Entrepreneurial Motivations and the 2008 Crisis: A Closer Look at Necessity- and Opportunity-Driven Entrepreneurs in Europe and the United States

(1)

1 Entrepreneurial Motivations and the 2008 Crisis: A Closer Look at Necessity- and

Opportunity-Driven Entrepreneurs in Europe and the United States

Master Thesis

Name João Ricardo Salgado Silva

Student Number s3001121

Educational Program Msc. International Business & Management

First Supervisor Prof. dr. D.H.S. Akkermans

Address Rua Dr. Afonso Cordeiro 899, 7D, 4450-007 Matosinhos, Portugal

Phone Number +351 937 643 969

(2)

2

ABSTRACT

Purpose: This research seeks to explore the effects of the 2008 crisis on the composition of motivations of entrepreneurs in the United States and Europe. Through the use of the necessity-driven and opportunity-necessity-driven entrepreneurship framework, conclusions are drawn about the representation of such entrepreneurs before and after the crisis.

Design/Methodology/Approach: The research includes 2 types of analysis. First, we made a general analysis of the graphs were we focus individual countries and the PIIGS region. In the graphical analysis we focus in trying to understand if there was observable response by the entrepreneurs to the crisis of 2008. The second part of the analysis is of quantitative nature. We perform panel data analysis to try to understand if some of the observable conclusion were statistically confirmed. The panel data focus mainly on the effect of the crisis on the US and PIIGS.

Findings: The graphical analysis lead us believe that countries where the crisis had more severe social and economic consequences showed a strong increase in the necessity-driven entrepreneurship. There are common patterns when it comes to the activities of necessity-driven entrepreneurs during and after the crisis. After the initial dip of the crisis, there is a substantial increase in the rate of necessity-related motivations. This suggests a lag between the effects of the crisis and the due response of necessity-driven entrepreneurs.

The quantitative part of the research does not corroborate the observation of the graphs. The results show that US necessity entrepreneurs respond to the crisis as expected but this does not happen with PIIGS necessity entrepreneurs. When it comes to the opportunity entrepreneurs, the US entrepreneurs do not show any response to the crisis as expected. However, the PIIGS entrepreneurs show that the crisis affected them negatively, which goes against what was hypothesized.

(3)

3 necessity and opportunity framework creates an mirror effect in the results, e.g. negative results for opportunity is going to a positive effect for necessity. Third, the dummy variable does not express the impact of a crisis with a continuous effect but through a binary effect. This is a problem because it tends to make results more homogeneous than it would be expected and thus does not represent the effects of the 2008 in the US and PIIGS with accuracy.

Originality/Value: The research is original because it is the first (to the authors knowledge) to relate and test the concepts of the necessity and opportunity with the 2008 crisis.

Keywords: Necessity; Opportunity; Entrepreneurship; Crisis; Recession

(4)

4

7.1 LIMITATIONS ...56 7.2 FUTURE RESEARCH ...57 8. FINAL CONCLUSION ... 59 8. BIBLIOGRAPHY ... 60 A.1 BELGIUM ...65 A.2 CROATIA ...65 A.3 DENMARK ...66 A.4 FINLAND ...67 A.5 FRANCE ...67 A.6 GERMANY ...68 A.7 GREECE ...68 A.8 HUNGARY ...69 A.9 IRELAND ...70 A.10 ITALY ...70 A.11 LATVIA...71 A.12 NETHERLANDS ...71 A.13 NORWAY ...72 A.14 SPAIN ...73

A.15 UNITED KINGDOM ...73

A.16 NORMALITY TEST - CHARTS ...74

A.17 LINEARITY TEST - CHARTS ...74

A.18 HAUSMAN TEST - CHARTS ...76

A.19 LOG MODELS TEST - CHART ...78

A.20 BEURSCH-PAGAN TEST – CHARTS ...79

A.21 CROSS SECTIONAL DEPENDENCE TEST (PESARAN TEST) - CHARTS ...80

(6)

6

A.23 AUTOCORRELATION/SERIAL CORRECTION TEST ...82

A.24 NECESSITY LIN-LOG FIXED EFFECT MODEL WITH PIIGS DUMMY AND US DUMMY ...83

A.25. NECESSITY LOG-LOG FIXED EFFECT MODEL WITH US DUMMY AND PIIGS DUMMY ...84

A.26 OPPORTUNITY LIN-LOG FIXED EFFECT MODEL WITH US DUMMY AND PIIGS DUMMY ...85

A.27. OPPORTUNITY LOG-LOG FIXED EFFECT MODEL WITH PIIGS DUMMY AND US DUMMY ...86

A.28. NECESSITY LIN-LOG MODEL WITH PIIGS DUMMY ...87

A.29. NECESSITY LOG-LOG MODEL WITH PIIGS DUMMY ...87

A.30. OPPORTUNITY LIN-LOG MODEL WITH PIIGS DUMMY ...88

A.31. OPPORTUNITY LOG-LOG MODEL WITH PIIGS DUMMY ...89

A.32. NECESSITY LIN-LOG MODEL WITH US DUMMY ...90

A.33. NECESSITY LOG -LOG MODEL WITH US DUMMY ...90

A.34. OPPORTUNITY LIN-LOG MODEL WITH US DUMMY ...91

A.35. OPPORTUNITY LOG-LOG MODEL WITH US DUMMY...92

A.36 NORMALITY FOR NECESSITY LOG-LOG MODEL WITH US TALLY AND PIIGS TALLY ...93

A.37 LINEARITY TEST FOR NECESSITY LOG-LOG MODEL WITH US TALLY AND PIIGS TALLY ...93

A.38 HAUSMAN TEST FOR NECESSITY LOG-LOG MODEL WITH US TALLY AND PIIGS TALLY ...93

A.39 WOOLDRIDGE TEST FOR AUTOCORRELATION IN NECESSITY LOG-LOG WITH USTALLY AND PIIGSTALLY ...94

A.40. BREUSCH AND PAGAN LAGRANGIAN MULTIPLIER TEST FOR RANDOM EFFECTS ...94

A.41. NECESSITY LOG -LOG MODEL WITH TALLYUS AND TALLYPIIGS ...94

A.42. LINEARITY FOR OPPORTUNITY LOG-LIN MODEL WITH US TALLY AND PIIGS TALLY ...95

A.43 NORMALITY FOR OPPORTUNITY LOG-LIN MODEL WITH US TALLY AND PIIGS TALLY ...95

A.44. HAUSMAN TEST FOR OPPORTUNITY-LOG MODEL ...95

A.45 FIXED ENTITY AND TIME EFFECT WITH US TALLY AND PIIGS TALLY ...96

(7)

7

1. INTRODUCTION

1.1 GENERAL INTRODUCTION

It is increasingly acknowledged that entrepreneurship plays an important role in advanced economies. Historically entrepreneurship has been linked to disruptive innovation, economic growth and increases in overall productivity and also nowadays the notion of entrepreneurship is discussed within these realms. When the economic crisis of 2008 hit the Western economies, the necessity and importance of entrepreneurial activity was brought to public debate. The economic shock caused large increases in the unemployment rates and working conditions for the general population deteriorated in many Western countries. For many companies this inevitably led to reorganizations of the productive factors within the firm, resulting in more working hours and salary cuts for employees. As a result, the general quality of the working population declined and factors such as stress and other mental health concerns increased. Therefore, many people in this situation saw the added benefit of starting up their own business. In the process of attenuation that followed the economic shock in 2008, entrepreneurship thus became a fundamental cornerstone in Western societies.

