2015 The Impact of Regulatory Burdens on Entrepreneurial Activity

(1)

S u p e r v i s o r : D r . R a q u e l O r t e g a A r g i l e s

2015

The Impact of Regulatory Burdens

on Entrepreneurial Activity

A quantitative analysis on the basis of the Entrepreneurship

Indicators Programme by the OECD

Rijksuniversiteit Groningen

Faculty of Economics and Business

Rinse Elsinga s1956310

(2)

1

Abstract

In this quantitative study, we analyzed the impact of six factors relating to regulatory burdens on entrepreneurship. The selection of variables is based on the EIP by the OECD that compiles numerous categories and factors, which impact entrepreneurship. The aim in this study is to find support for the framework and detect gaps in the literature. It was found that data to measure such a complex issue

is still missing which makes estimation difficult. Overall, it appears that time-intensive factors in starting a business are more impactful than financial matters. There are various implications for

future research to continue in this field.

(3)

2

1. Introduction

The topic of entrepreneurship is a much discussed issue that has gained more and more awareness in recent years. In today’s world, we are moving away from industrial and physical-capital economies to knowledge economies that are very much based on information (Audretsch, Grilo and Thurik, 2007; Powell & Snellman, 2004; Adler, 2001; Szerb et al., 2013). As Powell and Snellman (2004) noted, “numerous social scientists have documented the transition underway in advanced industrial nations from an economy based on natural resources and physical inputs to one based on intellectual assets”. This shift of focus brings with it a shift in where value lies; that is knowledge and information are embedded in people which makes them the most valuable force in economic growth (Powell and Snellman, 2004; Szerb et al., 2013). With people being at the core of bringing the economy forward, entrepreneurship soon comes to mind. The importance of entrepreneurship in this context is reflected in a shift from traditional protective policies that were prevalent in the industrial economies to a more innovation promoting and growth-facilitating approach in policy making (Szerb et al., 2013).

As mentioned above, entrepreneurship is a highly influential aspect to achieve economic growth (Páger, 2014; Szerb et al., 2013; Baumol & Strom, 2007; Stel, Carree and Thurik, 2005). Broadly speaking, one can make a distinction between different types of entrepreneurship (Baptista, Karaöz and Mendonça, 2013; Acs, 2006). Traditionally, there are two broad categories of entrepreneurship, namely necessity entrepreneurship and opportunity entrepreneurship (Baptista, Karaöz and Mendonça, 2013; Williams, 2007; Acs, 2006). While the effect of the former on economic development is zero, the latter has a significantly positive effect (Acs, 2006). It has even been suggested that the ratio of necessity to opportunity motivated entrepreneurs would be a good indicator of economic development, which goes so far as to suggest that there is a “positive relationship between the opportunity ratio and GDP per capita” (Acs, 2006). These findings clearly support the argument that entrepreneurship is an important factor in economic growth.

An important influencing factor of entrepreneurship is the government, the guidelines it imposes, and the policies it introduces that affect entrepreneurs (Gemconsortium.org, 2015; Lerner & Schoar, 2010). Research today is still lacking a comprehensive overview of the different ways the government can influence entrepreneurial activity in terms of defined policies (Gemconsortium.org, 2015; Lerner & Schoar, 2010, Ács et al., 2014). That is, we have no clear statistical evidence that shows in what ways and to what extent the government influences entrepreneurial activity. We do, however, have access to indicator lists that attempt to describe which certain regulatory and administrative procedures affect entrepreneurship (OECD, 2014; Acs, Autio, and Szerb, 2014). This includes, generally speaking, the time and effort that goes into the procedures of starting a company, which can be split up into more specific categories (OECD, 2014). The fairly recent shift to an information-focused economy and the strong emphasis on supporting entrepreneurs makes this a currently very hot topic. Identifying in what way the government, and in particular the regulatory guidelines, can hinder and facilitate entrepreneurship is an interesting research area (Lerner & Schoar, 2010). So, we can say that there have been attempts to identify the factors itself, but the question of how remains unanswered. And this is what this paper is trying to do: answer the question of how.

