Failure, the road to success : the relation between the stage of a startup, the presence of incubators/accelerators and the presence of licenses/patents and technological, operational and market failure reasons

(1)

Failure, the road to Success

Name: Stijntjes, Victor (V.J.L.)

Student number: 2605616 (VU) and 11431822 (UVA) Thesis supervisor: dr. N.A. Thompson

Date: August 8th, 2018 Word count: 12,515 MSc Entrepreneurship

Joint Degree: VU Amsterdam (VU) and University of Amsterdam (UVA)

The relation between the stage of a startup, the presence of incubators/accelerators and the presence of licenses/patents

and

(2)

Statement of originality

This thesis is written by Victor Stijntjes.

I declare to be fully responsible for the contents of this document, including mistakes. All text is originally written by me and the sources used are mentioned in the text and in the references. The VU University Amsterdam (VU) and the University of Amsterdam (UVA) are

(3)

Failure, the road to Success Thesis MSc Entrepreneurship 2018. 2

Abstract

Entrepreneurs more often fail than succeed. However, most entrepreneurship research is biased towards studying successful cases rather than difference reasons for failure. This thesis contributes to the body of entrepreneurial literature by answering the question of how the

stage of the startup, the presence of an incubator/accelerator and the presence of

licenses/patents correlate with the likelihood of experiencing technological, operational or

market-based failure reasons.

In the line with literature, I hypothesize that the likelihood to experience a technological failure reason is negatively related to the stage of the startup, while the likelihood to experience a market failure reason is positively related to the stage of startup. Furthermore, I hypothesize that incubators/accelerators and licenses/patents decrease the likelihood to experiencing a technological and/or operational failure reasons, while the likelihood of experiencing a market failure increases by the presence of an

incubator/accelerator.

Despite the literature based conceptual model, the logistic regression models in this thesis turn out to be insignificant. Therefore, I was not able to confirm the hypothesis as constructed in the literature review. Nevertheless, by creating a better understanding about failure reasons and their learning benefits, this thesis contributes to the study of failure and known failure reasons in entrepreneurship.

Keywords: Entrepreneurship, failure reasons, learning benefits, stage of startup,

(6)

Chapter 1: Introduction

‘’It’s fine to celebrate success, but it is more important to heed the lessons of

failure.’’

Bill Gates, Founder of one of the most successful startups the world; Microsoft.

There is no better way to introduce the importance of failure than to learn from one of the world’s most influential people. Bill Gates has become famous and successful due to his business mentality and his startup; Microsoft. Bill gates, and many more, are also partly responsible for all the positive connotations around the concept of entrepreneurship (Perren and Jennings, 2005; Kenny and Scrive, 2012; Stevens and Burley, 1997, Sorensen, 2008). Not only these financial success stories tend to attract people to go for an entrepreneurial

adventure, the optimism regarding risk taking within the entrepreneurial community contributes also to the overall utility that people gain from their entrepreneurial projects (Harrington 2010; Surowiecki, 2014). Thereby, entrepreneurs seem to enjoy the creating and controlling aspects of having their own business (Wasserman, 2008).

However, like this quote also mentioned, startups are not assured of a positive

outcome. Failure turns out to be more common in the world of entrepreneurship than success. According to Juan Carlos Domezain (2018), 75% of the startups fail within two years. The failure aspect of the entrepreneurial process is important to investigate because failure is the foundation for learning. Thereby, failed startups might also be seen as useful for the economy since they are starting points of new opportunities (Coad, 2013; Cope, 2011). For this reason, failure appears not to be harmful for the economy. Next to that, factors of success are mostly difficult to pinpoint, failure reasons on the other hand are easier identified (Sitkin, 1992). By investigating failed startups, we contribute to the body of literature and we might be able to avoid known failure reasons in the future, and we might resolve uncertainty in the

entrepreneurial community (Baker et al., 1997; Kets de Vries, 1985; McGrath, 1999; Reynolds, 1987; Romanelli, 1989).

In this thesis I will especially focus on three different factors that might correlate with the likelihood to experience these failure reasons. The failure reasons are categorized as

technological, operational, and market failure reasons and might all separately be responsible

for the discontinuation of operations. The first two failure reasons are in control of the entrepreneur because he or she is personally responsible for the product/service quality and the operational aspects of the startup. Market failure reasons are not in control of the

(7)

entrepreneur since these reasons depend on market related factors like governmental decisions, or natural disasters (Bruno et al., 1987; Gaskill et al., 1993; Everett and Watson, 1998).

The three independent variables that will be investigated are the stage of the startup

development, the presence of an incubator/accelerator, and the presence of licenses/patents.

These factors are chosen because they are main topics within the entrepreneurial literature. Hill et al. (2002) elaborates on the five-stage model from Churchill and Lewis (1983) by including different barriers that entrepreneurs experience in particular stages of the startup development. Hill et al. (2002) suggests that the leadership role of an entrepreneur changes over the stages from functional to strategic. I will elaborate on this theory by investigating whether the relation between particular failure reasons and the different stages changes during the firm-lifecycle. I expect that entrepreneurs will experience more technology related failure reasons in the early stages because in the early stages the product/service has to prove his value. Consequently, in the later stages I expect that the entrepreneur will experience more market related failure reasons because the startup success will then depend more market demand.

For the incubator/accelerator variable I make use of the constructs described by Lumpkin (1988) and Miller and Bound (2011). They developed theories that suggest that environments are created by incubators/accelerators that increase the probability of success for startups. However, these constructed environments are positively influencing the aspects that are in control of the entrepreneur (National Business Incubation Association, 1985). Therefore, the entrepreneurs that fail with the presence of an incubator/accelerator will not fail due to technological or operational failure reasons but, will mainly experience market failure reasons since these aspects are not in the control of the entrepreneur.

Finally, this paper goes beyond the classical view of the protection aspects of

licenses/patents and elaborates on the correlation between the presence of licenses/patents and the likelihood of experiencing particular failure reasons (Long, 2002). Startups who are in possession of licenses/patents are expected to have a high growth potential. Thereby, the time for these startups to obtain a venture capital investment reduced by 76%. The quality signals and the investment opportunity for investors, makes it possible it for entrepreneurs to get in contact with external advisors. This improves the access to information and advice, which improves the startups’ ability to deal with problems that can be controlled. Therefore, they

(8)

will experience less technological and operational related failure reasons (Haeussler et al., 2009).

By testing the influence of different independent variable on the likelihood to

experience certain failure reasons, this research contributes to the entrepreneurial literature as different relations might be confirmed. The regression results are derived from the data received from the Failure Institute (2017). The data is used to perform a logistic regression that might offer new insights which can contribute to the literature on failure reasons. The research method is different from most qualitative research on failure reasons that are based on multiple-case studies because it provides insights in the relative effect of different variables studies (Eisenhardt 1989; Eisenhardt and Graebner, 2007). This implies that the findings might suggest that the presence of an aspect has a particular effect on the likelihood to experience the dependent variable (the particular failure reason). These findings therefore perfectly link to the research goal of this paper, which is, as stated before, to investigate the presence of different factors on the likelihood to experience particular failure reasons.

