suppliers (Grewal
and Tansuhaj, 2001; Piercy et al., 2010). As such, bear markets offer firms
with a great number and range of threats and opportunities (Grewal and
Tansuhaj, 2001). This causes for an increased level of uncertainty within
organizations. Organizations can diminish some of this uncertainty by
expanding the scope of information acquisition (Sidhu et al., 2004). More
specifically, this implies gathering more boundary-spanning data for the
development of new approaches to handle the external developments. In
other words, firms that are directed to build explorative knowledge are likely
to be more able to flexibly adapt their overall operations in line with
unforeseen environmental change, in clear contrast to firms without
exploration activities (Grewal and Tansuhaj, 2001; Lee and Makhija, 2009).
In this respect, Sidhu et al. (2004) argue that the more turbulent the
environmental context (and the more severe the environmental selection
regime), the more important explorative learning becomes as this allows for
effective adaptation.
Figure 2.3: Illustration of hypothesis 2.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Exploitation-Exploration ratio
Effect on firm performance
Bear phase
Bull phase
a
c
b
c
H2: a < b
26
FIGHTING THE BEAR AND RIDING THE BULL
This idea is in line with observations by Steenkamp and Fang (2011)
that an increase in R&D activities during contractions is more effective for
building profit than increasing the R&D effort in expansionary periods.
Moreover, this idea resonates strongly with the corporate entrepreneurship
literature which suggest that firms, for reasons of effective organizational
adaptation, needs to engage in greater levels of entrepreneurial activities (i.e.,
exploration) as environmental hostility intensifies (e.g., Zahra, 1993; Zahra
and Covin, 1995). In a bear phase, organizational adaptation and explorative
activities are, therefore, relatively important. By contrast, by under-investing
in exploration, firms might fail to adjust to recovering and emerging
environmental situations and, therefore, lose their competitive advantage
(Srinivasan et al., 2011). As such, exploitation efforts, although necessary, are
considered less critical in the bear phase (D’Aveni and MacMillan, 1990;
Hambrick and Schecter, 1983).
Furthermore, it seems optimal for firms to engage in explorative
activities in a bear market when it conflicts less with production (due to the
decrease in demand), and wait until economic conditions improve before
introducing them (e.g., Barlevy, 2007). In the longer run, as the economy
improves, the company that engaged in exploration in a bear market will
have new offerings ‘shelf ready’ in the bull market (Steenkamp and Fang,
2011).
A bull market involves rising economic activity and increasing output
levels (Deleersnyder et al., 2004). In the bull phase, top management
typically attempts to bring sales and performance back to pre-recession levels
(i.e., toward the ‘Recovered’ point in Figure 2.1), or above. As the market is
expected to grow relatively stable in terms of customer preferences and
product demand, this is most likely achieved through more exploitative
activities. As such, the deflection point between a bear and bull market
signals the moment for top management to (ideally) re-divide the
exploitation-exploration ratio toward more exploitation. Thus, we expect that
in a bear market, compared to a bull market, the most profitable exploitation-
exploration ratio involves more exploration, and vice versa:
CHAPTER 2
27
Hypothesis 3 The exploitation-exploration ratio that is optimal in terms of
firm performance is higher (i.e., more explorative) in a bear
market than in a bull market.
Figure 2.4 builds on Figure 2.3 and illustrates the expected difference in
the most profitable ratio, by showing a different location of the vertex
between the two inverted U-shaped curves. Hypothesis 3 implies the vertex
shifts toward the left (i.e., more emphasis on exploitation, implying a lower
ratio) when moving from a bear to bull market; therefore, d < e in Figure 2.4.
Figure 2.4: Illustration of hypothesis 3.