This relationship between entrepreneurial activity and economic growth is not exactly new (Acs et al. 2009, 2010; Carree and Thurik 2010). Over time, the economy evolves in what are called economic cycles, which means that there are periods of economic thriving, periods of recession and depression (downturn) and the time in between such periods (Llopies et al. 2015). However, there is still little information on what types of entrepreneurship may (or may not) arise at times of economic downturn. More specifically, it is unclear if the decision to become an entrepreneur after a crisis is guided by a distinct motive as a result of that crisis.

(8)

8 dissatisfaction towards their current position and that “pull entrepreneurs” pursue an entrepreneurial career because they see opportunities to do so, even though they are satisfied with their current position. A different, but similar approach is to say that “push” refers to “necessity entrepreneurs”, and “pull” refers to “opportunity entrepreneurs” (Giacomin et al. 2011). This distinction will be further elaborated in the theoretical framework. A clear distinction between the concepts is critical as the entrepreneur´s intrinsic motivation may affect the way he or she runs their business and thus the performance of the company (Hessels, Van Gelderen and Thurik 2008). In order to comprehend the actions of push and pull entrepreneurs after an economic shock, data from the Global Entrepreneurship Monitor (GEM) is used. The sample is composed by several countries in Europe and the United States in the period of 2001 to 2015. The 2008 economic crisis offers a unique opportunity to put the push-pull theory to the test in a scenario of economy shock. Instead of the traditional push and pull concepts, GEM uses the concepts of necessity (push) and opportunity (pull) entrepreneurs to name their indicators. These indicators are also available before and after the crisis of 2008, which will help to draw conclusions about the behaviour of entrepreneurs around an economic depression.

1.2 RESEARCH QUESTION

This subparagraph will introduce the research question that guides the research. The general introduction already indicated that the push-pull theory and the crisis of 2008 could be linked together to uncover whether times of economic downturn affect the motivation towards entrepreneurship. Intuitively, it makes sense that the crisis affects potential push-necessity entrepreneurs, as unemployment usually rises. Simultaneously, potential pull/opportunity entrepreneurs might see more but also less opportunities to engage in entrepreneurial activities. Therefore, the research question to this study should capture the possibility that the crisis has any kind of effect on the motivation towards entrepreneurial activities. This results in the following research question:

(9)

9

1.3 RELEVANCE

According to developments in the push-pull theory, high unemployment results in self-employment (push), but simultaneously the uncertainty associated with high unself-employment might lead people to keep their saving in the bank rather than to start a venture in a depressed economy. On the other side of this theory, the perspective of “pull” seems a bit more complex. Most of the research does not seem to be focused around the effects of a crisis on the pull entrepreneur. But is this then relevant?

Understanding the dynamics of the entrepreneur's motives after an economic shock can help understand constraints and incentives that arise during this period. This is useful because it can help policy makers to better design policy destine to restart the economy in a scenario of uncertainty. Having a supportive policy toward entrepreneurs after an economic shock can deeply increase the potential for a faster recovery. The promotion of entrepreneurial practices can help to minimize the effects of unemployment, to control welfare expenditures related with the higher demand of social support due to unemployment and to potentially change entrepreneurial attitude of present and future generations. Entrepreneurial mindfulness can improve entrepreneurial market analysis (Gordon and Schaller 2014). Better understanding of the dynamics that play in a crisis can make the potential entrepreneur more capable to evaluate a possible entrepreneurial opportunity.

1.4 OUTLINE

(10)

(11)

11

2. RESEARCH FRAMEWORK

This chapter elaborates upon the underlying theories to this research. First, the concept of entrepreneurship is described, to create a clear understanding of the concept throughout the research. Second, the central theory discussed, the push-pull theory, is described. Subsequently, an elaboration is made on the crisis of 2008: how did it start, evolve and where are the economies in Europe and the United States heading now? All these theories and events are taken together at the end of the chapter in order to form several hypotheses with regard to the analysis that will be performed later on.

2.1 ENTREPRENEURSHIP

In order to have a clear view on what entrepreneurship entails, this first paragraph describes shortly what the view on entrepreneurship is throughout the research. Defining this concept in a straightforward manner is not simple, as there are many available views in existing literature.

Defining who an entrepreneur is and what he or she does might seem simple, but is still one of the great debates in economic and entrepreneurial research. Fields such as economies, management studies and sociology have contributed much to the development of this concept. Oddly, the most prevalent and enduring definitions have come from economists instead of from management studies or entrepreneurs themselves. The word entrepreneur was first used by the economist Jean-Baptiste Say, who felt that there was a word needed to describe enterprising businessmen (Anthony Brewer 1992). And thus the term entrepreneur was born in economic literature. The most traditional definition of what an entrepreneur does comes from Joseph Schumpeter, who posited that “entrepreneurs revolutionize the pattern of production through inventions, application of new technical solutions or by introducing new commodities” (Schumpeter 1934, p. 132).

(12)

12 Thus, if Schumpeter sees the innovation as the main driving force for entrepreneurship, Kirzner sees opportunity perception as the main driving force in the entrepreneurial process.

Another prominent player for defining the role of the entrepreneur is Frank Knight. He can be considered an expert in the realms of risk and uncertainty. Even though these terms might seem equal, there is a clear difference between then. Risk presupposes that enough information is available to allow the decision maker to know the odds of his decision, whereas uncertainty presupposes that there is no real knowledge about the odds of the decision. Relating this to entrepreneurship, Knight defines the entrepreneur as the one that can bear the uncertainties of the enterprise, which are related with profitability. According to his notions, the entrepreneur is motivated by the idea of “being his own boss”.

These three views together form a quite well rounded basis for the push-pull theory that is described in the next paragraph. Schumpeter and Kirzner´s views can be easily relation with the pull motivation, and Knight´s view can be related with both push and pull motives.

A final definition that needs to be added to these is the one from GEM, which is the provider of the data that is used in the research. Their definition is crucial, because their interpretation of entrepreneurship underlies the data. The GEM definition of entrepreneurship is aligned with the most key entrepreneurship scholars (see e.g. Shane and Venkatraman 2000, Hisrich et al. 2005). According to GEM, entrepreneurship is defined as “any attempt at new business of new venture creation, such as self-employment, a new business organization, or the expansion of an existing business, by an individual a team of individuals, or an established business” (GEM, 2012). Even though this definition may seem to define entrepreneurship as just “new business activity”, it also takes a more expansive view by recognizing what this new business activity could be.

All these definitions taken together thus form the basis for the interpretation of entrepreneurship throughout this research.

2.2 PUSH-PULL THEORY

(13)

13 decision to become self-employed. They saw that when the benefits (either monetary or not) resulting from self-employment surpass the salary they receive from an employer or the unemployment status, the decision to turn to entrepreneurship is made (Johnson and Darnell 1976). As a result of this study, the authors conclude that the decisive element in the decision to become an entrepreneur can be interpreted as the motivation along two forces: push and pull.

The studies by Johnson and Darnell (1945) formed the landmark for a new way of looking at entrepreneurship and many researches continued their work. Years later, Amit and Muller (1995) formalized the push-pull framework that was based on the work of Shapero and Sokol (1982) and Feeser and Dugan (1989) and applied it to entrepreneurship. They define the two concepts, push and pull, in a distinctive way. The “push entrepreneur” is seen as the one that moves to self-employment as a result of dissatisfaction with the current position. The “pull entrepreneur” is the one that is drawn towards entrepreneurial activity by the appeal of a business idea and the personal implication associated with the development of entrepreneurial activity. The push-pull model is frequently used in several studies, which has led to reinterpretations of the push and pull factors, as these two cannot always be seen as so distinctive. Oftentimes, the model is simplified in a way that puts unemployment, fear of job loss or instability of the current job as the main or only sources of motivation in the model. However, this does not always have to be the case.