The findings will be discussed in light of prior literature in the discussion part of this paper. By performing a combination of qualitative and quantitative research, the intermediate state of the literature can be improved since proposed relations will be tested. As argued by Edmondson and McManus (2007), this research design fits best for the state of intermediate theory. The role of incubators/accelerators and the role of licenses/patents are examples of subject that are still not fully understood (Lumpkin, 1988; Miller and Bound, 2011; Mann, 2005). Creating a better understanding about these subjects will improve our capability to deal with potential failure reasons that are related to these factors.

All this considered, I have constructed the following research question to test the correlation between the three dependent variables, and the three independent variables;

To what extent do different stages of startup development, the presence of

incubator/accelerator, and the presence of licenses/patents, correlate with the likelihood of experience technology, operational and, or market failure reasons?

This first chapter is used to introduce the topic of this thesis. In the second chapter the literary foundation for the research question will be argued. This literary foundation is used to construct different hypotheses, that will be tested in this thesis. In the Chapter 3 the

methodology will be described including the conceptual framework based on the literature. Besides, the methodology contains a description of the core statistics, the research design and

(9)

an explanation of the data collection and transformation. In Chapter 4 the results of the logistic regression models will be described and explained. Finally, in Chapter 5 the results will be discussed and argued in light of the literature. Thereby, the limitations and

(10)

Chapter 2: Literature review and hypothesis development

This literature review defines the topic of this thesis, failure reasons and the three

independent variables, the stage of the startup, the presence of an incubator/accelerator, and the presence of licenses and patents, that likely correlate with the failure reasons. This literature review argues the research questions and proposes the hypotheses for this thesis. With this thesis I seek to contribute to the body of literature of failure reasons in

entrepreneurship in order to gain more understanding of the learning benefits of failure in the entrepreneurial discourse.

2.1 Entrepreneurial failure

In existing literature, a wide variety of definitions is used to describe the concept of failure (Gimeno et al., 1997; Juan Carlos Domezain, 2018; Olainson and Sorensen, 2014;

Weerawardena and Sullivan, 2006). One example of a definition for the concept of failure is given by McGrath (1999): ‘’failure is the termination of an initiative that has fallen short of

its goals’’. However, equally as in the paper of Lussier (1996), it is not the goal of this paper

to form a generally accepted set of reasons why business fail. Instead, the goal of this research paper is to create a better understanding of failure reasons.

By learning from these failure reasons, the capability to deal with these potential failure reasons can be improved. As argued by McGrath (1999), we therefore might even celebrate the failure of business in the future. Olain and Sorensen (2014) contribute to this view by defining failure as a learning process for the individual entrepreneur and as a contribution to the economic growth of society.

According to previous studies of McGrath (1999), Shepherd (2003), and Zacharakis et al. (1999), the entrepreneurial literature has not paid enough attention to the learning benefits of entrepreneurial failure. On top of that, most research papers are based upon multiple-case studies instead of building on broad samples of failed entrepreneurs (i.e. Eisenhardt, 1989; Eisenhardt and Graebner, 2007). For this reason, this research will contribute to the body of entrepreneurial literature by performing a quantitative research to improve the understanding on certain failure reasons.

Various researchers have acknowledged that more research is needed within the field of entrepreneurship on entrepreneurial. It is argued that numerous questions about the role of different factors are still unanswered. For example, the role of incubators/accelerators has

(11)

changed a lot since Lumpkin (1988) wrote a paper about their role in startup development. Puzzles that are not fully understood are the way how incubators/accelerators could be used as an economic development tool, and what the long-term effects are on individuals (Miller and Bound, 2011). Another subject that needs more research is the role of licenses and patents. For instance, the signaling role of licenses/patents might be different across different sectors (Mann, 2005).

Overall it seems evident that the capability to deal with potential failure reasons can be improved when there is a better understanding of these subject and their influence on

entrepreneurial failure. 2.2 Failure reasons

Business failure reasons is examined in different studies (Berryman, 1983; McMahon et al., 1993). According to Everett and Watson (1998) business failure is mainly caused by endogenous (internal) factors, and only one third of business failure is caused by exogenous (external) factors. The endogenous causes for failure are particularly based on managerial, financial, and product/service related problems (Bruno et al., 1987; Gaskill et al., 1993). These factors are all in control of the entrepreneur. The external causes are less predictable since they depend on factors where the entrepreneur has no control over, like a recession in the economy that influences the purchase power of customers, governmental decisions that influences the market, or for example a natural disaster that makes the daily life impossible (Everett and Watson, 1998).

For the purpose of this paper I will divide the failure reasons in two endogenous factors and one exogenous factor. The two endogenous factors are defined as technological

failure reasons and operational failure reasons, both in control of the entrepreneur. The

technological failure reasons refer to quality of the product/service and the technology applied, whereas the operational failure reasons refer to the managerial and financial

decisions. Thornhill and Amit (2003) confirm these operational failure reasons, since starting entrepreneurs are regularly not in the possession of the adequate skills to make the best operational decisions for the startup. These inadequate skills are causing problems within the startup and might lead to overall failure (FEE, 2004). The exogenous factor will be defined as a market failure reason and refers to factors that are not in control of the entrepreneur. Other market related factors then mentioned in the previous paragraph are changing buying patterns,

(12)

shortage of raw materials, and for example substitution products that change preferences of consumers (FEE, 2004).

In this thesis I will test the likelihood to experience one of the three different failure reasons as mentioned in the previous paragraph, in correlation with three independent variables (different stages, presence of incubator/accelerator, presence of licenses/patents). The main research question in this research therefore is defined as:

To what extent do different stages of startup development, the presence of

incubator/accelerator, and the presence of licenses/patents, correlate with the likelihood of experience technology, operational and, or market failure reasons?

2.3 Hypothesis Development

2.3.1. The stage of the startup and relationship with failure reasons

Since there is some predictability within the pattern of organizational development, it is beneficial to use a lifecycle stage model. The lifecycle stage model provides a framework with useful insights in specific problems in specific stages of the firm development (Hill et al., 2002). There are many different lifecycle frameworks developed with numerous of different stages, however, I will use the model developed by Churchill and Lewis (1983) which is widely accepted (Hill et al., 2002; Yusuf, 1997). This model contains five-stages in the lifecycle (Churchill and Lewis (1983)):

1. Existence; 2. Survival 3. Success 4. Take-off; and 5. Resource mature

In the first three stages of the startup the return on effort is usually low and the focus is mostly on the vision of the proposed goal of the startup. Thereby the marketing focus of the startups will shift in the later stages from a more functional level to a more strategic level. Important in all stages is that the entrepreneur has a proactive attitude instead of a reactive attitude to gain and hold market share (Tyebjee et al., 1983). The study of Hill et al. (2002) concludes that this proactive approach is difficult to remain in the later stages of the firm development because an entrepreneur will experience a loss of focus on sales. In this study I will call this proactive approach the operational aspect of the entrepreneur.