2.3 Method
2.3.1 Data collection
Although business cycles affect the entire economy, not all industries are
equally effected (Deleersnyder et al., 2004; Steenkamp and Fang, 2011). The
IT industry is a fast-moving sector (e.g., continuous product innovation, high
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Exploitation-Exploration ratio
Effect on firm performance
Bear phase
Bull phase
Direction
of change
H3: d < eed
28
FIGHTING THE BEAR AND RIDING THE BULL
growth rates, and high product differentiation) (Mendelson, 2000). In this
respect, firms in high-tech markets tend to allocate greater resources to
exploration to manage the ongoing technological changes (Grewal and
Tansuhaj, 2001). Moreover, IT firms need to be more responsive to
environmental fluctuations and generate a return on (explorative)
investments faster than firms in many other industries (e.g., gas or food
industry) (Mendelson, 2000). Consequently, within the IT industry,
performance implications due to different exploitation-exploration
configurations (in different phases of the business cycle) are likely to be
observed more clearly, and within a shorter time span, than in most other
industries. As such, the Information Technology (IT) industry was selected
as the context for our empirical study.
To test the hypotheses developed in the preceding section, longitudinal
data covering both a bear and bull market are required. As such, we collected
data over the years 2007-2010 for companies active in the IT sector (16
quarters in total). In the Global Industry Classification Standard (GICS),
these firms are listed under code 4510-4530. In view of the global character
of the business cycle under investigation, we collected data on U.S. and E.U.
based companies. Using the GICS code previously mentioned, we selected
89 U.S. based firms from the Standard & Poor (S&P) 500 index and 11 E.U.
based firms from the S&P 350 EURO index. To improve the geographical
balance within the sample, we supplemented the data with all E.U. based IT
firms (not listed in the mentioned S&P indexes) that had a net income in
excess of 75 million dollars in 2007 (source: ‘Thomson ONE Banker’). These
21 firms are too small to be listed in one of the S&P indexes, but still align
well with the 100 firms extracted from these S&P indexes. All 121 selected
companies were publicly owned and traded at the beginning of 2007.
The resource richness and organizational structure of the companies
within the selected sample practically enables their top management to
simultaneously engage in exploitation and exploration (cf. O’Reilly and
Tushman, 2004). In this respect, the balance and pacing of exploitation and
exploration become more important than the absolute activity levels (as both
the resources and structure are typically available). As such, the choice to
CHAPTER 2
29
focus on large firms allows us to primarily focus on the exploitation-
exploration ratio as set out by top management.
The firm-level data were collected from two main sources: ‘Thomson
ONE Banker’ and the annual letters to shareholders. Fourteen firms were
omitted from the analysis because no letters were available. Moreover, two
firms comprised less than 6 (quarterly) observations (compared to an
average of over 14 per firm) and were omitted from the sample analyzed,
because such a limited number would provide a misfit with our longitudinal
research design. Nevertheless, inclusion of these two firms, as a robustness
check, resulted in highly similar findings – as reported in Appendix I under
the heading ‘Extra observations’. Another 125 quarterly financial
performance observations were not available, mostly due to stock market
exits. This resulted in a sample of 105 companies (incl. 75 U.S. and 30 E.U.
based) and 1555 valid observations over 16 quarters.
2.3.2 Measures
2.3.2.1 The recession and recovery phase
We analyzed the economic recession that started in 2007. According the
National Bureau of Economic Research (NBER), this specific recession lasted
18 months (Hall et al., 2010). The subsequent recovery that unfolded over
2009-2010 was of such strength and length that any subsequent recession
will be referred to as a new one (cf. Hall et al., 2010). This does not imply
that the economic conditions since the ‘Through’ point (see Figure 2.1) have
been particularly favorable. At the time of writing, the economic activity is
considered still to be below average (i.e., as found during a period of
expansion). In this respect, it was merely determined that the economic
recession ended and a period of recovery began.
Global economic upheavals tend to be synchronized at large (Claessens
et al., 2009), suggesting there is no need to accommodate for a delay
between E.U. versus U.S. based firms in the analyses. This idea is reinforced
by the fact that all firms in our sample are global players and, therefore,
affected by global crises simultaneously. Nevertheless, in order to verify this
30
FIGHTING THE BEAR AND RIDING THE BULL
statement for the selected firms, we calculated the average relative Tobin’s Q
for U.S. and for E.U. based firms. Subsequently, the Zivot and Andrews’
unit root test, which treats the breakpoint endogenously, was applied on the
two sub-samples (Zivot and Andrews, 2002). The breakpoint (i.e., the
minimum t-statistic, based on the slope) was found to be at quarter 9 for
both the U.S. based firms (t = -4.138, p < .10) and E.U. based firms (t = -
5.645, p < .01).
