Orientation evolution in start-ups : exploitation and exploration processes reviewed

(1)

Orientation evolution in Start-ups:

exploitation and exploration processes

reviewed

Master Thesis MSc Entrepreneurship

Author Evert Brolsma

Universities VU University & University of Amsterdan Student-numbers 2585483 (VU), 11055383 (UVA)

Supervisor Dr. J. Sol

Faculty Faculty of Economics and Business Administration

(2)

2

Acknowledgements

At first, I would like to express my sincere appreciation to all the start-up owners who took the time to fill out the survey they got send. Without it, this research would not be possible. My gratitude is everlasting.

I also would like to thank my supervisor dr. J. Sol. Without his guidance and advice in desperate times this thesis would’ve never been made.

(3)

3

4. Methodology ... 15 4.1 Research method ... 15 4.2 Data collection ... 15 4.3 Measurements ... 16 4.3.1 Dependent variables ... 16 4.3.2 Independent variable ... 17 4.3.3 Control variables ... 17 4.4 Statistical analyses ... 18 4.4.1 Assumptions ... 18 4.4.2 Analyses ... 19 5. Results ... 20 5.1 Assumptions ... 20 5.2 Descriptive statistics ... 21 5.2.1 Univariate analyses ... 21 5.2.2 Bivariate analysis ... 23 5.3 Hypotheses testing ... 25 5.4 Summary of results ... 28 6. Discussion ... 29 6.1 Theoretical discussion ... 29 6.2 Practical implications ... 30 7. Conclusion ... 32 7.1 Findings ... 32 7.2 Limitations... 32

(4)

4

8. References ... 35

8.1 Articles ... 35

8.2 Books ... 40

Appendix A: Original questionnaire ... 42

Appendix B: English questionnaire ... 45

Appendix C: Likert scale questions E&E; translation and referencing ... 48

Appendix D: Original mail survey-email sent (and translation) ... 49

(5)

5

Abstract

The aim of this research is to explore the different orientations within start-up companies, and to see how these orientations possibly change/evolve over time. The concepts central to this theses is are of exploration and exploitation. A vast amount of literature has already been written about these two orientations, and results are sometimes somewhat contradictory. This research adds to the existing literature by focussing on start-up companies only, whereas most of the research on exploration and exploitation processes has focussed on large companies.

Start-up company founders have been asked to fill out a survey. 82 respondents make up the dataset for this research. The results of this research show that older start-ups significantly focus less on exploitation then their younger counterparts do.

(6)

6

1. Introduction

Organizational learning entails both the exploration of new possibilities and the exploitation of old certainties (March, 1991). On the one hand, in order to keep up in touch with consumers, knowledge needs to be sought and integrated, while on the other also exploiting the already gathered information and to integrate that knowledge into certain routines and practices (March, 1991). This paper attempts to further discover this relationship between Exploration and Exploitation (from now on referred to as E&E) and organizations.

The key matters of studies of adaptive processes, namely the relationship between the exploitation of old certainties and the exploration of new possibilities (Schumpeter, 1934; Holland, 1975; Kuran, 1988), are also applicable to organizations (Burns and Stalker, 1961). Exploratory processes include the pursuit of novel mental models, and the development of new stereotypes and mind-sets regarding potential customers, whereas exploitative processes are building upon pre-existing routines and procedures that meet the needs of current customers (Baets, 1999). Benner and Tushman (2003) argue that exploration and exploitation differ on a fundamental level, and that these concepts require

profoundly different structures and strategies. In line with this, Gupta, Smith and Shalley (2006) argue that this fundamental difference between both processes make a synchronised pursuit of both very difficult. The requirements for profits of today are control, order and stability, while adapting for those of tomorrow require flexibility, change and creativity (Volberda and Lewin, 2003). Looking for an maintaining a balance between E&E is critical for the survival of a firm (March, 1991).

Although many research has been done on the achievement of a proper balance between E&E (March (1990), Cegarra-Navarro and Sánchez-Vidal (2011), Csaszar (2013), Stettner and Lavie (2014)), none so far have specifically focussed on the early stages and the development of these processes.

This paper attempts to shed light on precisely these early stages by looking at start-up companies. More specifically, this research attempts to explore whether these is some kind of coherent trend in the way this possible balance comes to fruition in a start-up company, and if there even is any balance at all. Schumpeter (1934) argues that entrepreneurship (and thereby start-up companies) is crucial in the understanding of economic development. Entrepreneurship is argued to be important for the

development of a healthy economy, in which prosperity is sustained and jobs are created (Henry, Hill and Leitch, 2003). This leads to the following research question:

How do the exploitation and exploration processes in a start-up develop over time?

Attempting to answer this question could provide insights as to which forces are at play in the important early stages of a firm. Additional questions to be asked are if the way a certain start-up is financed, the sector in which the up is operating and the number of people working for the start-up are of influence on the exploration and exploitation processes at play.

(7)

7 The following section will provide a theoretical background, including a literature review, for the research. After that the research method will be introduced and explained, followed by the research itself; presenting the results and an analyses. These results will be discussed and concluding remarks will be given, together with the research limitations and implications for future research.

(8)

8

2. Theoretical framework

Literature written about E&E, and even more so the literature about start-ups and the overarching entrepreneurship theme is vast and ever growing. Different perspectives and definitions are used to describe the phenomena at hand. To prevent misconceptions and to provide clarity as to where this research is actually about, this section will go more in depth about the definitions used for the different concepts. Starting off with going more in depth about start-ups and the overarching theme of

entrepreneurship.

2.1 Entrepreneurship and start-ups

2.1.1 Entrepreneurship

Start-up companies is part of the field of entrepreneurship, a field that has gained growing amounts of attention since the 18th century amongst scholars, practitioners and academics. Many authors

acknowledge the importance of the field and consider the value creation, technological improvements and the contribution to economic growth as the backbone of the economy (Miller, 1983; Gorman, Hanlon and King; 1997; Hisrich and Peters, 1992; Minniti & Lévesque, 2008). Entrepreneurship drives innovation and economic and societal advancement of nations (Van Praag and Versloot, 2007).

Entrepreneurship in general can be considered as a process with (at least) three phases:

• Phase 1: The phase in which the entrepreneur develops insights and identifies opportunities that might be feasible and viable business opportunities

• Phase 2: The stage at which the entrepreneur gathers the necessary resources for the actual start of a venture, e.g. the launch or development and execution phase.

• Phase 3: Post-launch phase in which the new venture is managed by the entrepreneur to make it grow and survive (Baron, Hannan and Burton, 1999)

These Phases will also be incorporated in this research. E.g., it adopts the vies that entrepreneurship is a process, rather than an instantaneous event (Low and Macmillan (1988).

For the phenomena of entrepreneurship there have been a broad range of definitions. For reasons of parsimony the definition of entrepreneurship adopted in this paper is the pretty straight-forward one (Jones, Macpherson and Jayawarna, 2013) given by Shane and Venkataraman

(2000):“[Entrepreneurship,] the field involves the study of sources of opportunities, the processes of discovery, evaluation, and exploitation of opportunities; and the set of individuals who discover, evaluate, and exploit them” (p.218).