(13)

Another contributing view on the lifecycle framework is made by Stevens and Burley (1997). From their research I conclude that the risks experienced in the development of the startup are a combination of technology and market risks. In the early stages of the startup it has to prove its product/service to the market, which depends on the technology, on the know-how, of the product/service. In the later stages, when the product has proven to add value, it will be sold to the greater public. Therefore, the entrepreneur has to be aware of the demand changes and therefore he must act more strategical towards to the market.

Consequently, from the literature I hypothesize that the problems that entrepreneurs face in the early stages are more related to the technology risks of the product/service and in the later stages the problems are more related to the market risks. Thereby, an important aspect is the personality of the entrepreneur. The dominant characteristics play an important role to prevent and to overcome periods of uncertainty. His leadership and management abilities, and especially his willingness to take risks in uncertain periods, independent on the development stage, makes huge differences in the ability of a firm to survive the periods of stagnation due to potential failure reasons (Hill et al., 2002). Thereby might ‘’autocratic leadership’’, and the prevention of participation in the decision-making process lead to bad management (Bruno et al., 1997). Bad management and operational decisions might cause business failure (FEE, 2004). For the purpose of this paper I will therefore include these aspects of personality related failure reasons in the operational failure reasons.

In this research paper I will test the failure reasons as constructed in the previous section in the different stages according to the lifecycle model. Thereby, I create a more comprehensive insight in the different stages and the failure reasons that entrepreneurs experience.

Hypothesis 1:

In the early stages of startup development entrepreneurs are more likely to experience technology related failure reasons, than market failure reasons, while in the later stages the

experience is the opposite

2.3.2. Presence of an incubator/accelerator and the relationship with failure reasons

Incubators are stimulating new business startups and they reduce the high failure rates within entrepreneurial projects (Lumpkin, 1988). Their role is defined by Lumpkin (1988) as; ‘’Business incubators have been organized to bring new businesses together to increase the

(14)

probability of success.’’. Other definitions of incubators are ‘’organizations that are designed specifically to attract and aid new businesses’’ (Smilor and Gill, 1984; Allen, 1985), ‘’an organized effort to bring together new and emerging businesses in a controlled environment’’’ (National Business Incubation Association, 1985). The main advantages of incubators are that they offer below-market-rate space on flexible terms (Money, 1984), they eliminate building maintenance obligations (Temali and Campbell, 1984), they increase the startup’s visibility (Dollinger, 1985), they provide equipment and services that are unaffordable for starting entrepreneurs (Bryan, 1984), and that they create an environment where startups can cooperate, thereby reducing the anxiety of starting new businesses (De Noble and Moliver, 1983).

Accelerators are apart from business incubators because they offer time-limited programs. Within these time-limited programs, entrepreneurs experience intensive mentoring and they work together in small teams rather than on an individual basis. Besides, the

application process of accelerators is open to all startups, which makes it highly competitive (Miller and Bound, 2011).

In contrast to the differences of both approaches, they both have the overall mission to increase the probability of success for the entrepreneurs (Lumpkin, 1988; Miller and Bound, 2011). Both approaches offer input from experienced advisors and develop conditions and support that will ensure successful business operations (Lumpkin, 1988).

Therefore, I assume that startups working with the presence of an

incubator/accelerator, face less difficulties in the technology and operational aspects of their business, and have relatively more failure reasons regarding the market risks since they have no control over these potential failure reasons. Therefore, I will test in this research paper whether the presence of an incubator/accelerator changes the failure reasons of the startups.

Hypothesis 2:

Startups working with the presence of an incubator/accelerator, are more likely to experience market related failure reasons than technology/operational related failure reasons

2.3.3. Presence of licenses and patents and the relationship with failure reasons

With this aspect I will contribute to the growing body of literature on the importance of patents and licenses for startups (Haeussler et al., 2009). The classical and best-known reason for startups to patent/license their product/service is to protect their intellectual property

(15)

(Long, 2002). However, patents/licenses are more than only a protection of intellectual property. According to Haeussler et al. (2009) ‘’patents are known to help companies to appropriate returns from investment in R&D, facilitate the commercialization of technology and shorten time to IPO’’. Thereby patents/licenses are quality signals for venture capitalists to invest in startups. Startups having at least one patent application, experience a time reduction of 76% for the first venture capital investment (Haeussler et al., 2009). Another effect of patents, is the barrier they create for new startups to enter the same market. The higher the density of patens, the more difficult it is for other firms to enter the market, so this increases the competitive advantages of startups with a patent in these markets (Cockburn and MacGarvie (2007).

Reduced time for a venture capital investment, better protected investment in R&D, and the competitive advantage of startups with patents/licenses increases the growth potential and the assumed overall quality of startups. The increased growth potential and the greater likelihood of venture capitalist that are interested in the startup, will lead subsequently to a higher possibility of information input from more external resources and advisors. This could lead to improvements of the technological aspect of the product/service of the startup

(Haeussler et al., 2009)

Consequently, patents and licenses increase the growth potential of startups due to their access to external (financial) resources and the competitive advantage they obtain with a patent/license. These aspects decrease the likelihood of experiencing technological and operational failure reasons. So, according to this argumentation I hypothesize:

Hypothesis 3:

Startups with licenses and, or patents, are less likely to experience a technological or operational related failure reason than those without one

(16)

Chapter 3: Methodology

In this section I will explain my choice of the methodology and the research context. In addition, I will explain how this data is categorized in useful variables according to the research purpose and how the key concepts are defined.

3.1 Research design

Prior research on entrepreneurial failure proposes relationships between new and established constructs which is typically for an intermediate state of literature. Edmondson and McManus (2007) argue that the best way to investigate the intermediate theory is by performing both qualitative and quantitative research. The qualitative research part is exposed in the literature review and is used to investigate the proposed relationships between the different variables. Subsequently, I have retrieved a dataset from the Failure Institute (2017) that is used for to perform the quantitative part of this research. They built a cross-sectional dataset by issuing questionnaires to more than 1000 unique failed entrepreneurial projects in more than 40 different countries. In line with the reasoning of Edmondson and McManus (2007), a quantitative research fits to the purpose of to test the proposed relations and to create a provisional theory that integrates the prior research work.

In this paper, a binary logistic regression model is used to determine the relative effect of several independent variables on a dependent variable. Consequently, a logistic test is also particularly suited to test dichotomous dependent variables (Lammers et a., 2007)

Lastly, in line with the survey of the Failure Institute (2017) and their intentions, this thesis defines failure as the discontinue of operations due to predominant failure reasons. 3.2 Conceptual and regression model

The conceptual model to the different failure reasons is visually exposed in Appendix A. The conceptual model is based upon the relations described in the Literature Review. The bottom line is that three different dependent variables (technological failure reason (TFR),

Operational Failure Reason (OFR), Market Failure Reason (MFR)), are tested with three

different independent variables (different stages of startup development (Stage), presence of

incubator/accelerator (Incub), presence of licenses/patents (Lic)) to determine whether the

previous composed hypotheses are true or false. In addition, the supportive control variables are gender (Gender), a distinction between the business type goods or services (Type), the

(17)

business sector in which the startup was operating (Sector), and the country (Country) where the startup was based.