(5)

4 entrepreneurship. In this context, the governmental support (or lack thereof) and motivation behind it has always interested me.

On a final note, it is important to address the issue of data availability. Even though entrepreneurship is now an important issue and is gaining in importance, there is still a major lack of data (Roberts & Peters, 2014; Lerner & Schoar, 2010). When measuring such a complex topic, much data is required to generate valid and reliable results that can be used for policy generation. However, collecting and compiling data is not only a lengthy process, but also a difficult undertaking. To this day, there is disagreement on how to best measure entrepreneurship (Ahmad & Hoffmann, n.d.). However, recent developments have generated some very useful indices and measures of entrepreneurial activities (Ahmad & Hoffmann, n.d.). One of these is the OECD’s list of indicators that was generated within the Entrepreneurship Indicators Programme (EIP) (OECD, 2014). According to Acs, Autio, and Szerb (2014), it is “perhaps the most systematic and comprehensive approach to measuring entrepreneurship policy frameworks thus far”. Therefore, this list was used as a basis for the selection of variables. Out of the very extensive list, the focus was put on one specific category that looks into regulatory and administrative burdens that entrepreneurs have to face. This category was considered the most relevant for governments. Therefore, it was suspected that the largest levers for improving entrepreneurial activity could be found within this category. It includes factors regarding the costs and required capital that come with starting a business, the perceived burden of the regulations, the time and number of procedures that are involved, and also the perceived burden of all these regulations (OECD, 2014). We will go into more detail about each factor that is included at a later point in this paper.

Overall, the guiding research question of this paper is as follows:

“In terms of administrative and regulatory factors (identified by the OECD), where are the levers for governments to improve entrepreneurial activity?”

The rest of the paper is structured as follows. First, we will review existing literature on the topics of entrepreneurship and aspects surrounding entrepreneurship. Then, we will move on to the methodology part of the paper, where the data and the methodological approach will be discussed. This will lead us to the description and then discussion of the results. The paper ends with some policy and research implications as well as a list of the most impactful limitations of the study.

2 Theory Review

In the following section, the theoretical foundations for this paper will be discussed. First of all, the concepts of entrepreneurship and entrepreneurship policy as well as the concept of an entrepreneurial ecosystem will be explored and defined. Finally, there will be a literature review of administrative burdens and the effect on entrepreneurship.

2.1 Entrepreneurship

2.1.1 Defining Entrepreneurship

(6)

5 (Cunningham & Lischeron, 1991; Oviatt & McDougall, 2005; Shane & Venkataraman, 2000). When the topic as an academic issue was still in its infancy, it quickly became clear that there are various different definitions and that reducing it to just one is a more or less impossible task. In Cunningham and Lischerin (1991), the authors explain that back then, entrepreneurship as a field of research was still very new and that various schools of thought existed that attempted to describe and define entrepreneurship. Consequently, the authors argue that there is no one definition that would fully incorporate all aspects of entrepreneurship, but rather that different angles to view the subject lead to various different definitions. They list six different entrepreneurship models, namely “Great Person”, psychological characteristics, classical, management, leadership, and intrapreneurship. Twenty years later, the definition of entrepreneurship is still debatable. Westhead, Wright and McElwee (2011) state that “academics have utilized a wide variety of definitions. It is, however, widely recognized that entrepreneurship involves the creation of new businesses and the development of new and established firms”. In line with the previous paper, the authors explain that entrepreneurship is a “fragmented field” (Westhead, Wright and McElwee, 2011).

The issue of finding a comprehensive definition of the subject has also been discussed in Oviatt and McDougall (2005), where the authors state that “the definition of entrepreneurship […] is a matter of continuing debate and evolution”. Researchers have focused on various aspects of entrepreneurship to find a definition, but each definition is constantly questioned and developed further (Oviatt & McDougall, 2005; Shane & Venkataraman, 2000).