(9)

9

2.1.2 Start-ups

The definition of entrepreneurship is quite general. For example, it does not exclude large firms, old firms, family firms, etc. To limit this broad range this paper only focusses on firm that are in the start-up/seed stages of the venture creation process, so called start-ups. The definition a start-up upheld in this paper is an adoption from what Van Praag en Versloot (2007) call entrepreneurial firms, which they define (e.g. a start-up) as “firms that satisfy one of the following conditions: (i) They employ fewer than 100 employees, (ii) They are younger than 7 years old; (iii) they are new entrants into the market” (p. 353).

Literature suggests that new ventures need different types of capabilities than established corporations (Zhara, Sapienza and Davidsson, 2006). The environments for start-ups are extremely versatile; only about half of all the start-up businesses manage to survive more than 5 years, and these failure rates appear to be stable over time (Headd, 2003; Cader and Leatherman, 2011; Lueg, Malinauskaite and marinova, 2014). It precisely this combination that makes it an interesting endeavour for this research.

2.2 Two ways of looking at exploitation and exploration processes

There are different schools of thought when it comes to the discussion about the processes by which opportunities are actually formed, and they are inherently different (Alvarez and Barney, 2007). The realization amongst scholars that the process by which opportunities are formed can differ

systematically is quite recent (Aldrich and Ruef, 2006; Venkataraman, 2003; Alvarez, Barney and Anderson, 2013) and these differences may have essential implications; both for how entrepreneurs can effectively exploit opportunities and for a broader societal and economic phenomena context (Alvarez and Barney, 2007, 2010, Alvarez et al., 2013).

One of the two ways to look at the opportunity creation/recognition process is the so called creationist theory. Creation theory reasons that opportunities are created by the actions, reactions and the

enactment of the entrepreneurs exploring ways to produce new products and services (Alvarez and Barney, 2007; Aldrich and Kenworthy, 1999; Aldrich and Ruef, 2006, Baker and Nelson, 2005; Gartner, 1985; Weick, 1979). An opportunity can only exist (and therefore be understood) until they are actually brought to existence by the actions and reactions of the entrepreneurs (Berger and

Luckmann, 1967; Weick, 1979). Variations –either blind or intentional- are the cause of the process of action and reaction to start (Aldrich and Ruef, 2006). The initial beliefs and perceptions of what resources and abilities are needed to exploit an opportunity are the things an entrepreneur uses to act upon (Sarasvathy, 2001), and these are therefore the building blocks from which opportunities arise.

The other school of thought sees opportunities as pre-existing objects that are already out there, ready to be discovered and exploited by entrepreneurs (McKelvey, 1999), the so called discovery theory. In

(10)

10 this view, opportunities exist independent from perceptions (Mckelvey, 1999). Most of the prior research has taken this so called discovery process perspective (Alvarez et al., 2013). Exogenous shocks cause consumer preferences to change, and cause changes in technology, industry and markets (Shane, 2003; Kirzer, 2015). This is where the entrepreneurs step in; they are the ones able to spot and actually exploit these opportunities, making them different from non-entrepreneurs (Kirzer, 2015; Shane, 2003).

Since this research uses qualitative data to test for hypotheses, e.g. assuming there is a pre-existing relationship ready to be discovered, the discovery theory approach is adopted in this paper (Bryman, 2003).

2.3 Exploitation and Exploitation

The paradigm of exploration-exploitation has already received much attention, and stems back to the central concern of the relation of exploring new possibilities and exploitation of old certainties (Schumpeter, 1934; Holland, 1975, Kuran, 1988; March, 1991, Stettner and Lavie, 2014). E&E has been a dominant research direction in the fields of strategic management (He and Wong, 2004; ; Uotila, Maula, Keil and Zahra, 2009), in innovation management (Jansen, Van den Bosch, and Volberda, 2006) and in organizational learning (March, 1991; Levinthal and March, 1993). Examples of organizational activities having to do with exploration focus on experimentation, searching, risk taking, flexibility and discovery (March, 1991). Contrary to that are organizational activities having to do with exploitation, examples are the strive for efficiency, production, refinement, execution and implementation (March, 1991).

E&E are not independent processes, they are constantly influencing each other (Holmqvist, 2004). The search for new capabilities (e.g. exploration) also contributes to a venture’s existing knowledge base (e.g. exploitation) (Katila and Ahuja, 2002). There appears to be an everlasting tension between exploiting what has already been learned, and exploring new knowledge (Crossan, Lane and White, 1999). The tension between E&E is likely to be the cause of fear for the unknown, distrust and discomfort, making it challenging to overcome (Homlqvist, 2004). Therefore, empowerment by exploration and exploitation processes is required in order to achieve organizational goals and complicated tasks, not obstruction by the tension between E&E processes (March, 1991; Cegarra-Navarro & Sánchez-Vidal, 2011).

Typically exploration efforts are more costly in the short term, since breakthroughs can take years to come to completion (Bierly and Daly, 2007; Moss, Payne and Moore, 2014). Efforts having to do with exploitation typically are more aimed at short term gains, as firms are striving for improving the quality and efficiency of their production (Benner and Tushman, 2003). Combining these factors with the notion that for early stage entrepreneurs finding early stage funding is a difficult task (Jones et al., 2013), results in the following hypothesis:

(11)

11 H1: A start-up that just started out will be more exploitative than a start-up that has been up and running for a longer period of time.

As a start-up grows older, the environment it faces likely becomes less versatile; business concept has been proven and a relatively stable network and customer -bases are in place (Jones et al., 2013). This will arguably enable a start-up to step away from the short-term focus and allows for more of a long-term focus. Following this reasoning leads to the following hypothesis:

H2: A start-up that has up been running for a longer period of time will be more explorative than a start-up that just started out.

Venkatraman, Lee, and Iyer (2007) argue that balancing E&E can have both positive and negative effects, effected mostly by time and market dominance. Lavie, Kang and Rosenkopf (2011) even find negative effects of balancing E&E because it can lead to conflicting organizational routines.

However, most studies argue that a balance between both processes has a positive effect on performance (March, 1991; Jansen et al., 2006; Lin, Yang, and Demirkan, 2007; Prieto, 2007; Belderbos, Faems, Leten and Looy, 2010; Stettner and Lavie, 2014; O’Reilly and Tushman, 2004; Raisch, Birkenshaw, Probst and Tushman, 2009; He & Wong, 2004; Ugur et al., 2013). Reasoning for this balance is that when a firm leans to much towards exploration it will suffer the costs of

experimentation and generate too many undeveloped new ideas without gaining a distinctive

competence, e.g. suffering the costs of experimentation without grabbing many of its benefits (March, 1991). When a venture puts too much emphasis on exploration processes, long term capabilities might be boosted, but it might become susceptible to a so called “failure trap”. This happens when a firms focus on exploration drives out exploitation processes, causing failure to meet profitability and cash demands in the short run (Levinthal and March, 1993; Ugur, 2013).

On the other side are firms that lean to much towards exploitation; they are likely to eventually find themselves trapped in suboptimal stable equilibria (March, 1991). Putting emphasis on exploitative processes can bolster efficiency and effectiveness which in turn contributes to short time performance, but it can also be the cause of a so called “success trap”, because solely focussing on exploitation might lead to ignoring exploration which can have negative effects on long-term performance (Levinthal and March, 1993; Ugur, 2013).