Following this finding and the definitions of the recession and recovery
phase adopted earlier, the data can be readily split into a bear and bull phase
with a deflecting point that ensues around quarter 9 at large. As such, for
quarter 1 until 8 (i.e., year 2007 and 2008) a dummy variable (‘Bear
dummy’) was coded zero to indicate a bear market, and for quarter 9 till 16
(i.e., 2009 and 2010) it was coded one to indicate a bull market. Figure 2.5
illustrates the tipping point from bear to bull market, in addition to the
average performance of the selected firms and the associated confidence
intervals (CI) in terms of relative Tobin’s Q. This figure also shows that the
average performance recovery of all firms at the end of 2010 (i.e., quarter 16)
equaled about 74 per cent.
Figure 2.5: Deflection point from bear to bull market (CI = Confidence Interval).
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Relative Tobin's Q
Bear phase
Bull phase
Average
80% CI
95% CI
80% CI
95% CI
Quarter
CHAPTER 2
31
2.3.2.2 Dependent variable: Relative Tobin’s Q
A variety of performance measures are used in the exploitation-exploration
literature. Some studies use self-reported subjective measures (e.g., Gibson
and Birkinshaw, 2004; Lubatkin et al., 2006) or accounting based-measures
(e.g., He and Wong, 2004), and yet others market-value based measures
(e.g., Uotila et al., 2009). In view of their retrospective bias, self-reported
subjective measures are not appropriate in a longitudinal research setting in
which historic data are collected (cf. Golden, 1992). Accounting based-
measures are also less suitable because of the long time lag for the results of
exploration to become manifest, compared to the more immediate effect of
exploitation (Lavie et al., 2010; Uotila et al., 2009). By contrast, market-value
based measures capture the short-term performance as well as long-term
prospects (Lee and Makhija, 2009; Lubatkin and Shrieves, 1986). In this
respect, empirical studies investigating performance effects longitudinally
have often utilized market-value based measures (e.g., Uotila et al., 2009).
As such, we calculated the widely utilized Tobin’s Q as the market-value
based index, by dividing the market value of a company by its book value
(Lee and Makhija, 2009). We considered the Tobin’s Q relatively because
relative metrics are more useful than absolute values in times of economic
upheaval (Reibstein and Wittink, 2005). This approach allows for direct
comparison of performance variation between firms arising from their
exploitation-exploration ratio in both the bear and bull market. More
specifically, all firms have the same Tobin’s Q (i.e., 1) at t = 1 and subsequent
values are calculated relatively to its initial value (see Figure 2.5). This
method serves to investigate the influence of the covariates on the relative
change in Tobin’s Q from the first quarter. A robustness check by means of
frequently used alternative operationalizations of the relative Tobin’s Q (e.g.,
Gozzi et al., 2008) demonstrated that the initially compressed variance did
not significantly influence the results (see Appendix I).
2.3.2.3 Independent variables: Exploitation-exploration ratio
Exploitation and exploration have been operationalized in many different
ways. For instance, the depth and breadth of technological search activity
have been used as a proxy (Katila and Ahuja, 2002). Other studies have
32
FIGHTING THE BEAR AND RIDING THE BULL
relied on questionnaires which target key personnel (He and Wong, 2004;
Jansen et al., 2006; Sidhu et al., 2007). These operationalizations are
frequently highly specific and, as such, lack generalizability and applicability
outside their respective contexts (cf. Uotila et al., 2009); moreover, it is
frequently unclear whether and how they resonate with the original
definitions of exploitation and exploration (March, 1991).
As has been argued, the capability of firms to simultaneously exploit
and explore inherently manifests itself in decision-making processes at the
top level of these firms. As such, we documented the CEOs’ attentional focus
– in terms of the exploitation-exploration ratio chosen – by content analysis
of the letters to shareholders (LTS). LTS are a relatively homogeneous
communication channel that is carefully controlled by top managers
(D’Aveni and MacMillan, 1990; Ocasio, 1997). These letters thus embody
the ‘corporate-speak’ of top management more than any other form of
communication. Moreover, content analysis of linguistic media is very useful
for reconstructing beliefs and perceptions of the authors (D’Aveni and
MacMillan, 1990). In this respect, previous research successfully engaged in
content analysis of LTS to uncover the strategic direction set by top
management (D’Aveni and MacMillan, 1990; Yadav et al., 2007). Also the
strategy literature indicates that TMT member to be suitable persons for the
measurements of organizational-level constructs (Sidhu et al., 2004). That
is, several studies confirmed and validated the link between the content of
LTS and actual firm activities (e.g., D’Aveni and MacMillan, 1990; Yadav et
al., 2007).