2.2.1 ADAPTING THE MODEL

(14)

14 In general, push entrepreneurship is associated with more negative motives from the potential entrepreneur to start a venture, whereas pull entrepreneurship is associated with positive motives (Shapero and Sokol 1982, Dawson and Henley 2012). Therefore, pull entrepreneurs are expected to work with a more proactive dynamic that derives from high professional aspirations, while the push dynamic results from a correcting perspective from the part of the entrepreneur (Giacomin, Guyot, Janssen and Lohest 2007).

In an attempt to more correctly conceptualize the two factors, Reynolds et al. (2002) introduced new interpretations of the two factors: necessity (push) and opportunity (pull) entrepreneurship. These terms gained widespread attention and positive critique in modern research. The disruption that leads to the change from a non-entrepreneurial situation to entrepreneurship by push and pull entrepreneurs is not always clear in the literature and the inclusion of the terms necessity and opportunity entrepreneur are important in this situation.

The adaption of the push/pull theory to the business cycle theory led to the terms recession-push and prosperity-pull. In their article, Benedict and Hakobyan (2008) report the inverse relation between self-employment and business cycle in order to argue for the recession-push hypothesis. On the other hand, the prosperity-pull hypothesis argues that the self-employment incomes decline with the decrease of aggregate demand.

In other studies, connections were drawn between push and structure, and pull and culture, in the sense that the ethnic entrepreneur can be either push (structural) or pull (cultural) (Bun and Hui 1995, Dana and Morris 2007, Min and Bozorgmehr 2000). These more culture related terms are beyond the scope of this research and will therefore not be elaborated upon. The terms used by Reynolds et al. (2002), are similar to the terms used by GEM, which is the provider of the data. They define both as “opportunity-based” (pull) and “necessity-based” (push) entrepreneurship. These are thus the definitions of push and pull that will be referred to throughout this research.

(15)

15 2.2.2 A DISTINCTION OF MOTIVES

To get a clearer view on the exact motivation for starting up a new venture, it is important to distinguish between the different types of motives that underlie the push and pull motivations. Push motivations underlies necessity-based entrepreneurship. Such motivations thus cause people to move away from the current situation with the hope to find something better within entrepreneurship. Examples of such motives are dissatisfaction with the current job (Schjoedt and Shaver 1986), unemployment (Lawrence and Hamilton 1997), scarcity of local jobs, pursuing better working conditions, meeting family expectations (Dawson and Henley 2012), career setbacks, difficulties in finding a job due to education, race, gender etc. (Gilad and Levine 1986) and so on.

Pull motivations underlies opportunity-based entrepreneurship and are therefore vastly different from the push motivations. Examples are increasing one's income (Mason 1989; Dawson and Henley 2012), developing new products, manufacturing products (Giacomin, Janssen, Guyot and Lohest 2011) and so on.

2.2.3 TRANSITION FROM POTENTIAL ENTREPRENEUR TO ENTREPRENEUR

Giacomin et al. (2011) point out the importance of the principle of action to explain the interaction of the potential entrepreneur, who is composed by strategical, cultural, historical elements, and the situation in which the action takes place (Amblard et al. 1996, Bernoux 1995). The authors pose that the potential entrepreneur as a strategic actor makes decisions with the reigning context in mind. The decisions they make are in accordance with their interests, perception of reality, perceived opportunity, available resources etc. Vinogradov et al. (2013) used a planned behavior theory to predict entrepreneurial intentions to start a business. This planned behavior theory says that individuals can select willfully from a number of alternatives. It poses that there are 3 main pillars that are used to determine entrepreneurial intentions:

1. Favorability (or not) for the act in question (attitude);

(16)

16 In a crisis situation, potential entrepreneurs have limited possibilities. Venture creation might be among these possibilities, which is then acted upon depending on the way they perceive these three pillars. For the current research, this is an important characteristic as this theory allows to speculate about the behavior of entrepreneurs. This will help to create hypotheses. In the next paragraph, this crisis will be further explained.

2.3 THE CRISIS

The crisis of 2008 got to be known as the biggest economic shock since the Great Depression during the 1930s (IMF 2008, UN 2011). The crisis originally started by a crisis in the subprime mortgage market, which would later be known as the subprime bubble (Reinhart and Rogoff 2008). Eventually this would develop into a banking crisis that would spread all over the US and Europe (Reinhart and Rogoff 2013). From this banking crisis it resulted in what would be called a “credit crunch” (Brunnermeier 2009). Particularly in Europe, there was a third crisis related with sovereign debt and the ability of some countries to maintain acceptable levels of debt, which required them to be assisted by external institutions (Gianviti, Von Hagen, Pisani-Ferry and Sapir 2010). This crisis would later be called the “sovereign crisis”. The total period of economic decline received the name “Great Recession”. Despite the size and scope of the 2008 crisis, there is no universally agreed narrative on the causes of the crisis and it is more likely that the crisis is the result of a series of causes than just a single cause (Helleiner 2011). Therefore, there are numerous theories to what triggered it, which makes it necessary to dedicate this paragraph to the 2008 economic crisis. This way, it becomes clear how the crisis might be a factor in changes in entrepreneurship or not. The current paragraph will be split in two subsections that subsequently describe the process that resulted in the crisis, followed by the effects of the crisis in the period where the crisis was the most severe and the years after.

2.3.1 PUMPING THE BUBBLE - 1990S-2008

(17)

17 mandated that commercial and investment bank should be separated. With the repelling of this Act, investment and commercial banks were brought together. There was quite some disagreement with this event, for instance by Joseph Stiglitz (Stiglitz 2009). He claimed that commercial banks are not supposed to be high-risk ventures, but instead should manage other people’s money very carefully. It is with this understanding that the government agrees to help should they fail. Investment banks, on the other hand, have traditionally managed rich people’s money – people who can take bigger risks in order to get bigger returns. When investment and commercial banks were brought together when the Glass-Steagall Act was repealed, a more investment-bank culture arose as well within the more commercial banks that managed people’s money. This created a demand for the types of high returns that could only be obtained through high leverage and big risk taking. (Stiglitz 2009) Investment and commercial banks used to be separated for a reason: to avoid a conflict of interest of the lending part (commercial banking) with the rating and financial-market oriented part (investment banking). A division that was no longer present without the Glass Steagall Act. This gave birth to the concept of “shadow banking”, which can be defined as “a diverse set of institutions and markets who collectively carry out traditional banking functions – but do so only loosely linked to the traditional system of regulated depository institutions (Bernanke, 2013).