He and Wong (2004) have found that the balance of E&E has a positive influence on sales growth. In line with that are the findings of Cao, Gedajlovic and Zhang (2009), who observe that a balance of E&E is positively related to the performance of a firm, especially under resource-constrained contexts. Ugur et al. (2013), Vagnani (2012) and Uotila et al. (2009) have found an inverted U-shape

relationship between exploration activities adopted, and a firms’ market value, which suggest there is a need for a balance (Ugur, 2013).

(12)

12 The literature clearly suggests that a venture benefits from balancing E&E (Gupta, 2006; Ugur, 2013). This notion is coupled to the high mortality rates amongst start-up companies (Headd, 2003; Cader and Leatherman, 2011; Lueg, et al., 2014), leading to the proposal of the following hypothesis:

H3: As start-ups grows older, they become more balanced when it comes to exploitative and explorative processes.

Stepping away from the notion that having a proper balance of E&E is beneficial for a firm, the discussion about how such a balance actually is acquires is a source of controversy (Gupta et al., 2006). According to the literature, two balancing mechanisms are in place: a structural one and a temporal one (Ugur, 2013). The structural view suggests that balancing E&E can be done

simultaneous though the specialization of certain subunits in either E&E activities (Jansen, Tempelaar, Van Dan Bosch and Volberda, 2009; Tushman and O’Reilly, 1997). Having a sole focus on either exploitation or exploration in certain units, avoids the tension between the both activities (Ugur, 2013). The other way of looking at acquiring balance in a venture, the temporal view, suggests that tension between E&E can be solved by a shift of focus over time within the same organization (Brown and Eisenhardt, 1998). In this temporal separation long periods of exploitation are followed by short bursts of exploration, creating a circle where E&E activities strengthen each other where a firm benefits from through learning and knowledge transfers (Gupta et al., 2006; Boumgarden, Nickerson and Zenger, 2012; Rothaermel and Deeds, 2004).

Empirical research shows that ventures that engages in both E&E, e.g. firms that are balanced, systematically financially outperform ventures that focus on only exploitation or exploration activities (He and Wong, 2004; Lubatkin, Simsek, Ling and Veiga, 2006; Uotila et al., 2009; Jansen et al., 2006; Vankatraman, Lee and Iyer, 2007; Lavie, Stettner and Tushman, 2010; Belderbos et al., 2010)

However, research comparing differences in financial performance of temporal and structural

balancing mechanisms has thus far not shown significant differences (Gupta et al., 2006; Ugur, 2013; Simsek, Heavey, Veiga and Souder, 2009).

2.4 Additional forces at play

Csaszar (2013) reasons that organizational structure is connected to E&E: larger polyarchies tend to be more explorative, whereas larger hierarchical firms are less explorative. Beckman, Haunschild and Philips (2004) argue that E&E processes are influenced by the type of interorganizational relationships they face: when experiencing market uncertainty, existing networks are exploited and reinforced.

Exploration of alliance networks is in turn triggered when a high degree of firm-specific uncertainty and a low level of market uncertainty is experienced.

(13)

13

In an attempt to resolve the tension between E&E, some firms form alliances to separate their E&E activities (Lin, Yang and Demirkan, 2007; Lavie, Kang and Rosenkopf, 2011; Im and Rai, 2008; Lavie et al., 2011). This separation of activities enables firms to structurally and simultaneously pursue E&E activities, which can have a critical effect on firm performance (He and Wong, 2004, Gupta et al., 2006; Jansen et al., 2009; Cao, Gedajlovic and Zang, 2009; Blome, Schoenherr and Kaesser, 2013; Tushman and O’Railly, 1997).

Organizational size is hypothesised to have an effect of E&E by different authors (Cao et al., 2009). Although the effects of size are somewhat contradicting (Csazar, 2013). Rothaermel and Deeds (2004) argue that as a firms grows, the likelihood of withdrawing from the path of discovery, development and commercialization of promising projects increases, thus lowering explorative processes. Beckman et. al. (2004) take a different stance, arguing that the larger the firm, the larger the resources it has at hand, the larger the ability to make network changes is, and the bigger the likelihood of making desirable partners is, thereby increasing explorative processes.

Several scholars have suggested that the mediating role of factors influencing E&E should be studied and in order to form a better understanding of the dynamic nature E&E has (Raisch et al., 2009; Ugur, 2013). O’Reilly and Tushman (2013) argue that a firm evolves and adapts its entire structure to changing market conditions. Environmental factors can have a moderating effect on E&E activities (Tamayo-Torres, Ruiz-Moreno and Lloréns-Montes, 2011). The moderating effects of R&D intensity (Uotila et al., 2009), absorptive capacity (Rothaermel and Alexandre, 2009) and organizational structural context (Thongpapanl, 2012) have also been shown in previous studies.

(14)

14

3. Summary of hypotheses and conceptual framework

The hypotheses that originated from the theoretical framework are put together and presented with the use of a conceptual framework in order to provide a clear picture.

H1: A start-up that just started out will be more exploitative than a start-up that has been up and running for a longer period of time.

H3: As start-ups grow older, they become more balanced when it comes to exploitative and explorative processes.

(15)

15

4. Methodology

This section of the paper will elaborate and describe the research method, the data collection ,the actions related to the preceding of the statistical analyses, and the statistical analyses itself. Gill and Johnson (2010) and Saunders (2011) argue that a structured methodology is important to facilitate replication, which is in turn important to ensure the reliability of the research. Replication adds to the controllability of a research, which is together with reliability and validity of utmost importance when it comes to judging the quality of a research (Yin, 1994; Swanborn, 1996; Van Aken, Berends and Van der Bij (2012). Because of their importance, the concepts of reliability and validity will be explained further in the following section, together with a discussion of the research method. The process of data collection and an description of the sample population is next. The end of this section will contain a detailed description of the statistical analyses that will be executed.

4.1 Research method

This research takes an objectivism approach to research, assuming that phenomena and their meanings exist independent of social actors (Bryman, 2003). The objective is to generate findings that can be generalized across different groups and individual behaviours (Cunliffe, 2010). The nature of the explanations are deductive; pre-existing theories are consulted and used in order to come up with a number of hypothesis that in turn get tested with a certain design (Saunders, 2011). In this research quantitative data has been collected to test the hypothesis. This type of research, called theory testing (Aken et al., 2012), seems suitable since there is already a vast amount of existing literature on E&E and on new venture creation and development (e.g. entrepreneurship), but the theories from both fields have never been combined and tested.

4.2 Data collection

To collect the required data for this research internet surveys have been made. The use of surveys is a popular method to collect large amounts of data (Saunders, 2011). With the use of surveys quantitative data can be collected and analysed statistically and thereby testing possible relationships between different variables.

Respondents were sent an email in which they were asked to fill out the survey (see appendix D for the text of the email that was send). The companies approached were found on the internet, with the use of www.foundedinholland.com and www.dealroom.co. In an attempt to come across as more personal and thereby increase the response rate the questionnaire was in Dutch, and only start-ups that

(16)

16 had a Dutch founder where approached. Additionally, anonymity was promised to the respondents. If provided or if it was possible to find out otherwise, the founder of the start-up company was mailed personally. If this was not possible, the email was sent to the firm’s address. The data was collected in the period of the 6th of June 2016 and the 23th of June 2016 with the help of survey-software program Qualtrics. 486 mails were sent in total, resulting in 82 actual responses. This means that the response rate was approximately 17%. Out of these 82 respondents, there were 59 questionnaires completely filled out. Only these were used for the analyses of the data, reasoning that the incomplete surveys were a sign of respondents not answering the questions seriously and that these surveys would therefore decrease the validity of the research.