The operational definition of exploitation and exploration in our content
analysis is based on March’s (1991) original definition and operationalization
of the two terms. This ensured that our operationalization of the
exploitation-exploration ratio aligns well with the conceptual definitions
adopted. Moreover, Uotila et al. (2009) demonstrated that March’s
dictionary statistically and accurately differentiates between exploitation and
exploration. As such, exploitation was captured by (the roots of) the
keywords: ‘refinement’, ‘choice’, ‘production’, ‘efficiency’, ‘selection’,
‘implementation’, and ‘execution’. Exploration was captured by the (roots of
the) keywords: ‘search’, ‘variation’, ‘risk’, ‘experimentation’, ‘play’,
CHAPTER 2
33
‘flexibility’, ‘discovery’, ‘innovation’. Moreover, manual inspection of a
randomly chosen selection of LTS, comprising five per cent of all 405 letters,
revealed that ‘new’ and ‘technology’ were keywords repeatedly indicating
attention toward exploration; and ‘cost’ and ‘reduction’ keywords
representing a focus on exploitation. As such, (the roots of) these four words
were also included in the investigation. A preliminary analysis of the LTS
pointed out that contractions of the keywords selected are rarely used in the
context of other meanings (except in case of ‘executive’, which was,
therefore, excluded from the analysis).
To construct the exploitation-exploration variable, other researchers
have utilized an array of mathematical methods (e.g., subtraction,
summation, or multiplicative interaction) (e.g., Auh and Menguc, 2005; He
and Wong, 2004). There is no compelling rationale for choosing one
operationalization over the other, yet this choice greatly influences the
results. The assumption that exploration-exploitation are two ends on a
continuum serves to circumvent this empirical challenge (cf. Lavie et al.,
2010). As such, the annual exploitation-exploration ratio (EE-ratio) was
designed as the total number of matched keywords for exploration divided by
the sum of matched keywords for exploitation and exploration. As such, a
firm exclusively directed toward exploitation will score 0, while a firm
exclusively conducting explorative activities will score 1. In total, the
keywords were matched to 4,799 instances (of which 42 per cent to
exploration). We used the year that a letter was published to denote the EE-
ratio of that year. In this respect, we assume that the LTS adequately
represent and reflect past, current, and future initiatives planned by top
management.
The length of the LTS may influence the independent variable distilled
(Yadav et al., 2007). In order to check for any interference of the length of
the LTS on the EE-ratio, we correlated the EE-ratio with the amount of
characters per letter. This robustness test implied the relationship is not
significant (r = .068, p > .1). It can also be argued that shorter letters may
result in extremer EE-ratios. That is, finding one additional keyword in a
shorter text, where relatively fewer keywords are likely to be identified
compared to a longer letter, would have a greater influence on the EE-ratio
34
FIGHTING THE BEAR AND RIDING THE BULL
compared to finding one additional keyword in a longer text. To test for this
possibly confounding effect, we took the absolute value of .5 (the mean of the
EE-ratio scale) minus the EE-ratio, and subsequently correlated this with the
amount of characters found in a letter. This effectively tests whether fewer
characters in LTS result in extremer (exploitative or explorative) EE-ratio’s.
This robustness check also produced a non-significant relationship (r = .010,
p > .1); therefore, the length of the LTS has no significant effect on the
distilled EE-ratio.
2.3.2.4 Control variables
We included several variables to control for possible confounding effects.