In accordance to the conceptual model, there have been constructed three different regression equations that will be used to test the hypotheses:

• ln𝑇𝐹𝑅𝑦𝑒𝑠 𝑇𝐹𝑅_𝑛𝑜 = ∝0+ 𝛽1𝑆𝑡𝑎𝑔𝑒 + 𝛽2𝐼𝑛𝑐𝑢𝑏 + 𝛽3𝐿𝑖𝑐 + 𝛽4𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽5𝑇𝑦𝑝𝑒 + 𝛽6𝑆𝑒𝑐𝑡𝑜𝑟 + 𝛽7𝐶𝑜𝑢𝑛𝑡𝑟𝑦 + 𝜀 • ln𝑂𝐹𝑅𝑦𝑒𝑠 𝑂𝐹𝑅_𝑛𝑜 = ∝0+ 𝛽1𝑆𝑡𝑎𝑔𝑒 + 𝛽2𝐼𝑛𝑐𝑢𝑏 + 𝛽3𝐿𝑖𝑐 + 𝛽4𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽5𝑇𝑦𝑝𝑒 + 𝛽6𝑆𝑒𝑐𝑡𝑜𝑟 + 𝛽7𝐶𝑜𝑢𝑛𝑡𝑟𝑦 + 𝜀 • ln𝑀𝐹𝑅𝑦𝑒𝑠 𝑀𝐹𝑅𝑛𝑜 = ∝0+ 𝛽1𝑆𝑡𝑎𝑔𝑒 + 𝛽2𝐼𝑛𝑐𝑢𝑏 + 𝛽3𝐿𝑖𝑐 + 𝛽4𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽5𝑇𝑦𝑝𝑒 + 𝛽6𝑆𝑒𝑐𝑡𝑜𝑟 + 𝛽7𝐶𝑜𝑢𝑛𝑡𝑟𝑦 + 𝜀

The overall correlation is that the likelihood of experiencing a specific failure reason (technological, operational, market) correlates with seven independent variables.

All variables are formed by the dataset provided by the FI (2017). As stated before, contains this unique dataset more than 1000 individual surveys filled in by entrepreneurs with failed businesses1. The dataset is converted to a dataset with relevant variables. In the next section I will explain how each variable used for this research is built up from the dataset of the FI (2017).

3.3 Explanation variables

3.3.1 Dependent variables

The TFR, OFR, and the MFR that are derived from the literature are not used in the same way by the FI (2017) in their survey. Instead, they used 11 different failure reasons for the possible causes of failure. These 11 different failure reasons from the survey have been categorized in the three failure reasons used in this thesis. TFR are concerning the aspects of the

product/service itself. OFR are reasons that have to do with the overall operational capabilities of the entrepreneur, the management of the firm and the personality of the

entrepreneur (e.g. ‘’Entrepreneur and associates’ business conviction’’). Both TFR and OFR are variables that are in control of the entrepreneur himself. The MFR variable contains the factors where the entrepreneur has no control over (e.g. ‘’Politicalissues’’ and/or ‘’Shifts in

(18)

market’’). In Appendix C is a detailed explanation given about the categorization of the

different failure reasons (FI, 2017).

3.3.2 Independent variables

3.3.2.1 Stage of startup development

In both the survey of the FI (2017) as in the literature there has been made use of a five-stage lifecycle model. Despite the fact that the stages have different names, I argue that the same interpretations are used for the stages in both cases2. The stages are ranked in order of existence, where the existence stage is the first in order, and resource mature is the last in order. This classification of the stages makes this Stage variable an ordinal variable.

3.3.2.2 Presence of an incubator/accelerator

The independent variable regarding the presence of an incubator/accelerator is easily adoptable from the dataset of the FI (2017). Namely, the dataset shows that there has been made use of support of an accelerator or incubator, or not. This makes this variable a dummy variable (1 = yes; 0 = no).

3.3.2.3 Presence of licenses/patents

Comparable to the previous variable, is the presence of licenses/patents a dummy variable too. The dataset made clear if firms were in possession of a license/patent, thereafter I made the dummy variable whether they were in possession (1) or not (0).

3.3.3 Control variables

Since the control variables are not in interest of this research they will briefly be described with an argumentation of their value in the regression. Gender is a control variable that controls for differences between men and women. This is control variable is important since the risk preferences between men and women are proved to be different (Croson and Gneezy, 2009). The second control variable controls for the business that is provided (goods or

services). Since service providers are not selling physical products, this variation needs to be included in the regression models. The third control variable controls for the business sector where the startup was operation. Due to the big differences between business sectors, it is important to control for these differences (e.g. Healthcare vs. Manufacturing). Lastly, a control variable concerning the country where the startup was based is added to the

(19)

regression. It is important to control for these cross-country differences since the culture, manners, and for example the way that countries are governed might influence the startup success (Mueller and Thomas, 2000).

3.4 Descriptive statistics

3.4.1 Core statistics

In Table 1 the descriptive statistics of all variables are exposed. As described before, there are three different dependent variables, and therefore three different regressions. The means shows the percentage of entrepreneurs that experienced a specific problem failure reason. Subsequently, it turns out that from the 1105 observations, 7% of the entrepreneurs experienced a TFR, 80% experienced an OFR, and 40% experienced an MFR.

For the independent variables Incub and Lic the same reasoning holds as for the dependent variables. Only 6% of the entrepreneurs that have been questioned, indicated that they have worked with an incubator/accelerator. Licenses/patens were obtained by 21 % of the startups. In Appendix E: Figure 1 an overview is exposed of the descriptive statistics for the

independent variable Stage. This figure shows that from the 1105 questioned entrepreneurs,

Descriptive Statistics N Minimum Maximum Mean Std. Deviation Variance

Dependent Variables TFR 1105 0 1 0,07 0,262 0,069 OFR 1105 0 1 0,80 0,397 0,158 MFR 1105 0 1 0,40 0,491 0,241 Independent Variables Stage 1105 1 5 2,32 1,339 1,793 Incub 1105 0 1 0,06 0,237 0,056 Lic 1105 0 1 0,21 0,407 0,166 Control Variables Gender 1105 0 1 0,61 0,489 0,239 Type 1105 0 1 0,41 0,493 0,243 Sector 1105 1 7 3,80 1,479 2,188 Country 1105 1 50 5,14 10,426 108,697

(20)

36,5% failed in the existence stage and 26% failed in the survival stage. This means that 62,5% of the entrepreneurs failed within the first two stages. The final three stages are reached by 37,4% of the questioned entrepreneurs (i.e. success stage: 17,4%, take-off stage: 9,0%, mature stage: 11,0%)3.