4.3 Measurements

This section contains an elaborate explanation of the variables given in the conceptual model (see section 2.4). See appendix A and B for the actual survey questions asked, and see appendix C for the origins (references) of the questions asked. Next up are explanations of the measurements used for the dependent variables, the independent variables and the control variables.

4.3.1 Dependent variables

To turn E&E into a measurable variables constructs were taken from studies by Cegarra-Navarro And Sánchez-Vidal (2011), Kohli, Jaworski and Kumar (1993) and Molina-Castillo, Jimenez-Jimenez, Munuera-Aleman (2011) (see appendix C). Adapted from these papers are statements of which respondents were asked to indicate on 7-point Likert-type scale in how far they disagree or agreed with them. This scale is developed by Likert (1932) and is usually used to produce ordinal data. However, the scale used in this study produces interval data that can later be analysed, something that can be considered common practice in research (Blumberg, Cooper and Schindler, 2014). The

statements of exploration and exploitation are intentionally mixed up in an attempt to increase the validity of the research, since the mixing up of questions keeps the respondents on their toes (Pelham and Blanton, 2012).

Exploitation is measured using 3 statements taken from Molina-Castillo et al., (2011) and Atuahene-Gima (2005), and one statement adapted from (Cegarra-Navarro & Sánchez-Vidal, 2011, Kohli et al., 1993). No statements were left out, since this resulted in the largest Cronbach’s alpha possible, namely 0,676 (N=4).

(17)

17 A Cronbach’s alpha (α) is a coefficient that indicates the internal consistency of the statements

(Cronbach, 1951), and can be used to check the reliability of the developed scale. An Cronbach’s alpha can range from zero to one, and the closer to one the more consistent (and thus reliable) the developed scale is. Generally an Cronbach’s alpha of above 0,70 is asked, however, arguments are made that an alpha of more than 0,6 is acceptable (Nunnaly, 1978).

Exploration is also measured using 3 statements taken from Molina-Castiollo et al. (2011) and Atuahene-Gima (2005), and one statement adapted from (Cegarra-Navarro & Sánchez-Vidal, 2011, Kohli et al., 1993). However, one question needed to be excluded in order to get to an Cronbach’s alpha of above 0,6, namely the statement “We have acquired manufacturing technologies and skills that are entirely new to the firm” (Molina-Castiollo et al., 2011; Atuahene-Gima, 2005) (See appendix C)). Excluding this statement resulted in a Cronbach’s Alpha of 0,627 (N=3).

In order to test the hypothesis regarding the balance of E&E, an additional variable has been

constructed. By extracting the responses to exploration by the responses to the exploitation statements, and afterwards standardizing these new numbers, and additional variable was made.

4.3.2 Independent variable

The independent variable used in all of the analyses is age. To take into account the activities that take place in the phase that precedes the actual and official start of a business (Baron et al., 1999), two question where asked regarding the age of a firm, namely how many months ago the business was officially started, and, to cover the nascent stage of the firm, how many months were spend to get the business officially up and running. The age used in the analyses is simply the answers to both of these questions added up.

4.3.3 Control variables

Personal growth and the way the start-up was funded are used as control variables. Since an

questionnaire has been used, personal growth appeared to be the most accessible measure/indication of start-up growth. To measure employee growth two questions were asked: one in which respondents were asked to indicate the number of people active in the firm at the start of it, and one in which respondents were asked to indicate the number of people active at the time they filled out the questionnaire. An division has been made between Full-time and Part-time members, in order to expose possible differences between the two. This results in continuous variables.

(18)

18 In order to measure the way a start-up was funded respondents were asked as to indicate the way they were financed, multiple answers were possible and choices were: Bootstrapping/private debt, Family & Friends, Private investors, Angel and seed investors, Crowdfunding, Incubators. These variables are recoded into dummy variables, in which 0 means the type of funding was not used, and an 1 means it was used.

4.4 Statistical analyses

As can been seen in the conceptual model in section 2.4, multiple relationships are measured in this research. The first hypothesized relationship is that between a start-up’s age and exploitation processes. The second is the effect that age has on exploration processes. The last relationship is the effect age might have on the possible balance between exploration and exploitation. Controlled for are the way a start-up is financed and employee growth.

4.4.1 Assumptions

When using a regression analysis there are certain assumptions that have to be met, otherwise the results might be inaccurate or even false (Field, 2013; Osborne and Waters, 2002; Agresti and Finlay, 2008). The assumptions tested for in this research are those of normal distribution, linear relationship, homoscedasticity, multicollinearity, and the normality of the residuals. The results of these tests can be found in section 5.1. In this section all the assumptions will be explained.

Normally distributed data is needed since regression assumes variables have a normal distribution. A non-normal distribution can distort relations and significance tests. Adding to that is the fact that regression analyses can only make accurate estimations of relationships between different dependent and independent variables if these relationships are linear in nature. If the relationship between dependent and independent variables is not linear, a regression analysis will under-estimate the true relationship, which in turn increases the risk of both a Type II error (for that independent variable) and for a Type I error (for other independent variables that share variance with that independent variable) (Osborne and Waters, 2002).

When doing an regression analysis, the variance of errors need to be the same for all levels of the independent variable (Osborne and Waters, 2002). This can be checked by using partial plots to check the so called homoscedasticity of the data. Although slight violation has little effect, if

(19)

19 thereby increasing the change of a Type I error (Osborne and Waters, 2002; Berry and Feldman, 1985; Tabachnick and Fidell, 2007).

A regression analysis assumes that independent variables are not independent from each other. This could possibly give an incorrect result as to which variable influences the dependent variable. Multicollinearity controls for this coherence between the independent variables. A high coherence leads to a high multicollinearity (Agresti and Finlay, 2008).

Tabachnick and Fidell (2007) and Agresti and Finlay (2008) stress the importance of also checking if the residuals are normally distributed. This assumption is used to check whether or not the residuals contain a certain structure that is (or is not) accounted for in the model.

4.4.2 Analyses

At first, the assumption just given are tested for. After that, the descriptive statistical data will be given with the use of univariate- and bivariate analyses. Then the different hypothesis will be tested with the use of 3 different hierarchical multiple regression analyses. In this research, 3 models are used. All of these models at least tested for the dependent variables of exploration, exploitation and the balance variable age. The independent variable of age is added in the first model. The control variables Employee growth and type of funding are added in the second and third model. Hypothesis 1, 2 and 3 are tested in different models.

(20)

20

5. Results

In this section multiple test are run to see if the assumptions (given in section 3.4.1.) hold true. After that descriptive statistics are given, with univariate analyses of the main variables. This will be followed up by a bivariate correlation analysis of the variables used for this research. This section will be concluded with the hypothesis testing with the use of different regression analyses.