(18)

18 The true recipe for disaster was created in the last few years preceding the crisis, when the idea of housing became convenient for all the parts involved. In the political sphere, the idea of homeownership became convenient as a demagogue move to win popularity and keep PAC’s (Political action committees) happy (Sowell 2011) In the private sphere, both investment and commercial banks got mesmerized by the high profits (Zuckerman 2008). The homeowner had cheap access to habitation (even if in normal circumstances they would not meet the criteria to the lending requirements). The real estate construction sector was healthier than ever, low skilled unemployment was low due to blue collar workers getting jobs in construction and additional jobs in real estate. Besides, saving-oriented investors saw their demands met for low risk and would purchase the high quality AAA mortgages (Petersen et al. 2011). Hedge funds were making above-average returns with the most risky “unrated CDO’s”. The economy thus seemed to be striving, but problems slowly started to arise in the form of a saturated housing market. Fewer and fewer investors who applied for a mortgage would meet the lending requirements. For this, short term solutions were brought into life as well, with the introduction of “sub-prime mortgages”, for people who may not succeed in maintaining the schedule for repaying the debt, sometimes as a result of setbacks, such as divorce, medical emergencies and unemployment (National Commission on the Causes of the Financial and Economic Crisis in the United States, 2011). This then allowed for the cycle of credit to keep growing.

Financial innovation was heavily blamed for the crisis by many economists and specialists (Krugman and Robin 2010, Spence 2008, Simkovic 2011,Volcker 2009, Financial Crisis Inquiry Commission 2011). Only after the crisis, the pointers were aimed towards the real estate debacle that inevitably had a large impact on the rise of the 2008 crisis.

2.3.2 EFFECTS OF THE 2008 CRISIS

(19)

19 collapse of financial systems throughout the developed world. The fall of these stock markets had severe social and economic consequences for both the US and Europe (Baily and Elliott 2009). Homeowners saw their equity disappear as house prices fell drastically. The worsening of the general economic situation led to a situation in which homeowners that already struggled to keep up with mortgage payments, started to default payments. This prompted a massive wave of evictions and decreased social standards of life for a considerable part of the population.

In the United States, the two main effects of the 2008 crisis were the rise in unemployment and in national debt. As a result, the government had to plan, since they were worried that this would cause a deflationary spiral. Such a spiral entail that a decrease in prices, leads to lower production, followed by lower wages, and demands and the restart of this vicious cycle by further lowering prices.

(20)

20 labour in order to keep the costs down. This, and other factors that put the economies under pressure, inevitably led to a vicious cycle of debt, low growth, people losing trust in the economy, and high unemployment (Blyth 2013). Due to this loss of trust in the economy, individual consumers start holding onto their money, which diminishes spending and slows the economy down even more (Blyth 2013). This occurred not only in the south of Europe, but it was most severe here. Therefore, the PIIGS are taken apart in this research, to find out whether there are major differences with regard to necessity-driven and opportunity-driven entrepreneurship.

2.4 HYPOTHESES

So far, the three concepts that were elaborated upon in the previous three paragraph are not yet put together to create a common perspective for this research. In the introduction of the research, the following research question was stated: “Does the economic crisis of 2008 change the intrinsic motivation towards entrepreneurial activity for potential entrepreneurs in Europe and the United States?”. This question incorporates the concepts of all these three paragraphs: entrepreneurship, the motivation of potential entrepreneurs (push-pull theory), and the 2008 crisis. In order to state some expectations for the research outcomes, the current paragraph will create links between the concepts in order to form a number of hypotheses about the results.

2.4.1. ENTREPRENEURSHIP DURING THE 2008 CRISIS

(21)

21 prosperity of the country create differences between the situation of for instance a potential entrepreneur in The Netherlands compared to a potential entrepreneur in Spain.

However, a period of economic downturn, as during the financial crisis, has similar effects for potential entrepreneurs everywhere. As a result of the crisis, banks stopped handing out loans with ease, and requirements for you to get a loan sharpened. Such a situation is called a “credit crunch” (Brunnermeier 2008). In the past fifty years there have been 28 credit crunches, which makes it a regularly occurring phenomenon (Herring and Wachter 1999).

Therefore, it became more important to have your own capital to start a venture, in the form of savings. There are two ways to handle such savings. On the one hand, the potential entrepreneur might decide that it is better to wait for a new job opportunity and use the savings as a support for the current period of job insecurity. On the other hand, it might seem logical that the potential entrepreneur might belief that he/she will not find a new job in the coming time and therefore waiting might lead to spending saving without having a long-term solution. These ideas can be strongly linked to the way push entrepreneurs think: they start a venture because they do not believe they will find a better job soon or are unhappy about their current employment situation.

When looking at the way different potential entrepreneurs spend their capital, economic downturns may have another substantial influence. Such a period may increase the availability of capital as firms enter insolvency and thus lower the cost of capital (Binks and Jennings 1986). Capital in this sense can be equipment, building, tools etc. To a potential entrepreneur this may send out different signals. On the one hand, it shows him/her that the current economic situation might not be easiest to thrive (especially if you are just starting a business). However, on the other hand it makes the cost of starting a business lower and gives a financial incentive to the potential entrepreneur to go forward with the start-up in that moment.

In the research we will be focusing specifically in the cluster PIIGS and US. Despite this we will include some observation of other European countries into the analysis in order to observe patterns in the behaviour of the entrepreneurs. The reason of the less emphasis on the European countries has to do with the research design reasons that is going to further explain the research design, research methods, and lastly further clarified in the limitations.

(22)

22 morph in European sovereign debt crisis show that this countries (PIIGS) expose the problems of competitiveness (Zemanek, H. 2010) that were distinct from the rest of Europe. The problems of competitiveness in the PIIGS lead to a situation where the impact of the crisis where felt with more intensity than the rest of Europe. This distinction between the PIIGS countries and the rest of the countries in Europe gives us a scenario to analyze the impact of a shock in the entrepreneurs. The rationale goes that if the shock hit the PIIGS countries hard then it will be more likely that an effect in the different entrepreneurs can be observed. The same goes for the US that was in the epicenter of the crisis of 2008. The US, as mentioned previously, suffered a massive impact from the series of economic consequences resulting from the crisis. The choice of making the US one of the focus of the research comes from the idea that the impact in the US was more severe since the country was where the 2008 crisis started and develop primarily before reaching Europe.

2.4.2. EFFECTS FOR NECESSITY-DRIVEN POTENTIAL ENTREPRENEURS

When taking the previous links between the crisis and entrepreneurs into account, it may become apparent that the effect of the crisis on motivations towards entrepreneurial activity might be different for push and pull entrepreneurs.

(23)

23 usually more affected with layout due to their time in the firm, competency, and substitutability (Cheung 2014).

People hold onto their savings in order to have a backup in case they lose their job. But this period of economic downturn soon turned out to be more severe than the typical fluctuations. Even though the effects of unemployment are always hard to forecast, due to the severity and attention that the 2008 crisis got, it is a logical response that necessity-driven entrepreneurs were directed to self-employment and business creation. Moreover, the effect of the availability of capital after the first shock of the crisis is likely to have affected necessity-driven potential entrepreneurs. Real Estate is usually one of the most expensive investments when starting new venture. After the wave of default of the subprime mortgages, a massive amount of real estate was set free at a low price (Coleman, LaCour-Little and Vandell 2008). These low prices might give a necessary push for necessity-driven entrepreneurs to start up their own ventures, because the typical size of such a venture is usually small at the beginning, which makes self-finance possible.

Besides, there are many state supported programs that help small entrepreneurs out at the beginning, by subsidizing or giving incentives to stimulate the creation of new businesses (OECD 2009). In theory, this availability of capital should not influence the decision of a necessity-driven potential entrepreneur, since he/she is mainly driven by dissatisfaction with the current situation. However, regardless of the idea that capital might not be the main trigger to the necessity-driven entrepreneur, it is still important that he/she needs certain level of capital in order to be able to go forward with the venture.