The descriptive statistics for two out of the four independent variables are easily explained with Table 1. The mean of the Gender variable implies that 61% of the sample are males, and 39 % are females. The business provided by the entrepreneurs was in 41% of the startups related to goods, and 59% related to services. The descriptive statistics of the Sector are exposed in Appendix E: Table A. The major group of the entrepreneurs were active in the

Hospitality and Leisure (33,0%), and in the Industry, Energy & Trade sector (29,3%). The

smallest group was active in the Healthcare sector with only 0,9%. At last, the control variable Country, consisting of 50 different countries (Table 1), is summarized in Appendix E: Table B. Notable is that 74,8% of the questioned entrepreneurs was based in Mexico, followed by Colombia, Argentina and Germany with respectively 3,5%, 3,3% and 3,2%.

3.4.2 Correlation matrix

In this paragraph the correlation matrix is shown in Table 2 (next page). Correlations give an indication about the relationship between different variables. Important, correlation does not say anything about the causal effects between variables (Newbold et al., 2012). According to Brooks (2008) correlations are not harmful for the regressions as long as they are within the boundaries of -70% and +70%. According to Table 2, it appears that the highest correlation variables in the regression models are Gender and Type. They have an estimated correlation of – 24,8 %, which is in the range determined by Brooks (2008). So, in line with the allowed level of correlations, the regressions are not harmed by any over-correlating variables.

(21)

3.4.3 Multicollinearity

By testing the data on multicollinearity, the models are tested to determine whether two or more variables are highly correlating with another variable. Highly correlating variables negatively influence the probability to estimate correct regression coefficients (Newbold, 2012). To test the multicollinearity, I have used the VIF-method (Variance Inflation Factors) as described by Field (2013). This theory indicates that variables are not influencing the regression coefficients, if the scores are within the boundaries of the VIF-scores: 10. The scores are automatically generated by SPSS and shown in Table 3. It appears that there are no independent variables that break the VIF-margin 10. Therefore, I conclude that multicollinearity is not influencing the regression coefficients in the logistic regression models.

Correlation Matrix

TFR OFR MFR Stage Incub Lic Gender Type Sector Country TFR 1,000 OFR -0,173* 1,000 MFR -0,029 -,443* 1,000 Stage 0,055 -0,033 0,031 1,000 Incub 0,118* 0,009 -0,021 -0,022 1,000 Lic 0,058 -,081* -0,016 0,071* 0,020 1,000 Gender 0,044 0,027 -0,067* 0,002 0,031 0,069* 1,000 Type -0,083* 0,012 0,022 0,003 -0,048 -0,085* -0,248* 1,000 Sector 0,016 -0,060* 0,047 0,010 -0,027 -0,002 -0,012 0,030 1,000 Country 0,025 0,030 -0,057 -0,127* 0,106* 0,016 0,100* -0,138* -0,010 1,000 Table 2: Correlation Matrix *Correlation is significant: p<0.05:

Multicollinearity table Independent Variables VIF

Stage 1,011 Incub 1,009 Lic 1,014 Gender 1,074 Type 1,078 Sector 1,002 Country 1,024 Table 3: Multicollinearity

(22)

Chapter 4: Results

This section contains the results of the regression models as constructed in the previous chapters. Table 4 summarizes the important results of the full regression models that are exposed in Appendix F. In Table 5 the hypotheses are summarized in combination with the outcomes of the regressions.

4.1 Significance test logistic regressions

The first important aspect of the regression is the Omnibus Test of Model Coefficients. This test incorporates the Chi-square test to investigate the likelihood ratio of the regression and to determine the significance. Subsequently, the most important outcome of these models are the p-values that determine whether the model can be used to draw significant conclusions or not. As shown in Table 4, it turns out that the p-values for the regression models are 0,131 for the

OFR-model, 0,149 for the OFR-model, and 0,057 for the MFR-model. So, with a significance

level of p < 0,05, the conclusion is that all regression models are insignificant. Consequently, there have not been found hard evidence that leads to significant correlations between the dependent and independent variables. However, this result is not devastating the entire research. It is still possible to interpret the results and find indications for possible

correlations, under the condition of insignificancy. The interpretation of the results will be done in the next session.

Dependent variable

TFR

OFR

MFR

Omnibus Test Chi^2 sign. (p) Chi^2 sign. (p) Chi^2 sign. (p) 74,612 0,131 73,580 0,149 80,529 0,057

Parameters Exp(B) sign. (p) Exp(B) sign. (p) Exp(B) sign. (p)

IV

Stage (1) 0,290 0,905 0,465 Stage (2) 0,959 0,904 0,892 0,577 1,190 0,294 Stage (3) 1,747 0,104 0,867 0,534 1,328 0,130 Stage (4) 1,292 0,579 1,119 0,710 0,904 0,676 Stage (5) 1,798 0,131 0,878 0,627 1,077 0,738 Stage AVG 1,449 0,402 0,939 0,671 1,125 0,461 Incub 3,760 0,000 *** 1,061 0,868 0,838 0,525 Lic 1,519 0,128 0,654 0,023 * 0,858 0,341 * p<0,05, ** p<0,01, *** p<0,001

(23)

4.2 Hypotheses results

Table 5 summarizes the hypotheses obtained from the literature review in combination with the corresponding regression results. In the last column the correlation is interpreted between the corresponding dependent variable and the independent variable from the regression result. This overview will be discussed in detail in the following paragraphs.

4.2.1 The stage of the startup and the relationship with failure reasons

Hypothesis 1: In the early stages of startup development entrepreneurs are more likely to experience technology related failure reasons, than market failure

reasons, while in the later stages the experience is the opposite

In Table 5 an overview of the first hypothesis is exposed in combination with the regression result. From the literature I hypothesized that Stage is negatively correlated with TFR. In other words, an increase in the stage of the startup would decrease the probability of experiencing a TFR. MFR is assumed to correlate positively with the stage of the startup since the likelihood to experience an MFR increases in the later stages.

From the results of Table 4 that are also incorporate in Table 5 and therefore it appears that the results of the single variables are not significant as well as the full regression model.

Literature review

Regression results

Interpretation

Hypo. DV IV Relation Chi-sq. p (model) Exp (B) p Relation direction 1. TFR Stage (avg) Negative (-) 74,612 0,131 1,449 0,402 + MFR Stage (avg) Positive (+) 80,529 0,057 1,125 0,461 + 2. TFR Incub Negative (-) 74,612 0,131 3,760 0,000 *** + MFR Incub Positive (+) 80,529 0,057 0,838 0,525 - OFR Incub Negative (-) 73,580 0,149 1,061 0,868 + 3. TFR Lic Negative (-) 74,612 0,131 1,519 0,128 + OFR Lic Negative (-) 73,580 0,149 0,654 0,023 ** - * p<0,05, ** p<0,01, *** p<0,001

(24)

The average significance for the Stage variables are 0,402 (TFR), 0,671 (OFR), and 0,461 (MFR).