The autoregressive component (y
t-1
) was included in the analyses to control
for firm past performance. Time dummies (for every quarter) were included
to prevent the most likely form of cross-individual correlation:
contemporaneous correlation (Roodman, 2009b). R&D spending is likely to
positively influence firm performance in times of economic upheaval
(Hoang and Rothaermel, 2010; Srinivasan et al., 2011; Steenkamp and Fang,
2011). As such, the standardized value of R&D spending as percentage of
sales was included (‘R&D expenditure’). However, not all companies
reported their R&D spending. Therefore, if a firm did not report its R&D
expenses, it was treated as being zero (effectively replacing the missing value
with the sample’s mean) and a dummy variable (‘R&D missing dummy’)
was coded as one (cf. Cohen et al., 2003; Uotila et al., 2009). Furthermore,
larger firms may be better able to mitigate the effects of economic recessions
and recoveries due to their large amount of resources (Lee and Makhija,
2009; Steenkamp and Fang, 2011). As such, we controlled for firm size,
measured by calculating the standardized value of the number of employees
(‘Firm size’). Also, older firms are likely to be more inert and so less able to
adapt to changing environmental circumstances (e.g., Steenkamp and Fang,
2011). Therefore, firm age in terms of the standardized value of the number
of days since initial public offering was included in the analyses. Moreover,
we incorporated the geographic location by coding and including a dummy
variable for U.S. versus E.U. based firms (‘U.S. location dummy’) and we
CHAPTER 2
35
also controlled for industry subsector by coding and including two dummy
variables: ‘GICS 4510 dummy’ and ‘GICS 4520 dummy’.
2.3.3 Analysis
A longitudinal research design can draw on sophisticated econometrical
methods that serve to control for endogeneity and unobserved heterogeneity
(Blundell and Bond, 1998; Roodman, 2009b; Uotila et al., 2009). In this
respect, simple dynamic panel models estimated with standard General
Method of Moments (GMM) estimators have often produced unsatisfactory
results (cf. Blundell and Bond, 2000). This is caused by a weak instrument
problem if the dynamic panel autoregressive coefficient is highly persistent,
causing large finite-sample biases (i.e., downward and imprecise) (Blundell
and Bond, 1998). As such, testing the hypotheses with the data at hand
required the use of system GMM estimation (Arellano and Bover, 1995).
System GMM estimation makes the endogenous variables predetermined
and, therefore, not correlated with the error term, which prevents
endogeneity problems. Moreover, system GMM estimation controls for
(unobserved) heterogeneity (Roodman, 2009b).
Roodman (2009b) recommends putting all regressors (and their lags)
in the instrument matrix. As such, almost all variables were treated as
predetermined (cf. Uotila et al., 2009); exceptions were the time dummies,
the ‘Bear dummy’, the industry dummies, and the ‘U.S. dummy’, which
were all treated as exogenous variables. This approach, combined with the
number of variables used in the analyses, resulted in a large number of
instruments and therefore in over-identification. Although over-
identification does not compromise the coefficient estimates, it does weaken
the Sargan/Hansen test and, as such, raises the need for robustness tests
(Roodman, 2009b). The models were, therefore, also tested by varying the
number of instruments. Appendix I reports these tests. These robustness
tests demonstrated that the key coefficients mostly remain comparable, in
terms of sign, effect size and significance level, with those of the model used
for hypotheses testing.
36
FIGHTING THE BEAR AND RIDING THE BULL
2.4 Results
Table 2.1 presents the descriptive statistics and correlations for the variables
used in this study. Table 2.2 presents the results of the system GMM
regression analyses.
In order to test for the (assumed) inverted U-shaped relationship, the
squared term of the independent variable under investigation (EE-ratio) is
included in the model (Aiken and West, 1991). As such, the first model
introduces ‘EE-ratio’, the ‘EE-ratio squared’, and the ‘Bear dummy’. The
second model examines the moderating effect of the phase of the business
cycle (Bear dummy) on the relationship between the EE-ratio and firm
performance by including interaction terms (Aiken and West, 1991).
First of all, the second model (with interactions) has a significant better
overall model fit than the first model (with direct effects only) (p < .001). As
such, the second model will be further discussed. The autoregressive
component – the relative Tobin’s Q
t-1
– is highly persistent (b
6
= .885, p <
.001), which justifies the use of system GMM estimation (Blundell and
Bond, 1998). Next to this, both R&D expenditure (b
7
= .025, p < .01) and firm
age (b
10
= -.027, p < .01) significantly influence the dependent variable. That
is, the more R&D investments are made during times of economic upheaval
and/or the younger the firm is, the better its performance. The former
finding replicates the empirical results by Steenkamp and Fang (2011).