Taking this all into consideration, it seems to be the expectation that the 2008 crisis has an impact on the motivation of the necessity-driven potential entrepreneur. Therefore, hypothesis 1 a:

Hypothesis 1a (H1a): Necessity-driven potential entrepreneurs will increase their entrepreneurial

activity as a result of 2008 crisis in the US and Europe.

can be stated as follows:

(24)

24

Hypothesis 1a (H1b): Necessity-driven potential entrepreneurs will increase their entrepreneurial

activity as a result of the 2008 crisis in the United States.

Hypothesis 1b (H1c): Necessity-driven potential entrepreneurs will increase their entrepreneurial

activity as a result of the 2008 crisis in the European PIIGS countries.

2.4.3. EFFECTS FOR OPPORTUNITY-DRIVEN POTENTIAL ENTREPRENEURS

The opportunity-driven potential entrepreneur, as stated before, is pulled towards entrepreneurial activity due to the opportunities they see there and the appeal of a business idea. This is a vastly different motivation to become an entrepreneur when compared to the necessity-driven potential entrepreneur. Therefore, expectations with regard to the influence of the 2008 crisis are also different.

This relation with the 2008 crisis is less obvious than with necessity-driven entrepreneurship. The profile of an opportunity-driven entrepreneur is radically different, since they tend to earn more, be more educated (Fossen and Büttner 2012), are more risk tolerant (Block, Sandner, Spiegel 2015) etc. These character traits should not differ over time, and therefore quite robust to influences of periods of economic downturn as was the case with the crisis.

As opportunity-driven potential entrepreneurs do not participate in entrepreneurship due to unemployment and other negative situations, it could be expected that the 2008 crisis does not affect the trigger to start a venture. In the existing literature, there is no conclusive answer to this matter so far.

When taking the character trait of this type of potential entrepreneur into account, it seems logical that the 2008 crisis should not drastically influence the intrinsic motivation towards becoming an entrepreneur. As a result, the following hypotheses can be stated:

Hypothesis 2a (H2a): Opportunity-driven potential entrepreneurs will increase their

(25)

25 The profile of the opportunity entrepreneurs seems to be vulnerable to the effects of unemployment, low growth, uncertainty. Therefore, the effects of the crisis on the opportunity entrepreneurs of US and PIIGS, despite their differences in terms of institutions, culture and effects of the crisis, lead us to hypothesize that:

Hypothesis 2a (H2b): Opportunity-driven potential entrepreneurs will not be affected in their

entrepreneurial activity as a result of the 2008 crisis in the United States.

Hypothesis 2c (H2c): Opportunity-driven potential entrepreneurs will not be affected in their

(26)

26

3. Conceptual framework

In this section, we will present the variables used to operationalize the experiment, as well as they interact. The interaction will be explaining through the use of the a visual representation of the model follow by a short explanation. The criteria for the choice of this variable will be expanded in the methodology section. But first we are going to define the constructs in the research.

3.1 CONSTRUCTS AND VARIABLES

To make it coherent, we will start by presenting and definition the variables by dependent, independent, control and interactions.

3.1.1 DEPENDENT VARIABLES

The research is going to include 2 dependent variables in order to study 2 different types of motivational entrepreneur. The concepts of necessity and opportunity entrepreneurship as was mentioned previous have origin in the research of Reynolds et all (2002) and the data is original from the GEM consortium (more information about the process of data collection in the

respective section).

The Necessity entrepreneurship (or Necessity-driven early-stage entrepreneur) is defined by Reynolds et all (2002) as the “Percentage of those involved in TEA who are involved in

entrepreneurship because they had no other option for work”. When explaining the origin of the construct in the literature we mentioned that there many variations of the construct that do not focus exclusively in the work factor but we will be using this definition since our data is original from the GEM.

(27)

27 From these descriptions, it becomes clear that GEM altered the outcomes of the survey so the final data file showed percentages for both necessity-driven (1) and opportunity-driven (2) entrepreneurship.

3.1.2 INDEPENDENT VARIABLE

The variable “Tally” is the representative of the impact of the crisis in the research. The variable had origin in the paper “This time is different: eight Centuries of Financial Folly” (2009) by Carmen M. Reinhart and Kenneth S. Rogoff. Despite the tally being used to represent the impact of the crisis of 2008 it is not individual variable but a compound one since it compiles 5 differents types of crisis. The five “varieties” of economic crises are “external default”, “domestic default”, “banking crises”, “currency crashes”, and “inflation outbursts”. The variable and sub-variables are both dummy variables which means that they take on the value of one when there is a crisis and zero when there not observation of a crisis. Therefore the variable tally varies between a range of 0 and 5. A value of 5 would mean that the 5 types of crisis are happening in simultaneous. Reinhart and Rogoff (2009) define limits or thresholds that define if we are, for example, just in a fluctuation of inflation or in an inflation crisis. See Appendix 46.

The parameters to define if we are a crisis or not for the diverse variable are:

Inflation outburst: “An annual inflation rate 20 percent or higher. We also examine separately the incidence of more extreme cases where inflation exceeds 40 percent per annum.”

They split the sub-variable “Currency crashes” in “Currency debasement type 1” and “Currency debasement type 2”

Currency crash: “An annual depreciation versus the US dollar (or the relevant anchor currency—historically the UK pound, the French franc, or the German DM and presently the euro) of 15 percent or more”

Currency debasement type 1: “A reduction in the metallic content of coins in circulation of 5 percent or more”

(28)

28 External Debt crises: “A sovereign default is defined as the failure to meet a principal or interest payment on the due date (or within the specified grace period). The episodes also include instances where rescheduled debt is ultimately extinguished in terms less favorable than the original”

Domestic Debt crisis: “The definition given above for external debt applies. In addition, domestic debt crises have involved the freezing of bank deposits and or forcible conversions of such deposits from dollars to local currency.”

Much like the “Currency crash” sub-variable, the banking crisis also presents a split. Reinhard and Rogoff consider that the a “Banking crisis” can type 1: systemic/severe or, type 2: Financial distress/ Milder.

Banking crisis type 1: “bank runs that lead to the closure, merging, or takeover by the public sector of one or more financial institutions”

Banking crisis type 2: “if there are no runs, the closure, merging, takeover, or large-scale government assistance of an important financial institution (or group of institutions), that marks the start of a string of similar outcomes for other financial institutions.”

The variable tally takes the averages across all countries (or region) and “construct a composite index of financial instability that is multidimensional”.

USDummy and PIIGSDummy are dummies variables that were created in order to capture the country specific effects. In the USDummy the value 1 is assigned to the US and all the other countries are assign the value 0. The PIIGSDummy assigns the value of 1 to Italy, Spain, Greece and Ireland, and 0 to the rest of the countries in the sample. Unfortunately, the data for Portugal (the P in the acronym PIIGS) was missing and it was not possible to include the the country in the analysis. Even though Portugal was not included in the analysis we are going to keep the acronym PIIGS.

3.1.3 CONTROL VARIABLES

(29)

29 Human Development index, unemployment etc. However, when we proceed into the quantitative part of the research we ended up having problems with missing data and multicollinearity

(discussed more thoroughly in the methodology part). The results was that we end up with 2 control variables, productivity and housing prices. See Appendix 46.

The variable productivity (or “labour productivity”) is uses the concept of the GDP per hour in order to get a measure of how efficient the labour input is when combine with other production factors. OCDE defines labour input as “total hours worked of all persons engaged in production”. Productivity as define by the OCDE “only partially reflects the productivity of labour in terms of the personal capacities of workers or the intensity of their effort”. Another important thing to point out is that the measure is dependent on other inputs that just labour (e.g. capital,

intermediate inputs, technical, organizational and efficiency change, economies of scale). The construct is measured in USD (constant prices 2010 and PPPs).