Subsequent, I am not able to draw any significant conclusion for the correlation between the independent variable Stage and the dependent variable TFR, or MFR. Due to the insignificance it is only possible to cautious interpret the direction of the correlation within these regression models. That is why, I observe a positive relation between Stage and TFR (Exp(B): 1,449, p=0,402), and a positive relation between Stage and MFR (Exp(B): 1,125, p=0,461). These Exp(B)’s indicates that if the start-up grows and reaches the next stage (Stage: +1), the average probability of experiencing a TFR is increased with factor 1.449 (44,9 %) and the average probability of experiencing an MFR is increased with factor 1.125 (12,5 %). However again, due insignificance this indication is based on weak evidence.

Besides Hypothesis 1, the results in Table 4 show evidence for the insignificant relation between operational failure reasons and stages as proposed in the literature review. Thereby, must be mentioned that the full regression model is insignificant (significance level of p<0,05 (p=0,149)). So, despite the insignificance of the total model, the single variable

Stage shows indications to be highly insignificant for all stages, which confirm the literature

review that there is no relation between the stage of the start-up and an OFR.

4.2.2 Presence of an incubator/accelerator and the relationship with failure reasons Hypothesis 2. Startups working with the presence of an incubator/accelerator, are

more likely to experience market related failure reasons than technology/operational related failure reasons

Hypothesis 2. as constructed in the Literature review leads to a positive relation between Incub and MFR since entrepreneurs working with an incubator/accelerator are tend to have

less TFR and OFR due to the environment that is shaped for them, and therefore are more likely to have experienced an MFR.

According to the regressions, I observe that the relations show indications that they correlate in the opposite way as expect from the literature. Incub is positively related to TFR (Exp(B): 3,760), and OFR (Exp(B): 1,061), and negatively related to MFR (Exp(B): 0,838). Although, besides the insignificance of the model, the single variable Incub is also extremely insignificant in the OFR-regression (p=0,868) and the MFR-regression (p=0,525). Therefore, it is not possible to draw any conclusions regarding the correlation between Incub and OFR,

(25)

Contrary to these insignificancies within the OFR- and MFR- regression models, it appears to be different in the TFR-regression model. The Incub variable seems to be significant (p=0,000) in this regression model. Therefore, I examined if the removal of the control variable Country would lead to a significant regression model. The control variable

Country was included to control for cross country differences however, it turned out that this

removal led to the significant regression model as shown in Table 6 (next page) (significance level p<0,01) (Mueller and Thomas, 2000). The effect the removal of Country on all the variables is shown in Appendix G. According to these results I conclude, that the presence of an incubator/accelerator increases the likelihood to experience a TFR with 351,9%

(Exp(B)=3,519; p=0,000).

Dependent variable

TFR

Excluded

variable None Country

Omnibus Test Chi^2 sign. Chi^2 sign. 74,612 0,131 31,638 0,005 **

Parameters Exp(B) sign. (p) Exp(B) sign. (p)

IV

Stage (1) 0,290 0,178 Stage (2) 0,959 0,904 0,978 0,945 Stage (3) 1,747 0,104 1,892 0,048 * Stage (4) 1,292 0,579 1,215 0,667 Stage (5) 1,798 0,131 1,786 0,113 Stage AVG 1,449 0,402 1,467 0,390 Incub 3,760 0,000 *** 3,519 0,000 *** Lic 1,519 0,128 1,487 0,131 * p<0,05, ** p<0,01, *** p<0,001

Table 6: Regression model TFR vs. TFR (No Country)

4.2.3 Presence of licenses and patents and the relationship with failure reasons Hypothesis 3: Startups with licenses and, or patents, are less likely to experience a

technological or operational related failure reason than those without one

From the literature the hypothesis is constructed that the presence of licenses/patents (Lic) are negatively correlated with the likelihood to experience both a technological as an operational

(26)

failure reason. Unfortunately, the regression models are not showing significant evidence to confirm this hypothesis. However, just like the previous independent variables, there are indications that can be taken from the regression models.

In the regression with the TFR as dependent variable, the presence of licenses/patents (Lic) is positively related (Exp(B)= 1,519). This means that Lic increases the probability of experiencing a TFR. Nonetheless, the p-value of this single independent value is furthermore insignificant (p=0,128). For that reason, it is not possible to attach any value to this indication.

In the OFR-model the single variable Lic (p=0,023) is significant with a significance level of p<0,05. The indication that might be derived from this result is that Lic is negatively related to OFR, with an Exp(B) of 0,654. This means that an increase in Lic, increases the probability of experiencing an OFR with 65,4%. And thereby, the likelihood to experience an

OFR with Lic reduced with 34,6 % (100%-65,4. So, the insignificant model (p=.,149), shows

indications that the presence of licenses/patents might decrease the likelihood to experience an operational failure reason.

(27)

Chapter 5: Conclusion, Implications and Contribution

This section will elaborate on the thesis’ results by arguing these results in light of prior literature. Next to that this section contains the contribution, the recommendations for further research and the limitations.

5.1 Conclusion

The main subject in this thesis is to determine whether three different independent variables correlate with the likelihood of experiencing three dependent variables. To determine these correlations, the following main research question has been constructed:

To what extent do different stages of startup development, the presence of incubator/accelerator, and the presence of licenses/patents, correlate with the likelihood of experience technology, operational and, or market failure reasons?

The results of the regression models have shown that there are no significant correlations observable within the logistic regression models as constructed from the conceptual model. Despite this conclusion, it is possible to extract indications about the correlations between the likelihood to experience different failure reasons and the concerning independent variables.

5.1.1 The stage of startup development and the relationship with failure reasons

The regressions show that the single independent variables concerning the stage of the startup are insignificant as well. This might implicate that there is no correlation between the stage of the startup and the likelihood to experience an operational failure reason, which is in line with the literature. This might also confirm the findings of Hill et al. (2002), that an important aspect within startups is the personality of an entrepreneur and his capabilities to lead the operational tasks, regardless of the stage where the startup is situated.

Due to the high insignificance of the Stage variable in the regression models

concerning technological and market failure reasons, it is difficult to find any indications for possible correlation between the variables. For the market failure reason regression this might be explained by the origin of this failure reason. Since market failure reasons are caused by factors that are not in control of the entrepreneur, and thereby occur independently of the stage of the startup, the randomness of this factor might explain the overall non-correlation between market failure reasons and the startup stage.

(28)

5.1.2 Presence of incubator/accelerator and the relationship with failure reasons

The presence of an incubator/accelerator is expected to be negatively correlated with technological (TFR) and operational failure reasons (OFR). Since incubators/accelerators create the perfect environment to develop products/services and to develop the operational aspects of the startups, these problems could be prevented (Lumpkin, 1988). The constructed hypothesis can therefore not be confirmed by the results. In the OFR-model the Incub variable appears to be insignificant. This might be the case because the operational performance of the firm is mainly depending on the entrepreneurs’ personality. Those personality characteristics are not changed by the presence of an incubator/accelerator. So, therefore there might be no correlation between the likelihood to experience an operational failure reason and the presence of an incubator/accelerator.