5.1 Assumptions

In a regression analysis the dependent variable should be coherent with all the independent variables in order to uphold the assumption of normal distribution, linearity and homoscedacity. To control for this histograms have been made for all of the dependent variables (see appendix E). These show that each dependent variable; exploration, exploitation and the balanced variable of E&E, are normally

distributed. Scatterplots showing the standardized predicted values and the standardized values of all three of the dependent variables (see appendix E) show that no clear patterns appear. All variables are randomly scattered around the average of zero, which means the assumptions of linearity and

homoscedasticity are not violated.

An additional test has been used to see in the residuals are normally distributed. With the use of P-P plots, it is made clear that a slight violation of this assumption has been made, since not all the estimated residuals are on one line. A slight s curve appears in all the P-P plots (for a further elaboration, see section 7.2).

To check for multicollinearity, the coherence between the independent variables, the variance inflation factor (VIF) is calculated. A maximum acceptable level of a VIF is 4, otherwise this assumption is violated and a conclusions regarding the relationship between variables might be false (Agresti and Finaly, 2008). In table 1 All the VIF scores are given. No VIF score is higher than 4, which means that the requirements for this assumption are met.

(21)

21 Collinearity Statistics

Tolerance VIF

Age 0,797 1,254

Bootstrapping/private debts 0,807 1,239

Family & Friends 0,834 1,199

Private loans 0,844 1,185 Seed investors 0,773 1,293 Crowd-funding 0,836 1,196 Incubators 0,864 1,157 Growth Full-time 0,589 1,698 Growth Part-time 0,822 1,217

Table 1; VIF and tolerance values

5.2 Descriptive statistics

5.2.1 Univariate analyses

The descriptive statistic are an attempt to summarize the sample an to give a first impression of the direction of the variables. In this section an indication of the main variables at play are given, with the use of univariate analyses. For example, table 2 shows that that both means of E&E do not differ much (2.14 & 2,24). The average age of the start-ups in this dataset is about 35 months (close to 3 years). Another thing that seems remarkable is that employee growth minima are negative; this means that the dataset contains start-up that started out with more employees than active at the time they responded to the questionnaire. The biggest employee growth total is 32, and the average is 3.56. This relatively low average might cause effects sizes having to do with this variable to be quite low (Aken et al., 2012).

(22)

22 Variabele

(coding)

Mean (standard deviation) Minimum Maximum N total

Exploitation 222,14 (0,76) 1 4,25 59 Exploration 222,24 (0,93) 1 5,67 59 Age 34,97 (17,70) 8 80,00 59

Growth

Full-time 222,25 (4,43) -3 26,00 59

Growth

Part-time 221,31 (2,03) -2 8,0 59

Growth

Total 223,56 (5,39) -3 32,00 59

Table 2: Descriptive statistics of the final variables – Mean, standard deviation, minimum, maximum and total number of respondents (N).

Table 3 shows the frequencies of the categorical questions asked in the questionnaire. This shows that Bootstrapping and the use of private debt, and private investors are relatively popular ways of funding (57,6% and 40,7% respectively), whereas crowdfunding and incubators are used less often (8,5% and 20,3% respectively). As to which sectors are represented most is are start-ups active in the “internet of things” sector.

Variabele Frequency Percent N total

Funding

Bootstrapping/private debts No 25 Yes 34

42,4%

57,6% 59

Family & Friends No 42 Yes 17 71,2% 28,8% 59 Private investors No 35 Yes 24 59,3% 40,7% 59 Seed investors No 41 Yes 18 69,5% 30,5% 59 Crowd-funding No 54 Yes 5 94,5% 8,5% 59 Incubators No 47 Yes 12 79,7% 20,3% 59

(23)

23

Variabele Frequency Percent N total

Sector

E-commerce 8 13,6%

FinTech 7 11,9%

Sharing Economy 4 6,8%

Travel & Mobility 3 5,1%

Social Local Mobile 2 3,4%

Lifestyle 1 1,7%

Internet of Things 10 16,9%

Energy 3 5,1%

Health- & MedTech 8 13,6%

Marketing- & AdTech 8 13,6%

3D 2 3,4%

EdTech 3 5,1% 59

Table 3: categorical variables

5.2.2 Bivariate analysis

In table 4 below the correlation analysis is included. Some expected correlations are insignificant. What can be seen is that exploitation and exploration are highly significantly correlated. This means that when a start-up scores high for one of these variables, it will also score high on the other. Also exploitation and the age a start-up has appear to correlate significantly, meaning that there is a direct significant relationship between age and exploitation variables. This significant correlation is absent when looking at exploration and age. Also highly significant are start-up age and the growth of full-time employees. Full-full-time employee growth and part-full-time employee growth correlate significantly, and in line with that, are the fact that the correlation of both full-time and part-time employee –growth is highly significant. Two of the control variables also appear to correlate significantly, namely that of full-time employee growth (and total employee growth), and seed investment.

(24)

24 Exploitation Exploration Age Growth

Full-time Growth Part-time Growth Total Bootstrapping/ private debts Family & Friends Private investors Seed investors Crowd-funding Incubators Exploitation - 0,435*** -0,256* -0,093 -0,059 -0,054 -0,073 -0,096 -0,057 -0,034 -0,115 -0,160 Exploration - -0,177 -90,016 -0,136 -0,064 -0,089 -0,042 -0,063 -0,143 -0,010 -0,052 Age - -0.361*** -0,086 -0,329* -0,078 -0,154 -0,107 -0,022 -0,027 -0,102 Growth Full-time - -0,294* -0,933*** -0,245 -0,028 -0,117 -0,330* -0,107 -0,096 Growth Part-time - -0,618*** -0,092 -0,145 -0,183 -0,027 -0,075 -0,160 Growth Total - -0,167 -0,031 -0,165 -0,261* -0,116 -0,140 Bootstrapping/ private debts - -0,136 -0,128 -0,102 -0,015 -0,078

Family & Friends - -0,083 -0,178 -0,210 -0,136

Private investors - -0,051 -0,244 -0,161

Seed investors - -0,063 -0,031

Crowd-funding - -0,149

Incubators -

Table 4; Overview of the correlations of the variables used in the model Correlation is significant ≤0,001 ***

Correlation is significant ≤0,01 ** Correlation is significant ≤0,05 * N=59

(25)

25

5.3 Hypotheses testing

In this section the hierarchical regression analyses are given. As a reminder, all the assumption for the regression analysis have been met, with an exception of the residuals not being normally distributed (slight S curve in P-P plot, see section 4.1 and Appendix E).

In the first regression analysis model the first hypothesis has been tested, which is formulated as follows:

A start-up that just started out will be more exploitative than a start-up that has been up and running for a longer period of time

The model shows that there is a statistically significant effect (b=-02,52 (0,22);p≤0,05) of Exploitation on age. It further shows that this effect is negative; meaning that as a start-up gets older, the degree of exploitation processes used becomes less. This in turn means that the first hypothesis can be accepted. However, when the control variables of ways a start-up has been financed and that of employee growth are added, it weakens this relationship effect and it lowers the significance level to the point that it is no longer significant. R² is quite low overall and increases slightly when the control variables are added (0,066, 0,110 and 0,115), which means that the amount of variance explained by the model is quite low and get a bit higher as the control variables are added.