The data used for “House prices” construct was the adjusted version of the House prices

(nominal). We used the data relative to “Real House prices” which is the “ratio of nominal price to the consumers’ expenditure deflator” since it would eliminate some of the possible

disturbances on the data. See Appendix 46.

3.1.4 COUNTRY SPECIFIC VARIABLES

The interaction variables were included in order to study the effect of crisis (“Tally”) on the US and PIIGS countries, this lead to the creation of the variables “tallyUS” and “tallyPIIGS”. This variables we only included later on the research since by mistake we assume that just the inclusion of dummies variable without the interaction would be enough to test the tally against the countries. We will explain later on in the statistical techniques the why of the later inclusion of the interactions.

(30)

30 3.1.5 Equation Model

There are 2 main equation model relative to the 2 dependent variables. However, the use of logarithmic models in the statistical part would theoretically change the presented equations. The utility of a equation model is to simply the understanding of the model. The inclusion of all the different variations of the model would go against this idea. Therefore, we are only going to present the simplified versions of the equation.

The model equation for the necessity entrepreneurship dependent variable is:

𝑁𝑒𝑐𝑒𝑠𝑠𝑖𝑡𝑦_𝑖,𝑡 = 𝛽₁+ 𝛽₂𝑇𝑎𝑙𝑙𝑦 + 𝛽₃𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦 + 𝛽₄𝐻𝑜𝑢𝑠𝑖𝑛𝑔𝑃𝑟𝑖𝑐𝑒𝑠+𝛽₅𝑈𝑆𝐷𝑢𝑚𝑚𝑦 + 𝛽₆𝑃𝐼𝐼𝐺𝑆𝐷𝑢𝑚𝑚𝑦 + 𝛽₇𝑇𝑎𝑙𝑙𝑦𝑈𝑆 + 𝐵₈𝑇𝑎𝑙𝑙𝑦𝑃𝐼𝐼𝐺𝑆

The model equation for the Opportunity entrepreneurship dependent variable is:

𝑂𝑝𝑝𝑜𝑟𝑡𝑢𝑛𝑖𝑡_𝑖,𝑡 = 𝛽₁+ 𝛽₂𝑇𝑎𝑙𝑙𝑦 + 𝛽₃𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦 + 𝛽₄𝐻𝑜𝑢𝑠𝑖𝑛𝑔𝑃𝑟𝑖𝑐𝑒𝑠+𝛽₅𝑈𝑆𝐷𝑢𝑚𝑚𝑦 + 𝛽₆𝑃𝐼𝐼𝐺𝑆𝐷𝑢𝑚𝑚𝑦 + 𝛽₇𝑇𝑎𝑙𝑙𝑦𝑈𝑆 + 𝐵₈𝑇𝑎𝑙𝑙𝑦𝑃𝐼𝐼𝐺𝑆

Where tallyUS and tallyPIIGS are:

𝑇𝑎𝑙𝑙𝑦𝑃𝐼𝐼𝐺𝑆 = 𝑃𝐼𝐼𝐺𝑆𝐷𝑢𝑚𝑚𝑦×𝑇𝑎𝑙𝑙𝑦 And

𝑇𝑎𝑙𝑙𝑦𝑈𝑆 = 𝑈𝑆𝐷𝑢𝑚𝑚𝑦×𝑇𝑎𝑙𝑙𝑦

In order to understand the nomenclature in the logarithmic models it might be useful to used the follow chart.

Linear variable Logarithmic variable

Necessity Necesslog

Opportunity Opportlog

Productivity Prodlog

(31)

31 3.1.6. Visual Model

The visual model allows to have a better understanding of how the concepts interact in the model.

4. RESEARCH DESIGN

(32)

32

4.1 RESEARCH METHODS

The research is going to be split in two parts. The first are going to make use of data relative to necessity and opportunity entrepreneurship to conduct a visual analysis of the reactions of both entrepreneurs to the period pre and post crisis of 2008. Second, we are going to use Panel data analysis to try to analyze if the hypothesis can be statistically confirmed. The use of Panel data analysis is necessary to objectively prove (or not) the effect of the 2008 crisis on the different types of entrepreneurs.

The dataset that was used for this research was obtained from the Global Entrepreneurship Monitor (GEM) and was slightly adjusted to fit the current research. A selection was made to only include countries of which data was mostly available for the years 2001-2015 and to only include European countries and the United States. Countries that had more than two years missing were excluded from the dataset. In the end, the dataset includes 17 European countries and data from the United States. This sample was chosen because of the impact that the 2008 crisis had on the US and Europe. As a result of economic globalization, many other areas were also affected by the crisis. However, it was only in Europe and in the US that the effects led to structural vulnerabilities and a necessity towards extreme economic reforms (Edey 2009). The main criteria for the sample selection was the fit and number of observations. In a lot of cases it would have been desirable to have complementary data in order to make a more complete analysis. For example, the lack of observations for Portugal prevents a complete analysis of the economic group PIIGS, that was introduced before.

The available and relevant data points were summarized in a new Excel dataset. These points were then used to create graphs for each country in the range 2001-2015 for necessity-driven entrepreneurship and 2005-2015 for opportunity-driven entrepreneurship. These graphs reveal the pattern in which both necessity-driven and opportunity-driven entrepreneurship develop throughout the years before, during and after the hardest hit of the 2008 crisis. Besides, an all the European countries are aggregated all together, in order to be able to make a single graph for the whole European area. The same calculations were made for the PIIGS area, of which Portugal was not included due to a lack of data. This aggregation is performed with the following formulas:

(33)

33 𝐸𝑛𝑡𝑟𝑒𝑝𝑟𝑒𝑛𝑒𝑢𝑟𝑠ℎ𝑖𝑝 𝐼𝐼𝐺𝑆 20𝑋𝑋 = 𝐸𝑛𝑡𝑟𝑒𝑝𝑟𝑒𝑛𝑒𝑢𝑟𝑠ℎ𝑖𝑝 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒𝑠 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑖𝑛𝑐𝑙𝑢𝑑𝑒𝑑 𝐼𝐼𝐺𝑆 − 𝑐𝑜𝑢𝑛𝑡𝑟𝑖𝑒𝑠 𝑖𝑛 20𝑋𝑋

𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐼𝐼𝐺𝑆 − 𝑐𝑜𝑢𝑛𝑡𝑟𝑖𝑒𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑑𝑎𝑡𝑎𝑠𝑒𝑡

The XX in the formula can be filled in with the corresponding year for which the average is calculated. The data points for all the years (2001-2015 for necessity-driven and 2005-2015 for opportunity-driven entrepreneurship) form the input for the graphs that show the path that entrepreneurship followed around the crisis.

4.2 DATA COLLECTION

The country-level data concerning necessity-driven and opportunity-driven entrepreneurs was thus obtained from GEM. Therefore, all the data used in the current research is secondary data from GEM that was slightly adjusted in the manner mentioned before. This data is part of a GEM consortium project, which is a yearly occurring assessment of the entrepreneurial position of countries. This assessment is divided into the National Expert Survey (NES), in which 36 “experts” collect socioeconomic data related to entrepreneurship, and the Adult Population Survey (APS), which collects surveys concerning entrepreneurial aspirations, attributes and capabilities from a sample of at least 2000 entrepreneurs from diverse countries. It is assumed that GEM draws representative samples from these countries, but the samples of 2000 entrepreneurs give the idea that this is the case. The surveys that GEM makes are composed of a number of standardized questions with a multiple-choice format. Subject that are interviewed fall within the population group of ages between 18 and 64, the working part of the population. The conduction of these surveys is directed by the Global Entrepreneurial Research Association (Xavier, Kelley, Kew and Herrington 2013).