Overall, the picture emerges that since its inception, the definition of entrepreneurship has been rather difficult due to its fragmented character. Depending on your point of view, the definition of entrepreneurship can change. What all definitions do have in common is that they first of all recognize this issue and second of all include new business creation. Since we are focusing on the government perspective, it is interesting to consider how we would define entrepreneurship from a government point of view. As mentioned before, the goal of the government in supporting entrepreneurship is to strengthen the economy and to create jobs. Accordingly, the definition of entrepreneurship used in this paper is as follows: Entrepreneurship involves the creation of new businesses and new jobs to further innovation and economic development from a governmental point of view.

2.1.2 Measuring Entrepreneurship

(7)

6 includes for instance papers by Lee, Florida, and Acs (2004) and Stuart and Sorenson (2003). In these cases, entrepreneurship and new firm formation are used synonymously indicating, that it has been applied as a reliable proxy before. Therefore, these two indicators (self-employment and new business formation) will be used here to measure entrepreneurship.

2.2 Entrepreneurship Policy

2.2.1 The Role of the Government

Now that it has been discussed how entrepreneurship is defined, it is interesting to consider how the government can influence it. This is mainly done by introducing policies directly aiming at entrepreneurship but also indirectly aiming it at entrepreneurship, which we will come back to in the next subsection of this paper. First of all, it needs to be specified what entrepreneurship policies are. Before we go into finding a definition of the subject, let us discuss why the government has an interest in supporting entrepreneurs. Feldman, Lanahan and Miller (2011) explain that “policymakers realize the importance of supporting entrepreneurial ventures in large part because of their potential to grow and create good jobs”. Entrepreneurship is therefore an important factor in economic development and as mentioned before has an impact on the wealth generation of a country. This is recognized and supported by the government, whose goal is to propel their economy forward. Consequently, the government wishes to create an entrepreneurship-friendly environment. There are different ways of doing so, but one major tool of governments is policy aimed at entrepreneurship. To illustrate this point further, let us go over some specific examples.

In a paper by Bornefalk and Du Rietz (2009), the authors discuss the cases of Denmark and Sweden in the light of EU Agendas that have been implemented to further entrepreneurship in the member countries. While Denmark defined a very ambitious strategy that involved substantial research, defining policy areas, and introducing policies to reach targets in improved entrepreneurship, Sweden has after an initial willingness to make policy changes ceased to focus on this issue (Bornefalk & Du Rietz, 2009). Denmark has since introduced various initiatives that target entrepreneurs in their country including the Danish Foundation for Entrepreneurship or the Global Entrepreneurship Week (Danish Business Authority, 2015). In Sweden, the Swedish Entrepreneurship Forum was founded that serves as a source of information and has a strong focus on research and on connecting the academic and the real life facets of entrepreneurship (Swedish Entrepreneurship Forum, 2015). There are many more examples of specific initiatives taken by governments to support the issue, not only at national but also at other levels. McCann and Ortega-Argilés (2013), for instance, mention a list of regional government initiatives, such as the Endeavor Programme in County Kerry, Ireland or Euregional start-up initiative in Rhein-Maas-Nord, Germany.

2.2.2 Defining Entrepreneurship Policy

(8)

7 policy and public governance. Whereas public policy includes actions taken by the government and institution, public governance focuses on more informal means of supporting entrepreneurs. These two aspects combined form a thorough base for strong entrepreneurial growth by providing official as well as communal support. Apart from these two types of interventions, there are a number of indirect aspects affecting entrepreneurial behavior, such as education policy or other economic policies.