Model 1 Model 2 Model 3

b (error) t b (error) t b (error) t

Intercept -2,52 (0,22) -11,68*** -2,67 (0,32) -8,46*** -2,66 (0,35) -7,89***

Age -0,01 (0,01) -2,00* -0,01 (0,01) -1,79 -0,01 (0,01) -1,69

Bootstrapping/ private debts

-0,11 (0,21) -0,50 -0,10 (0,23) -0,44

Family & Friends -0,09 (0,24) -0.39 -0,08 (0,25) -0,32

Individual investors -0,15 (0,22) -0,68 -0,17 (0,23) -0,74 Seed investors -0,02 (0,22) -0,08 -0,02 (0,25) -0,08 Crowd-funding -0,24 (0,39) -0,61 -0,25 (0,40) -0,62 Incubators -0,26 (0,26) -0,97 -0,23 (0,27) -0,86 Growth Full-time -0,00 (0,03) -0,06 Growth Part-time -0,03 (0,06) -0,49 R Square 0,066 0,110 0,115 F change 4,012* 0,421 0,141

(26)

26 Since the second model involves a different dependent variable (exploration), a different regression analysis has been made to test the second hypothesis, which is:

The results of the regression analysis for hypothesis 2, with exploration as a dependent variable and age as an independent variable, are displayed in table 6 below. The hypothesized relationship of an increase in age meaning an increase in explorative processes is not found. The results indicate that this relationship is negative, meaning that as a firm gets older, explorative processes are displayed less. This is contrary to what was hypothesized. However, this effect is insignificant. This does mean that the hypothesis 2 is rejected. No other significant effects have been found

As was the case in the first regression analysis, R² is quite low overall (0,031, 0,082 and 0,142) and increases slightly when the control variables are added, which means that the amount of variance explained by the model is quite low and get a bit higher as the control variables are added.

Intercept -2,56 (0,27) -9,52*** -2,94 (0,39) -7,48*** -3,24 (0,42) -7,67***

Age -0,01 (0,01) -1,36 -0,01 (0,01) -1,41 -0,01 (0,01) -1,84

-0,22 (0,26) -0,84 -0,40 (0,28) -1,44

Family & Friends -0,24 (0,30) -0,82 -0,26 (0,30) -0,87

Individual investors -0,17 (0,27) -0,63 -0,19 (0,27) -0,71 Seed investors -0,35 (0,28) -1,24 -0,56 (0,30) -1,86 Crowd-funding -0,20 (0,49) -0,41 -0,18 (0,48) -0,37 Incubators -0,07 (0,33) -0,20 -0,07 (0,33) -0,20 Growth Full-time -0,06 (0,04) -1,66 Growth Part-time -0,09 (0,07) -0,28 R Square 0,031 0,082 0,142 F change 1,835 0,469 1,703

(27)

27 Again, since the third model involves a different dependent variable (the balanced standardized scale of exploration minus exploitation), a different regression analysis has been made to test the second hypothesis, which is:

H3: As start-ups grow older, they become more balanced when it comes to exploitative and explorative processes.

The results of the regression analysis for hypothesis 3, with the standardized scale of exploration minus exploration as a dependent variable and age as an independent variable, are displayed in table 6 below. The hypothesized relationship of an increase in age meaning an increase in balance of

exploration and exploitation is not found. The results indicate that this relationship is negative, meaning that as a firm gets older, explorative processes and exploitative processes move further away from each other. Reminder: the balanced variable is made by extracting exploitative response variables from the explorative response variables. This is contrary to the hypothesized balance of older start-ups. Additional proof: When looking at both means ((see section 5.2.1), we can see that the mean for exploitation is 2,14, and that of exploration is 2.24. As table 5 and 6 show, both variables have a downward trend as they get older, but the trend of exploitation is bigger than that of exploration, causing a bigger unbalance. However, this effect is statistically insignificant. This does mean that the hypothesis 3 is rejected. No other significant effects have been found

As was the case in the first 2 regression analyses, R² is quite low overall (0,001, 0,084 and 0,160 and increases slightly when the control variables are added, which means that the amount of variance explained by the model is quite low and get a bit higher as the control variables are added.

Intercept -0,07 (0,29) -0,23 -0,18 (0,42) -0,42 -0,53 (0,45) -1,18

Age -0,00 (0,01) -0,25 -0,00 (0,01) -0,03 -0,00 (0,01) -0,45

-0,13 (0,28) -0,45 -0,33 (0,29) -1,11

Family & Friends -0,37 (0,32) -1,16 -0,37 (0,31) -1,17

Individual investors -0,02 (0,29) -0,08 -0,03 (0,29) -0,10 Seed investors -0,36 (0,30) -1,21 -0,60 (0,32) -1,85 Crowd-funding -0,48 (0,52) -0,91 -0,47 (0,51) -0,91 Incubators -0,35 (0,35) -1,00 -0,33 (0,35) -0,94 Growth Full-time -0,06 (0,04) -1,67 Growth Part-time -0,12 (0,07) -1,73 R Square 0,001 0,084 0,160 F change 0,064 0,765 2,217

(28)

28

5.4 Summary of results

This paragraph contains a short summary of the results of all the statistical analyses. The testing for the reliability of the questions asked regarding E&E, using Cronbach’s alpha, revealed that internal reliability of the questions and the constructs used to measure E&E was in order, all yielding a Cronbach alpha’s above 0,6.

The hypothesis stemming from the literature review have been tested with the use of multiple

regression analyses. The assumptions that have to be met in order for a regression analysis to provide accurate data (Field, 2005; Osborne and Waters, 2002; Agresti and Finlay, 2008) have all been met, with the exception of the normal distribution of the residuals. The data as a whole is normally distributed, and there is homoscedasticity and multicollinearity.

The descriptive statistics, which are used to give a first impression of the data and the directions of the variables, show that correlation is only significant between E&E, for exploitation and age, and for and seed investment as funding and employ growth (both full-time and total -growth). The predicted correlation between exploration and age appeared to be statistically insignificant.

The regression analysis used to test the first hypothesis reveals that the hypothesis is statistically significant and can therefore be accepted, meaning that as younger start-ups use more exploitation processes then older firms do. However, what is noticeable is that when the control variables of employee growth and type of funding used for the start-up are added, the results are no longer significant.

The second hypothesis, assuming that as a venture gets older it will use more explorative processes, is rejected due to opposing statistically insignificant results given by the regression analysis. The same goes for the third hypothesis, which assumes that as firms grow older they become more balanced when it comes to exploitative and explorative processes being used; statistically insignificant opposing results stem from the regression analysis, which means that the hypothesis is rejected.

(29)

29

6. Discussion

The results of the statistical analyses are discussed in this section. In table 8 below a summary of the different hypotheses and their outcomes are given: all but the first hypotheses are rejected. Speculation and discussion about the how and why is next. In this process, the findings are coupled with the theory. The section is divided in two parts: the first part contains a theoretical discussion about the different hypotheses and variables involved, and the second part gives practical implications that come forward from this research.

Hypothesis Accepted/rejected

H1

A start-up that just started out will be more exploitative than a start-up that has been up and running for a longer period of time.

Accepted (partially)

H2

A start-up that has up been running for a longer period of time will be more explorative than a start-up that just started out.