(34)

34 most important motive for pursuing this opportunity?”. Again, fixed answers were provided to the interviewee. Possible answers are here “Greater independence”, “To increase personal income”, or “To maintain income”. Also, an open field was provided in the case the individual felt there was another reason that was not listed above.

4.3 STATISTICAL TECHNIQUES

In this section, we are going to present the results for the tests of the assumptions and methods used in panel data specific models. The protocol used was based on the Oscar Torres-Teyna “Panel data analysis - Fixed and Random effects using Stata” (2007). The analysis of the data started with checking for balance within the data. The correlation was accounted for first by reproduction and analysis of the pairwise correlation chart and later through testing for multicollinearity. The test for Multicollinearity led to the elimination of some of the variables that highly correlated.

The choice for a fixed or random effect model was made by using the “Hausman fixed random test”. To test if a pooled regression would be more adequate than random effects estimator we used the “Breusch- Pagan Lagrange multiplier”.

The “Pesaran CD Test” was used to check for cross sectional dependence on panel data. Even though the phenomena of cross sectional dependence are more concerning in panel data with 20 or more years (ours is only 10 years), we decided to perform the test anyway. The heteroskedasticity was only checked for fixed effect models since random effect models. Due to the nature of the random effect estimator we do not test for heteroskedasticity. Much like the tests for cross-sectional dependence, the problem of serial correlation (or autocorrelation) in panel data is usually more associated with long time-series (20 or more years). Despite this, the test for serial correlation was conducted as a precaution.

One of the biggest concerns of the data collection was to be sure that we had little to no observations missing. The “strongly balanced” tells us that there are data for all the countries and years.

Balance of dataset

Panel variable: Ncountry (strongly balanced) Time variable: Year, 2004 to 2014

(35)

35 The next chart shows the summary of the data with some basic reference indicators.

Variable Obs Mean Std. Dev. Min Max

opportlog 165 1.713885 .1087437 1.264346 1.911179 Opportunity 165 53.27658 12.25569 18.38 81.50396 necesslog 165 1.177712 .2516429 .5084794 1.594873 Necessity 165 17.44662 8.773517 3.224626 39.34347 prodlog 165 1.994523 .020099 1.914279 2.047157 Productivity 165 98.85039 4.471704 82.0878 111.4697 houprlog 162 2.001422 .0639845 1.829244 2.27541 HousingPrice 162 101.4461 15.83254 67.49066 188.5429 USDummy 165 .0666667 .2502032 0 1 PIIGSDummy 165 .2666667 .4435628 0 1 tallyUS 165 .0242424 .1897211 0 2 tallyPIIGS 165 .2545455 .5911393 0 3

(36)

36 Necessity Opportunity prodlog houprlog Tally PIIGSDummy USDummy Necessity 10,000 Opportunity -0.5943* 10,000 0.0000 prodlog -0.0021 0.0667 10,000 0.9788 0.3945 houprlog -0.1800* 0.2632* -0.3308* 10,000 0.0219 0.0007 0.0000 Tally 0.1580* -0.2777* 0.0744 -0.0907 10,000 0.0427 0.0003 0.3425 0.2511 PIIGSDummy 0.1853* -0.3114* -0.0231 -0.1358 0.1984* 10,000 0.0172 0.0000 0.7680 0.0849 0.0106 USDummy 0.0308 0.1716* -0.1277 0.1431 -0.1186 -0.1612* 10,000 0.6948 0.0275 0.1023 0.0692 0.1291 0.0386

and for the logarithmic variables

necesslog opportlog prodlog houprlog Tally PIIGSDummy USDummy necesslog 10,000 opportlog -0.5445* 10,000 0.0000 prodlog -0.0123 0.0549 10,000 0.8757 0.4834 houprlog -0.1422 0.2973* -0.3308* 10,000 0.0711 0.0001 0.0000 Tally 0.1512 -0.2828* 0.0744 -0.0907 10,000 0.0526 0.0002 0.3425 0.2511 PIIGSDummy 0.2230* -0.3364* -0.0231 -0.1358 0.1984* 10,000 0.0040 0.0000 0.7680 0.0849 0.0106 USDummy 0.0653 0.1739* -0.1277 0.1431 -0.1186 -0.1612* 10,000 0.4046 0.0255 0.1023 0.0692 0.1291 0.0386

(37)

(38)

38 Multicollinearity analysis

Variable VIF 1/VIF

prodlog 2.21 0.452808 houprlog 2.20 0.453789 Tally 3.04 0.329173 PIIGSDummy 5.87 0.170213 USDummy 1.98 0.504203 Ncountry 2 1.90 0.526744 3 1.96 0.509733 4 1.88 0.530914 5 1.97 0.507621 6 2.00 0.500033 7 1.83 0.545402 8 1.97 0.506609 9 1.94 0.514300 10 2.25 0.443999 11 1.87 0.533867 12 1.95 0.511797 14 1.90 0.526102 Year 2005 1.95 0.513433 2006 2.27 0.440297 2007 2.61 0.383579 2008 4.06 0.246285 2009 2.55 0.392111 2010 2.84 0.351947 2011 2.98 0.335161 2012 2.72 0.367114 2013 2.62 0.381520 2014 2.99 0.334667 Mean VIF 2.46

(39)

39 square of the variable or the use of logarithmic models to the dependent and independent variables (add references). To add up to the problems of nonlinearity and nonnormality, some of the logarithmic models that would not pass the Hausman test. In the end, we had to try different combinations of models to see which one would produce the best results. We created a chart that tested the linearity though the “Ramsey reset test” (See Appendix A.17), normality thought the “Shapiro-Wilk test” and “jarque-bera test” (See Appendix A.16), and the appropriate estimator though the Hausman test. See (Appendix A.19) In the “Hausman test” we must test the null hypothesis: random effect is appropriate and compare it with the alternate hypothesis: Fixed model is appropriate. If the Prob>chi2 is higher than 0.05 then we are in the presence of a random effect model, if the value is smaller than 0.05 then the model should use fixed effect estimator. See (Appendix A.18). Another aspect that we also tested was a full model which includes the variable “Fear of failure” and a partial one where the variable is not included. We discovered that including (or not) this variable would impact the normality and linearity in a significant way.

The Log-Lin model was not considered in the chart, since for most of the combinations it would not pass the Hausman test and revealed to be less fit than the presented models.

Unfortunately, we were unable to produce a model that would pass all the tests. Curiously, some models passed the “jarque-bera test” but not “Shapiro-Wilk test” even though both tests are intended to detect non-normality in the distribution of the residuals. In the end, we identified a couple of models for opportunity and necessity entrepreneurship that seem to deal with normality and linearity the best while still passing the Hausman test. The model that best fit the assumptions for necessity entrepreneurship was the Log-Log random partial model since it passed the linearity test, one of the normality tests and it did not violate the assumptions of the Hausman test. For the opportunity entrepreneurship, the best model (according to the previously mentioned criteria) would be the Lin-Log fixed model. However, in order to compare dummy variables, we decided to do fixed and random for all models since only the random model gives outputs for the dummy variables. This has to do with the theoretical nature of the model. The fixed effect model was also included in the research for comparative motives.