Another piece of research defines entrepreneurship policy as “government initiatives that influence the formation, viability, and commercial success of new firms” (Feldman, Lanahan and Miller, 2011). The authors bring up another interesting point namely the different levels of entrepreneurship policy and how they interact, namely the national and local level, for instance (Feldman, Lanahan and Miller, 2011). Since entrepreneurs tend to remain in the same geographical location once started, local and regional policies play a highly relevant role (Feldman, Lanahan and Miller 2011, Ács et al., 2014). But at the same time, it is crucial that the policies that are developed at a national level address the macro-level and build a base for the local policies (Feldman, Lanahan and Miller 2011). Accordingly, policies that are developed at a supra-national level have to be in line with national efforts to create an effective and efficient mechanism to support entrepreneurship.

Illustrating a different point of view is an article by Audretsch, Grilo and Thurik (2007), in which the authors argue that entrepreneurship policy is not actually focused on just promoting SMEs. To them, entrepreneurship policy is a misleading term that should be replaced by ‘entrepreneurial economy’, which goes beyond policies implemented by certain institutions but rather encompasses various facets of society. This is an interesting point, since it goes beyond the more restricted views that were previously presented. While we are trying to find a concise definition of what entrepreneurship policy is, this paper suggests that this narrow perspective is the wrong approach, Instead it seems to be a broader and less tangible concept that encompasses significantly more factors that what it would initially suggest.

Overall, the concept of an ‘entrepreneurial economy’ is a valid point and good to keep in mind. For now, we will still attempt to define the concept of entrepreneurship policy. Of course, this definition is not exhaustive and is mainly applicable for the research at hand. From the discussion above, a number of overlapping factors are noticeable. That is, to many researchers, entrepreneurship policy happens within the governing body, is aimed at new business creating activities, and aims to support these actions directly. To these, we would like to add two aspects that we deem important. First of all, the governing body is not only relevant at a federal level, but also at other levels, such as locally or regionally. Second of all, entrepreneurship policy not only exists as a direct means, but also takes indirect forms. That is, certain policies that aim mainly at a different factor can still have a large effect on entrepreneurship. Having said this, the final definition of entrepreneurship policy we will work with is as follows: Entrepreneurship policy encompasses all actions taken and policies introduced by governing bodies at all levels that significantly affect new firm creation directly and indirectly.

2.3 Entrepreneurial Ecosystems

(9)

8 system. This can be seen as an attempt to shine a new light on the topic, where past attempts based on a one-dimensional view failed. That is, various supporting factors need to be given in order for entrepreneurship to work effectively. So, viewing entrepreneurship as an isolated issue has not met much success. In the light of this view, the concept of entrepreneurial ecosystems is interesting to address and explore, which we will do in this section.

The theory of entrepreneurial ecosystems describes a collection of factors that are part of the field of and combined form entrepreneurship. In Vogel (2013), the entrepreneurial ecosystem is defined as “an interactive community within a geographic region, composed of varied and inter-dependent actors (e.g. entrepreneurs, institutions and organizations) and factors (e.g. markets, regulatory framework, support setting, entrepreneurial culture), which evolves over time and whose actors and factors coexist and interact to promote new venture creation.” Generally, entrepreneurial ecosystems consist of various clusters focusing on the various factors influencing entrepreneurship. These include among others markets, human capital, finance, training, culture, and also government (Vogel, 2013; World Economic Forum, 2014; Isenberg, 2011). Diagram 1 shows an example of how an entrepreneurial ecosystem looks like.

As explained previously, government is our main focus. Within the area of government, there are various relevant factors to consider. Due to the complex nature of the entrepreneurial ecosystem and entrepreneurship itself, we focused on a specific area within the field of government, namely the regulatory framework. We chose this specific focus for two reasons. First of all, within the government category, the regulatory framework stood out as an area that could potentially represent a big lever for governments to increase entrepreneurship. That is, it is an area in which many problems still persevere that should be addressed in the future. Second of all, it seems to be an area where many questions are still unanswered, meaning we do not yet understand how exactly entrepreneurship is influenced by the administrative regulations and where the government can effectively apply changes to reach positive results. After having established the background of the paper and the focus points, it is vital to discuss how the factors in questions are measured.