Rejected

H3

As start-ups grow older, they become more balanced when

it comes to exploitative and explorative processes. Rejected

Table 8: Acceptation of rejection of the hypotheses

6.1 Theoretical discussion

As a firm faces high market uncertainty, it tends to solidify and even balkanize the present network structure (Beckman et. al., 2004), i.e. make use of more exploitation. The reasoning is in line with the first hypothesis, since younger firms arguably face a higher market uncertainty than older firms do. However, concluding that as a firm gets older it abandons exploitative processes is wrong. This is because it might be the case that start-ups that actually have less of exploitative focus in the first place are the only ones who get to an older age. However, based on the literature, it is likely that as a start-up ages, it becomes more stable: a certain proof of concept and a customer base are in place, which lowers market uncertainty (Jones et al., 2013). This in turn lowers the need for existing network to be exploited and reinforced, and thus lowering the exploitation focus in general (Beckman et al., 2004).

The hypothesized effect of an older start-up company being more explorative than their younger counterparts was not found. Reasons could be that start-up companies already adopted a certain appropriate focus on explorative processes from the get-go, leaving this level unchanged for the rest of

(30)

30 the ventures existence. Or that the explorative processes only get more emphasis as a venture

surpasses the start-up fase.

Authors Gibson and Birkinshaw (2004) argue that exploration only occurs for short periods of time. This might also be the case for start-ups, and it might even be true that these short bursts of

exploration only happen when a venture reaches a certain age, or that they happen more often as a firm gets older. This would mean that the second hypothesis is wrongfully rejected. However this theory cannot be tested with the research design adopted in this paper (see section 7.2; limitations).

The hypothesized balance between E&E was not found. There might be different reasons as to why this has not been found. It could be that start-ups are simply too young to effectively find and actually bring about this balance. The average age if the start-ups used for this research is about three years, the balance might only occur in a later stage, when more experience and security is in place. The rejection of this hypothesis could also mean that there are no structural balancing mechanisms in place at start-up companies, since no significant differences have been found between younger and older firms. Balancing mechanisms in start-up companies could possibly be more temporal in nature, where cycles of E&E exist. These cycles, with short bursts of exploration and long periods of exploitation (Gupta et al., 2006; Gibson and Birkinshaw; 2004), would be hard (if not impossible) to detect with the research design adopted in this paper, explaining the rejection of the hypothesis.

A so called “hybrid structure” might allow for achieving a high degree of both E&E at the same time and place (Csaszar, 2013). A start-up company might simply be too small to be able to put such a hybrid structure in place. Gupta et al. (2006) argue that radically different knowledge structures make a simultaneous pursuit of both very difficult (Cegarra-Navarro & Sánchez-Vidal, 2011). Raisch et al. (2009) suggest that an organizational separation is needed in order to successfully accomplish balancing exploration and exploitation, with organizational units exclusively dedicated to one of both activities (O’Reilly and Tushman, 2004; Lavie et al., 2011). However, authors reasoning for this dual governance are generally addressing large firms. It is not known, and arguably a bit unlikely, that the same can be accomplished by small ventures (i.e. start-ups).

6.2 Practical implications

Firm founders might benefit from being aware that putting too much emphasis on exploitation processes might cause them to be unbalanced. Being balanced is important to improve firm survival and both financial and innovative performance, e.g. a firm benefits both in the short and the long term (March, 1991; Belderbos et al., 2010; Ugur, Belderbos, Leten and Kelchtermans, 2013; Jansen et al., 2006; He & Wong, 2004; O’Reilly and Tushman, 2004; Raisch et al., 2009). An important notion that is that of Gupta et al. (2006), who argue that organizational members might grow used to exploitation,

(31)

31 limiting their critical abilities and narrowing their definitions of problems. Having an focus on

exploitation for longer periods of time causes organizational members to be stuck in a paradigm, limiting future problem solving abilities. Additionally, an exploitation focus might lead to an ignorance of exploration, which hinders long-term performance; the so called “success trap” (Levinthal and March, 1993; Ugur, 2013).

On the other hand, a focus on solely explorative processes can possibly lead to an ignorance of exploitation, causing short-term problems (cash targets, profitability); the so called “failure trap” (Levinthal and March, 1993; Ugur, 2013).

How can this be battled? Cegarra-Navarro and Sánchez-Vidal, 2011 argue that unlearning outdated knowledge and relearning updated knowledge is very importing when trying to achieve an appropriate balance between E&E processes.

Start-ups can encourage exploration processes by setting up both formal and informal meetings, or with the creation of external communities of practice where both customers and sellers are able to work together and interact on a particular objective (Bagozzi and Dholakia, 2006, Cegarra-Navarro and Sánchez-Vidal, 2011). An internalization of such knowledge can be achieved by organizational members developing relational trust, common language and confidence to materialize it (Selnes and Sallis, 2003, Cegarra-Navarro and Sánchez-Vidal, 2011). Increasing the capacity of a start-up in order to create new or enhanced products/services, which is the main aim of exploitation processes (Bierly and Daly, 2007), can be done by fostering commitment, install training programs, and focussing on using what has already been learned within the start-up (Macdonald, Assimakopoulos and Anderson, 2007, Cegarra-Navarro and Sánchez-Vidal, 2011).

(32)

32

7. Conclusion

In this section the final conclusions are given, answering the research question proposed in the

introduction. After that, the limitations of this research are given. The final part of this section contains suggestions for future research.

7.1 Findings

To recap, the research question proposed in the introduction is repeated, namely:

How do the exploitation and exploration processes in a start-up develop over time?

This research has shown that younger firms have a significantly higher focus on exploitation processes than older firms. Furthermore, it has been found that no significant changes take place when it comes to the focus on explorative processes. The same goes for the balance of both E&E processes; no significant differences have been found between younger and older when it comes to the balance of E&E processes, suggesting that there is no noticeable balancing (or unbalancing) effect.

This research contributes to the existing literature because it has zoomed in at the very early stages of ventures, shedding light on the birth and the origins of E&E processes. In the field of

entrepreneurship, the E&E view can possibly contribute to answering the many questions that this young research discipline still has to ask. Also, the novel approach to testing pre-existing assumptions about E&E processes in the field of starting and developing ventures only can possibly prove

previously held understandings wrong, and deepen the understanding of how these processes actually work.

7.2 Limitations

Overall, the number of respondents in total is quite low. This lowers the power of the research, since it causes the change of a Type II error quite high (VanVoorhis and Morgan, 2007), which means failing to reject a null hypothesis if it is actually wrong, i.e. finding significant results can prove to be difficult. Also, the low number of respondents makes drawing generalizations problematic.

Respondents were asked to indicate in which industry they were active (see appendix A and B). This was done in attempt to control for this factor. However, some respondents said that they could not find a suiting label for their industry. For example; one person replied that he/she did not fill out the survey, because “chemistry-industry” was missing. This might have led to an increase in the dropout rate (recap: 83 respondents, but 24 questionnaires were left out because they were not completely

(33)

33 filled out). Adding to this, the fact that apparently some industries were missing might have negatively influenced this variable, and this is why it is not used in the analysis. In hindsight, making a more extensive industry answering option list, or simply adding an answer option saying “other, namely …”, might have solved this.