According to the results it turns out that variable Incub is positively related to

technological failure reasons. Even in the significant model, without control variable Country, there is evidence for this correlation. In contrast to the literature, the likelihood of

experiencing a TFR increases with the presence of an incubator/accelerator. This observation might be explained by the fact that incubatorsand accelerators are not timeless supporting programs. Incubators/accelerators shape the perfect environment for startups to increase the probability of success. However, the growth that might occur in a successful startup, could lead to the situation that the startup has to continue independently from the

incubator/accelerator. At that moment the startup has to control the growth by itself, thereby it also has to control for the quality of the product/service. The (unexpected) growth, might cause problems in maintaining the original product/service quality on a larger scale. Secondly, startups that grow, force entrepreneurs to take a more strategic leadership role than a

functional leadership role. This might force him to leave the real workplace where products and services are made. His/her experience and adequate quality control that than will be missed at the workplace might decrease the quality and thereby technological failure reasons might occur.

For the correlation between market failure reasons (MFR) and the presence of an incubator/accelerator I have not found evidence in the empirical results. In line with the correlation between MFR and Stage, this insignificance might be caused by the randomly, and unexpected occurrence of market coincidences. The presence of an incubator/accelerator does not influence the control that an entrepreneur has on market factors. Therefore, is the

(29)

likelihood to experience a market failure reasons not correlated with the presence of an incubator/accelerator.

5.1.3 Presence of licenses/patents and the relationship with failure reasons

The presence of licenses and patents are expected to increase the overall quality of the startup. According to this assumption there should be a negative correlation between the likelihood to experience a technological, or operational failure reasons and the presence of a license/patent. Nonetheless, the results have not shown any evidence that supports this correlation.

According to technological failure reasons, there are indications that the likelihood to experience a TFR is increased with the presence of licenses/patents (Lic). However, this relation is very doubtful due to the insignificance of the Lic variable in this regression model. In contrast to the Lic-variable in the TFR-model, the Lic-variable is significant in the OFR-model. The Lic-variable coefficient indicates that the presence of licenses/patents decrease the likelihood of experiencing an operational failure reason with 36,5%. Therefore, the

assumption that licenses/patents are positive indications of the overall quality of the operational capabilities of the startup, might be confirmed.

5.2 Implications

As argued by Everett and Watson (1998) entrepreneurial failure is mainly caused by endogenous factors. The endogenous failure reasons mainly include the operational and financial problems of startups and/or problems regarding the quality of the product/service. According to Thornhill and Amit (2003), these operational failures are explained by the lack of adequate skills of starting entrepreneurs to make the best operational decisions for their startup. For the purpose of this paper, these endogenous factors are described as technological and operational failure reasons.

Subsequently, the theory of Lumpkin (1988) suggests that the incubators/accelerators create controlled environments for startups to acquire the resources to overcome these operational and technological failure reasons. They do this by providing equipment in the controlled environment which is unaffordable for starting entrepreneurs, offering training programs to learn entrepreneurs the adequate operational skills, and by reducing the anxiety of starting new businesses. However, the theory is unclear about whether incubators/accelerators might also increase entrepreneurial failure. My study suggests that the likelihood of experiencing a technological failure reason is increased by the presence of an incubator/accelerator. An

(30)

explanation for this increased probability might be that the product and/or service

development is accelerated by the presence of incubators/accelerators, which may lead to the early discovery of fatal flaws in products/services and/or the unworkability of the product service.

The relation between the experience of technological and operational failure reasons and the presence of licenses/patents is also described in more recent literature. According to Haeussler et. al (2009), licenses and patents are not only intellectual property protections, but also

quality signals for investors. For those reasons, they suggest that licenses/patents could help entrepreneurs to acquire the needed financial resources, to facilitate the commercialization of the technology and to obtain access to multiple information resources. These aspects should reduce the likelihood to experience technological and operational failure reasons. Due to the result of my research that the presence of licenses/patents might reduce the likelihood to experience an operational failure reason, I suggest that the relation between operational failure reasons and the presence of licenses/patents can be confirmed.

5.3 Limitations

To understand the contribution of this thesis, the limitations of the research project must be understood. These limitations will be discussed in the coming paragraphs.

The first limitation of this thesis is the distribution of the 11 failure reasons retrieved from the questionnaire into the three dependent variables as constructed from the literature. Several failure reasons may leave room for discussion whether they are categorized in one category or the other. The reasons are beliefs why entrepreneurs fail, and not per se causes. These reasons and causes might overlap but might also be different. This is a subjective interpretation of the failure reason and the possible causes is a measurement issue.

This interpretation of failure reason is also involved in the questionnaire filled in by the failed entrepreneurs. They might have interpreted some concepts in different ways. This bias might have influenced not only the failure reason choice, but as well for other variables. For example, it might be that entrepreneurs categorized themselves in stages since they believed they reached a certain stage. However, due to all sorts of reasons this choice might leave room for discussion. So, the overall interpretation bias for entrepreneurs within the questionnaire might be seen as the second limitation.

(31)

The third limitation, is the fact that most startups were based in Mexico (74,8%). Since there are cross-country differences, this factor might cause that the results of this thesis are not suitable for other countries (Mueller and Thomas, 2000). In addition, the interpretation of the variable regarding the presence of an incubator/accelerator might also be influenced by these cross-country differences because the environment created by incubators/accelerators might differ between countries.

The fourth limitation is that this research not based on a time series connected database. Therefore, it is not possible to test the relation over time. These relations might be changed from the past and might change in the future which could influence the view on the relations. Namely, it might for example be that the relation between incubators/accelerators and certain failure reasons is improving since the environment created by the

incubators/accelerators is developing (Vanderstraeten and Matthyssens, 2012). This trend is not noticeable in this data. Therefore, it is important to see the results within the current environment.

5.4 Contribution and avenues for future research

Although this thesis has not found evidence that support the hypotheses, it contributes to the body of entrepreneurial literature by indicating some expected correlations and suggest some new, unexpected correlations. The goal of this paper is in line with Lussier (1996) not to determine a certain set of reasons why business fail. The value of this paper lies within the understanding of entrepreneurial failure. A better understanding of why business fail and understanding the learning benefits, are the foundation for startups to be successful because it improves the entrepreneurs’ ability to deal with potential failure reasons. Thereby, I

contribute to the learning process for individual entrepreneurs and even to the economic growth as proposed by Olain and Sorensen (2014) (Baker et al., 1997; Kets de Vries, 1985; McGrath, 1999; Reynolds, 1987; Romanelli, 1989). Especially the result that

incubator/accelerators might increase the likelihood of experiencing a technological failure reason, might lead to discussion and this subject definitely needs further research. The intensified discourse about the role of this, and other, factors will increase the knowledge in the field of entrepreneurship.