Source: World Economic Forum, 2014

(10)

9

2.4 Literature Review

2.4.1 Administrative complexity

Research in entrepreneurship in recent years has increased, but the amount of research focusing on the influence of administrative burdens and the influence of the government is still rather limited even though it has clearly been established that the government does have an effect and should use this power to improve entrepreneurial activity (Grilo & Irigoyen, 2006; Braunerhjelm & Eklund, 2014; Van der Zwan, Thurik & Grilo, 2010; Van der Zwan et al., 2013). When looking specifically at administrative burdens, several studies have found a negative effect of administrative complexity, or more specifically the perception of it, on entrepreneurship (Van Stel & Stunnenberg, 2004; Grilo & Thurik, 2008; Grilo & Irigoyen; Gohman, 2010; Van der Zwan et al., 2007; Van der Zwan, Thurik & Grilo, 2010; Grilo & Thruik, 2005; Van der Zwan et al., 2013; Luthje & Franke, 2003). There appears to be a lack between the actual and the perceived complexity of procedures, so as a first step, governments need to reduce this gap and adjust the perceived and actual situation (Van der Zwan et al., 2013; Van Stel & Stunnenberg, 2004). This negative effect of perceived administrative burdens holds true for the stage of entrepreneurial intent (Luthje & Franke, 2003; Gohman, 2010; Grilo & Irigoyen, 2006). Furthermore, there is evidence that active entrepreneurs are also negatively affected by the perceived administrative complexity (Van Stel & Stunnenberg, 2004; Grilo & Thurik, 2008; Grilo & Irigoyen, 2006; Van der Zwan et al., 2007; an der Zwan, Thurik & Grilo, 2010; Grilo & Thurik, 2005; Van der Zwan et al., 2013). Another aspect closely related to the process side of things is the time and effort put into paying taxes. One study found that on the one hand the imposed taxes themselves pose a barrier to entrepreneurship but also the administrative procedures related to taxes have a negative impact (Braunerhjelm & Eklund, 2014). Based on these results, we would expect that variables relating to administrative complexity will have a negative effect on entrepreneurship. However, in this present study, the administrative burdens are broken down into more detailed variables, which has not been done by any of the studies previously discussed. Therefore, a very general hypothesis results from this discussion that related to all of our variables that refer to procedures and processes as part of administrative burdens.

H1: The more process and procedure focused aspects of government regulation have a negative relationship with entrepreneurship.

2.4.2 Financial Factors

(11)

10 the granulated level of this present study is not reflected in the previously discussed studies. There are several variables in this study that refer to the financial side of government regulations. Since the evidence is not conclusive, but points into the direction of no significance of financial aspects, the following hypothesis results:

H1: The financial aspects of government regulation do not have a significant relationship with entrepreneurship.

2.3.4 Control variables

Apart from the administrative complexity variables and the financial factors, which together form the administrative burdens, certain control variables were included in the model. These can be very generally grouped into economic, cultural, and demographic factors.

From an economic point of view, GDP per capita has been found to influence entrepreneurial behavior (Braunerhjelm & Eklund, 2014; Van der Zwan et al., 2013). The amount of money available can have an impact on an individual’s decision to engage in entrepreneurship. Furthermore, the perceived corruption has been included (Anokhin and Schulze, 2009), as well as market dominance. Both of which can have a negative impact on entrepreneurship. In terms of cultural factors, the two Hofstede dimensions of uncertainty avoidance as well as individualism were included. Uncertainty can be seen as a counterpart to risk tolerance, which has been shown to have an impact on entrepreneurship (Van Stel & Stunnenberg, 2004; Grilo & Thurik, 2008; Verheul et al., 2010). In terms of demographic factors, tertiary education as well as unemployment was considered. Both of these, and in particular education, have been shown to have an impact on entrepreneurship (Stel & Stunnenberg, 2004; Grilo & Thurik, 2008; Grilo & Irigoyen, 2006).

For a complete overview of all papers included in this theory discussion, please refer to Appendix 1.