Moving away from trying to answer the question of why some questionnaires were partially filled out, this fact on itself might be a threat to the reliability and the validity of this research. The somewhat low response rate to the emails is a point of attention as well: it could very well be that the respondents used for this research are not representative for the average start-up founder. The reasons why about 17% of the people contacted decided to respond, and why approximately 83% decided not to do so, are unknown and could therefore be a threat to the validity and the reliability of this research (Pelham and Blanton, 2012).

The fact that most of the respondents were active in the “internet of things”, and several other industries were perhaps underrepresented (see section 5.2.1) might have caused, or be a sign of, a selection bias (Pelham and Blanton, 2012), which possibly tainted the data. Generalizations are therefore hard to draw.

The fact that this questionnaire was just a one of event with no follow-up questionnaires sent also limits this research. Comparisons between start-ups have been made, because of this the wrongful assumption that the older a firm gets the more it abandons exploitative processes is easily made (see the theoretical discussion section 5.1 for a more elaboration on this subject).

Residuals are not normal and independently distributed with a (some) constant variance. This possibly means that amongst the residuals resides a structure that is not accounted for in the model. Identifying this structure and adding term(s) that represent it in the originals model might will lead to a better model (Beckman et al., 2004). However, the way the questionnaire was made up, the limited data and the low number of respondents made this not possible in this research. The low number of respondents is a likely cause for this violation (Beckman et al., 2004).

Growth is measured by looking at the number of employees. This might not be the best measure of venture growth, but since for this research surveys were used, it appeared to be the best option. Asking for revenue growth or turnover growth etc. would likely have a negative influence on the response rate of the questionnaire.

7.3 Future research suggestions

The notion that cycles exist, where longer periods of exploitation are followed up by short bursts of exploration, cannot properly be studied with this research design. Extensive qualitative research is

(34)

34 needed in order to answer questions like that. Qualitative studies using multiple measures (follow-up questionnaires) at different points in time might also prove to be insightful. This way actual

differences and developments within start-up companies themselves can be revealed, instead of the mere comparison between different start-ups which has been done in this research.

This research might serve as a tiny piece of the puzzle as to looking for the reasons why so many start-ups fail in the early stages of their development. Given the fact that young firms focus significantly more on exploitation, temporal studies could unveil if this dominant exploitation focus can be linked to the high mortality rates amongst start-up companies. Are unbalanced firms being outperformed by more balanced firms? Do the start-ups that have an balance E&E from an early age survive more often than those that do not? Can a venture’s E&E focus serve as a prediction as to future success/failure? Etc. Specifically linking financial performance to E&E in start-ups is also quite an interesting future research endeavour.

Sectoral/industry differences can also be brought to light by intensive research within one specific sector or industry, or by performing research on a bigger scale than this research. Entry and exit rates of a specific industry might have a big influence on which E&E processes are adopted; linking these rates to E&E might bring certain differences and relationships to light.

(35)

35

8. References

8.1 Articles

Aldrich, H. E., & Kenworthy, A. (1999). The accidental entrepreneur: Campbellian antinomies and organizational foundings. Variations in organization science: In honor of Donald T. Campbell, 19-33.

Alvarez, S. A., & Barney, J. B. (2007). Discovery and creation: Alternative theories of entrepreneurial action. Strategic entrepreneurship journal, 1(1‐2), 11-26.

Alvarez, S. A., Barney, J. B., & Anderson, P. (2013). Forming and exploiting opportunities: The implications of discovery and creation processes for entrepreneurial and organizational research. Organization Science, 24(1), 301-317.

Atuahene-Gima, K. (2005). Resolving the capability—rigidity paradox in new product innovation. Journal of marketing, 69(4), 61-83.

Bagozzi, R. P., & Dholakia, U. M. (2006). Antecedents and purchase consequences of customer participation in small group brand communities. International Journal of research in Marketing, 23(1), 45-61.

Baker, T., & Nelson, R. E. (2005). Creating something from nothing: Resource construction through entrepreneurial bricolage. Administrative science quarterly, 50(3), 329-366.

Baron, J. N., Hannan, M. T., & Burton, M. D. (1999). Building the iron cage: Determinants of managerial intensity in the early years of organizations. American sociological review, 527-547.

Belderbos, R., Faems, D., Leten, B., & Looy, B. V. (2010). Technological activities and their impact on the financial performance of the firm: Exploitation and exploration within and between firms. Journal of Product Innovation Management, 27(6), 869-882.

Beckman, C. M., Haunschild, P. R., & Phillips, D. J. (2004). Friends or strangers? Firm-specific uncertainty, market uncertainty, and network partner selection. Organization science, 15(3), 259-275.

Benner, M. J., & Tushman, M. L. (2003). Exploitation, exploration, and process management: The productivity dilemma revisited. Academy of management review, 28(2), 238-256.

Bierly, P. E., & Daly, P. S. (2007). Alternative knowledge strategies, competitive environment, and organizational performance in small manufacturing firms. Entrepreneurship Theory and Practice, 31(4), 493-516.

(36)

36 Blome, C., Schoenherr, T., & Kaesser, M. (2013). Ambidextrous governance in supply chains: The impact on innovation and cost performance. Journal of Supply Chain Management, 49(4), 59-80.

Boumgarden, P., Nickerson, J., & Zenger, T. R. (2012). Sailing into the wind: Exploring the relationships among ambidexterity, vacillation, and organizational performance. Strategic Management Journal, 33(6), 587-610.

Cader, H. A., & Leatherman, J. C. (2011). Small business survival and sample selection bias. Small Business Economics, 37(2), 155-165.

Cao, Q., Gedajlovic, E., & Zhang, H. (2009). Unpacking organizational ambidexterity: Dimensions, contingencies, and synergistic effects. Organization Science, 20(4), 781-796.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. psychometrika, 16(3), 297-334.

Crossan, M. M., Lane, H. W., & White, R. E. (1999). An organizational learning framework: From intuition to institution. Academy of management review, 24(3), 522-537.

Csaszar, F. A. (2013). An efficient frontier in organization design: Organizational structure as a determinant of exploration and exploitation. Organization Science, 24(4), 1083-1101.

Cunliffe, A. L. (2010). Crafting qualitative research: Morgan and Smircich 30 years on. Organizational Research Methods.

Gabriel Cegarra-Navarro, J., Sánchez-Vidal, M. E., & Cegarra-Leiva, D. (2011). Balancing

exploration and exploitation of knowledge through an unlearning context: An empirical investigation in SMEs. Management Decision, 49(7), 1099-1119.

Gartner, W. B. (1985). A conceptual framework for describing the phenomenon of new venture creation. Academy of management review, 10(4), 696-706.

Gibson, C. B., & Birkinshaw, J. (2004). The antecedents, consequences, and mediating role of organizational ambidexterity. Academy of management Journal, 47(2), 209-226.

Gorman, G., Hanlon, D., & King, W. (1997). Some research perspectives on entrepreneurship education, enterprise education and education for small business management: a ten-year literature review. International small business journal, 15(3), 56-77.

Gupta, A. K., Smith, K. G., & Shalley, C. E. (2006). The interplay between exploration and exploitation. Academy of management journal, 49(4), 693-706.

Orientation evolution in start-ups : exploitation and exploration processes reviewed