That is why I propose, in line with McGrath (1999), Shepherd (2003) and Zacharakis et al. (1999), that more research must be done to the role of entrepreneurship and the learning

(32)

benefits of entrepreneurial failure. Especially I suggest to intensify the research in factors that are in control of the entrepreneur, the technological and operational factors, instead of the market failure reasons. Finally, I recommend to intensify the research about the personality characteristics of entrepreneurs. This research should be done to confirm the importance of personality characteristics within startups, and the way how different personalities are used to increase the probability of success.

(33)

References

Allen, D.N. (1985) An entrepreneurial marriage: Business incubators and start ups, Paper presented at the Babson Entrepreneurial Research Conference, Philadelphia, PA. Baker, T., Aldrich, H., Langton, A., and Cliff, J.E. (1997), Tempered by the flame? What do

entrepreneurs learn from failure?, Paper presented at the annual meeting of the

Academy of Management, Vancouver, British Colombia.

Berryman, J. (1983), Small Business Failure and Bankruptcy: A Survey of the Literature, European Small Business Journal, 1(4), 47-59.

Brooks, C. (2008), RATS Handbook to accompany introductory econometrics for finance, Cambridge Books.

Bruno, A., Leidecker, J., and Harder, J. (1987), Why firms fail, Business Horizons, 30: 50-58. Bryan, E.L. (1984), Why small business fail: Testimony proposes, Management Accounting,

Vol. 16, 73.

Churchill, N.C. and Lewis, V.L. (1983), The five stages of small business growth, Harvard Business Review, May-June, pp. 30-49.

Coad, A.l. (2013), Death is not a success: reflections on business exit, International Small Business Journal, Vol. 0 No. 0, pp. 1-12.

Cockburn, I. and MacGarvie, M. (2007), Patents, Thickets and the Financing of Early-Stage

Firms: Evidence from the Software Industry, NBER Discussion Paper, No. 13644.

Cope, J. (2011), Entrepreneurial learning from failure: an interpretative phenomenological

analysis, Journal of Business Venturing, Vol. 26. No. 6, pp. 604-623.

Croson, R., and Greezny, U. (2009) Gender differences in preferences, Journal of Economic literature, 47(2), 448-474.

De Noble, A. and Moliver, D.M. (1983), The small business dilemma: Can cooperation

help?, American Journal of Small Business, 49.

Dollinger, M.J. (1985), Environmental contacts and financial performance of the small firm. Journal of Small Business Management, pp. 24-30.

(34)

Edmondson, A.C. and McManus, S.E. (2007), Methodological fit in management field

research, Academy of Management Review, 32(4), pp. 1246-1264.

Eisenhardt, K.M. (1989), Building theories from case study research, Academy of Management Review, 14: 532-550.

Eisenhardt, K.M. and Graebner, M.E. (2007), Theory building from cases: opportunities and

challenges, Academy of Management Journal, 50: 25-32.

Everett, J. and Watson J. (1998), Small Business Failure and External Risk Factors, Small Business Economics, 11: 371-390.

Failure Institute (2017), retrieved at 19-01-2018 from https://thefailureinstitute.com/. FEE (Fédération des Experts Comptables Européens) (2004), Avoiding business failure: A

guide for SMEs, A paper issued by the European Federation of Accountants.

Field, A. (2013), Discovering statistics using IBM SPSS statistics, sage.

Gaskill, L.R. and Van Auken, H.E. (1993), A Factor Analytic Study of the Perceived Causes

of Small Business Failure, Journal of Small Business Management, 31(4): 18-31.

Gimeno, J., Folta, T., Cooper, A., and Woo, C. (1997), Survival of the fittest?

Owner-managerial human capital and the persistence of underperforming firms,

Administrative Science Quarterly, 750-783.

Haeussler, C., Dietmar, H., and Mueller, E. (2009), To be Financed or Not: The Role of

Patents for Venture Capital Financing, ZEW Discussion Papers, No. 09-003.

Harrington, M. (2010), Are you a Pirate, Techcrunch.

Hill, J., Nancarrow, C., and Wright, L.T. (2002), Lifecycles and crisis points in SMEs: a case

approach, Marketing Intelligence & Planning, Vol. 20 Iss 6 pp. 361 – 369.

Juan Carlos Domezain (2018), Causes of Failure in Mexican Social Enterprises, The Failure Institute.

Kenny, K. and Scrive S. (2012), Dangerously empty? Hegemony and the construction of the

Irish entrepreneur, Organization, Vol. 19, No. 5, pp. 615-633.

Kets de Vries, M.F.R. (1985), The dark side of entrepreneurship. Harvard Business Review, pp. 160-168.

(35)

Lammers, J., Pelzer, B., Hendrickx, J. and Eisinga, R. (2007), Categorische Data Analyse met

SPSS, Inleiding in loglineaire analysetechnieken, Assen, Van Gorcum.

Long, C. (2002), Patent signals, University of Chicago Law Review, Vol. 69, pp. 625-679. Lumpkin, J.R. (1988), Screening Practices of New Business Incubators: The Evaluation of

Critical Success Factors, American Journal of Small Business, pp. 59 – 81.

Lussier, R.N. (1996), Reason why small Businesses fail: and how to avoid failure, The Entrepreneurial Executive, Vol. 1, Numb. 2, pp. 10-17.

Mann, R.J. (2005), Do patents facilitate financing in the software industry?, Texas Law Review, 83, pp. 961-1030.

McGrath, R.G. (1999), Falling Forward: Real Options Reasoning and Entrepreneurial

Failure, Academy of Management Review, Vol 24, No. 1, 13-30.

McMahon, R.G.P., Holmes, S., Hutchinson, P.J. and Forsaith, D.M. (1993), Small Enterprise

Financial Management: Theory and Practice, Sydney, Australia: Harcourt Brace.

Miller, P. and Bound, K. (2011), The Startup Factories: The rise of accelerator programmes

to support new technology ventures, NESTA.

Money (1984), Bringing up businesses, 14.

Mueller, S.L. and Thomas, A.S. (2000), Culture and Entrepreneurial Potential: A Nine

Country Study of Locus of Control and Innovativeness, Journal of Business

Venturing, 16: 51-75.

Newbold, P., Carlson, W. and Thorne, B. (2012), Statistics for Business and Economics, Pearson, 8th edition.

Olainson and Sorensen (2014), The abject of entrepreneurship: failure, fiasco, fraud.,

International Journal of Entrepreneurial Behavior & Research. 20(2). pp. 193 - 211

Perren, L. and Jennings, P.L. (2005), Government discourses on entrepreneurship: issues of

legitimization, subjugation, and power, Entrepreneurship, Theory and Practice, Vol.

29 No. 2, pp. 605-635.

Reynolds, P.D. (1987), Firms, Societal Contribution vs. potential, Journal of Business Venturing, 2: pp. 216-231.

Failure, the road to success : the relation between the stage of a startup, the presence of incubators/accelerators and the presence of licenses/patents and technological, operational and market failure reasons