(40)

40 estimator, otherwise a OLS pooled regression is more appropriate. All the models passed the test and thus this suggests that there are significant differences between the countries.

The “Pesaran CD test” tests the existing correlation in the residuals which would mean that there is cross sectional dependence. (See Appendix A.21) In this test, the null hypothesis refers to residuals across entities not presenting correlation. The null hypothesis is no cross sectional dependence and the alternate hypothesis is the existence of cross sectional dependence. All the tests fail to reject the null hypothesis. All the values of the tests are higher than 0.05 thus none of the models presents cross sectional dependence.

The test for heteroskedasticity tests the statistical dispersion of the residuals. (See Appendix A.22).This is a fixed effect estimator specific test. In case of heteroskedasticity, the assumptions of the Gauss markov theorem are broken and the significance test will not be valid. The null hypothesis is homoscedasticity (constant variance of the residuals) and the alternate hypothesis is heteroskedasticity. All the models present a Prob>chi2 smaller than 0.05, thus there is heteroskedasticity in the models.

Serial correlation or autocorrelation affects the standard errors of the coefficients and causes them to appear smaller than they are and at the same time present a higher than expected R-squared. (See Appendix A.23). The test for serial correlation in panel data is the “Wooldridge test for autocorrelation in panel data”. The outputs of the “Wooldridge test for autocorrelation” show that in all models the Prob>F and thus there is serial correlation in all the models.

In order to solve the problems of heteroskedasticity and serial correlation we used the cluster option to control for this effect. The cluster option solves the problems and heteroskedasticity and serial correlation by clustering the standard errors Cameron and Trivedi (2009). We applied the cluster option to all the models.

(41)

41 best when tested without the UStally and PIIGS dummy. However, after trying to apply the Lin-Log model to the opportunity entrepreneurship with the extra variables, it turned out to have problems with the Hausman test. In the end, we used the Log-Lin format, because it was the only that passed the Hausman test, even though we still had problems with linearity (See appendix 42) and normality (See appendix 43) much like the previous opportunity models. Even though the strategy of using the log model from the previous model did not work well for the opportunity entrepreneurship, it did work well in the necessity entrepreneurship, since we managed to get a model that was linear with a normal distribution of the residuals and passed the Hausman test (See Appendix 44). The tests for the classical assumptions for the necessity Log-Log with tallyUS and tallyPIIGS were similar to the ones with the dummies. This means that the Log-Log with tallyUS and tallyPIIGS is linear (see Appendix A.36), normally distributed (See Appendix A.37) and passes the Hausman test (See Appendix A. 38). The necessity model presented serial correlation (See Appendix. A.39) and the random effect estimator is adequate (See appendix A.40)

4. ANALYSIS

This chapter is divided in three subparagraphs. The analysis of necessity-driven and opportunity-driven entrepreneurship will be divided over two subparagraphs, followed by another subparagraph that compares these two. Paragraph 5.1 indicates the graphs that were made based on the dataset and describes the patterns that can be distinguished for necessity-driven entrepreneurs. These patterns are then connected to the expectations that were stated in the hypotheses of chapter 2. Paragraph 5.2, the same is performed for the opportunity-driven entrepreneurs. Paragraph 5.3 then contains the quantitative analysis of the models

5.1 ANALYSIS NECESSITY-DRIVEN ENTREPRENEURSHIP

The dataset consists of a large number of European countries, amongst which the PIIGS countries, and the data for the United States.

(42)

42

Figure 1: Necessity-Driven entrepreneurial activity in the United States

This pattern in the United States is quite different when compared to those in Europe, even though the crisis had similar effects in both areas. The crisis in the US started in 2007, but from 2007 to 2008 there is still a decrease in the percentage for necessity-driven motivation. However, after the drop in 2008, there is an enormous peak from approximately 10% in 2008 to more than 35% in 2010. This is a tremendous increase, considering that this percentage is supposed to cover the total group of entrepreneurs in the United States. What is surprising is that from 2010 onwards, this percentage slowly starts to decrease over the years, all the way down to approximately 14% in 2014. Therefore, it can be said that hypothesis 1a cannot be completely be accepted, but partially. There is a very large increase shortly after the hardest point of impact of the crisis, and this is in line with the hypothesis. However, the drop after 2010 is not according to what was expected. Therefore, hypothesis 1a can only be partially accepted.

(43)

43

Figure 2: Necessity-Driven entrepreneurial activity in Europe

When analyzing the curves in Figure 2, it can be quickly seen that an M-pattern stands out. Apparently, right after 2003, the necessity-related motivation to become an entrepreneur becomes stronger, which diminishes again slightly until a large drop in 2007. This is an interesting finding, as the crisis started in Europe in 2008. After this drop, large increase can be seen for the most severe years of the crisis and the years following. This is in accordance with hypothesis 1b, that stated that necessity-driven potential entrepreneurs will increase their entrepreneurial activity as a result of the 2008 crisis. The data in this dataset discussed the motivation of entrepreneurs, which is the reason of these individuals for becoming an entrepreneur. Therefore, it can be said that when the percentages in this graph rise, the pool of necessity-driven entrepreneurs rises in the given period. This rationale applies to all following graphs, which are based on the same type of data.

(44)

44

Figure 3: Necessity-Driven entrepreneurial activity IIGS

(45)

45

Figure 4: Necessity-Driven entrepreneurial activity individual European countries

An interesting outlier to the Europe-graph is Belgium. Its curve illustrates that only after the year 2011 a massive increase in necessity-driven entrepreneurial motivation takes place: from 10% in 2011 to 30% in 2014. This is a very different pattern compared to Europe. Another example of a country that does not follow the pattern is Hungary. The curve of Hungary has one missing value for 2003, but for the rest all data are available. The pattern around the crisis is quite different compared to Europe. From 2005 to 2006, there is a large drop from 40% to 22%. However, in the years after the crisis there is no distinct increase of the motivation for necessity-driven entrepreneurship per se. This pattern in even stronger in the United Kingdom. As the graph shows, the motivations seem to shift up and down before and after the crisis without a clear trend. These three countries thus form outliers to the European graph. Most of the other countries had a similar pattern when compared to the average for Europe, or had changes that were only different for a few percent.

5.2 ANALYSIS OPPORTUNITY-DRIVEN ENTREPRENEURSHIP

(46)

46

Figure 5: Opportunity-Driven entrepreneurial activity in the United States

To investigate the change in the motivation towards opportunity-driven entrepreneurship, GEM only had data available for the years 2005-2015. This is not a problem, since the years before, during and after the crisis are available. The pattern in Figure 5 shows that in the United States there have been two trends: a downward trend from 68% in 2005 to 51% in 2010, followed by an upward trend to almost 70% in 2015. What is interesting to see is that from 2007 to 2008, there was a slight increase in the motivation towards opportunity-driven entrepreneurship. Linking this directly to the 2008 crisis, it actually seems to be the case that when the crisis hit hardest in the US, people seemed to be more motivated to become entrepreneurs because they see opportunities. In the two years following this, the percentage for the opportunity-driven entrepreneurship motivation decreases quite fast. This would mean that people seem less motivated in the aftermath of the most severe time of the crisis. All of this is not in accordance with hypothesis 2a, that stated that opportunity-driven potential entrepreneurs will not be affected in their entrepreneurial activity as a result of the 2008 crisis.