Model 2 points at a curvilinear relationship between the EE-ratio and firm
performance in both the bear and bull market context. That is, the required
coefficients are statistically significant and have the correct signs: EE-ratio (b
1
= .826, p < .01), EE-ratio squared (b
2
= -.633, p < .01), EE-ratio – Bear dummy
interaction (b
4
= -.627, p < .05), and EE-ratio squared – Bear dummy
interaction (b
5
= .438, p < .05). The vertexes are located within the
theoretically plausible exploitation-exploration range (.65 for the bear phase
and .51 for the bull phase), providing evidence that the relationships are non-
monotonic. This implies that firms possessing a more balanced exploitation-
exploration ratio are likely to achieve a better relative performance compared
their ‘non-balanced’ competitors, in both bear and bull market.
CHAPTER 2 37
*Correlations are significant the .05 level. Significance levels reported are two-tailed.
12
11
10
9
8
7
6
5
4
3
2
1
Table 2.1: Means, Standard Deviations, and Correlations* (t1 till t16).
U.S. location dummy
GICS 4520 dummy
GICS 4510 dummy
R&D missing dummy
Firm age
Firm size
R&D expenditure
Bear dummy
(EE-ratio)
2
EE-ratio
Relative Tobin’s Qt-1
Relative Tobin’s Q
.738
.328
.412
.136
.000
.000
.000
.477
.487
.677
.779
.776
mean
.44
.47
.492
.343
1.00
1.00
1.00
.5
.222
.172
.319
.321
St.dv.
-.029
-.085*
.011
-.013
-.022
-.021
.123*
-.332*
.072*
.087*
.871*
1
.032
-.078*
.005
-.025
-.032
-.021
.081*
-.433*
.092*
.109*
2
.042
-.141*
.187*
.055*
.104*
.026
.223*
-.155*
.984*
3
.043
-.145*
.197*
.049
.094*
.029
.223*
-.144*
4
.001
-.009
.021
.009
-.012
.016
.034
5
.088*
-.265*
-.016
.001
.074*
-.242*
6
.102*
.122*
.036
-.031
-.288*
7
-.132*
-.265*
.235*
.002
8
-.097*
-.161*
.363*
9
.139*
-.585*
10
-.052*
11
38 FIGHTING THE BEAR AND RIDING THE BULL
This finding replicates the findings by Uotila et al. (2009) and underpins
that exploitation and exploration (in the context of this study) can be
considered as two ends on a continuum (i.e. by a ratio), characterized by an
inverted U-shaped relationship with firm performance.
Table 2.2: Results of the system GMM regression analysis
(half of the available lags used).
Dependent variable:
Relative Tobin’s Q
Model 1:
Model 2:
Coeff.
(S.E.)
b
Coeff.
(S.E.)
b
b
1
– EE-ratio .396
(.196)* .826
(.306)**
b
2
– (EE-ratio)
2
-.359
(.157)* -.633
(.225)**
b
3
– Bear dummy .154
(.021)*** .368
(.099)***
b
4
– EE-ratio * Bear dummy
-.627
(.307)*
b
5
– (EE-ratio)
2
* Bear dummy
.438
(.233)*
b
6
– Relative Tobin’s Q
t-1
.885
(.020)*** .885
(.019)***
b
7
– R&D expenditure
a
.024
(.010)** .025
(.010)**
b
8
– R&D missing dummy
-.012
(.035) -.006
(.036)
b
9
– Firm size
a
-.019
(.021) -.013
(.018)
b
10
– Firm age
a
-.029
(.011)** -.027
(.011)**
b
11
– U.S. location dummy -.009
(.014) -.010
(.014)
b
12
– GICS 4510 dummy
.027
(.018)+ .022
(.017)
b
13
– GICS 4520 dummy
.000
(.017) .000
(.016)
b
14
– Constant -.146
(.063)** -.305
(.098)**
Hansen test of over-identification