3 Data and Methods

This chapter fully focuses on the methodological side of the study. It is split into four sections. The first section deals with the data characteristics. In section two, the variables that were chosen, the data, and descriptive statistics are discussed. In the third part, we discuss the final models. The last part deals with the method that was used and the issues that occurred with the model.

3.1 Data Characteristics

Overall, the dataset that was used consists of 86 countries in a six-year time period from 2007 to 2012. Several databases were used including the World Bank, the GEM report, the Global Competitiveness Report, UNESCO, Hofstede, Transparency International, and World Economic Forum.

(12)

11

3.2 Variables

This part will focus on the variables that were used in this research. We will first go over the dependent variables, then the independent variables, and finally we will discuss the control variables. 3.2.1. The Dependent Variables

First of all, it is important to define how we measure entrepreneurship. In this study, it was decided to measure it two different ways, namely new businesses registered adjusted for population, self-employment. On top of that, an additional variable was generated to act as a support for the two dependent variables, namely NBO (new business owners). We will discuss each in turn.

New businesses registered adj. for population (BRPOP). The first measure of entrepreneurship can be defined as the number of new limited liability corporations that are registered each year. The data was taken from the World Bank. To better be able to compare the numbers, the data was adjusted for the population size, that is, it is measured as a percentage of the adult population. The reason this variable was chosen to measure entrepreneurship is because active entrepreneurs that newly enter the market will most likely register a new company (Lee, Florida & Acs, 2004; Stuart and Sorenson, 2003). Using the amount of additional businesses rather than stock data truly represents the trend in entrepreneurial activity, since the numbers are not influenced by the amount of businesses closing. Therefore, using the number of these registered companies can be seen to be representative of entrepreneurial activity as seen from the government perspective.

Self-employment (SELF). A second measure of entrepreneurship is self-employment. As mentioned before, self-employment is often used in the literature to measure entrepreneurship (Stel, Carree & Thurik, 2005; Blanchflower, 2000; Gohman, 2010). The self-employment variable with data taken from the World Bank represents a percentage of the population that is currently self-employed. As opposed to the previous variable, this variable does not only measure new entrepreneurial activities, but includes how many people in an economy are currently self-employed. Whereas the previous variable shows yearly trends and numbers, the self-employment variable shows the total of entrepreneurial behavior in a country.

New Business ownership rate (NBO). The support variable is new business ownership rate. The data was taken from the GEM report and is part of the TEA. NBO is similar to our first dependent variable, namely new business registered, except that it does not only include limited liability companies. New business ownership rate is a percentage of the 18-64 population that are currently managing and owning a business and made some form of payment for more than 3 months but no longer than 42 months (GEM, 2008).

3.2.2. The Independent Variables

(13)

12 splits this important subcategory into eight further subcategories, namely patent systems, business and capital taxes, income and wealth taxes, social and health security, court and legal framework, product and labor market regulations, bankruptcy regulations, and administrative burdens (OECD, 2014). However, the specifying of indicators does not stop here: each of these eight categories, again, consists of various different factors. So to summarize this complex structure: We have established that entrepreneurship should not be seen as a single-dimension topic, but rather as a complex ecosystem, within which exist different areas that each form part of entrepreneurship. One of these areas deals with policies. Within this area, there is the level of government, which again is subdivided into a range of categories. One of these categories is the regulatory framework, which can again be described by a range of determinants. One of these is the category of administrative burdens. This last category consists, as defined by the OECD, of eight different factors to this day. These eight factors form the basis of our research.

It becomes clear how complex of an issue entrepreneurship really is and how many different angles one can take to analyze it. It is important to find a defined focus for several reasons. First of all, when targeting policy implications, a specific recommendation should be given focusing on a particular issue. Second of all, data and especially longitudinal data in the field of entrepreneurship is still lacking. Therefore, one has to focus initially on fields where data is available or collect data yourself. Also, the larger and the more complex the model, the more data is required to be able to generate valid results. Since there is only limited data, the model itself is limited. A final note on the indicators selected: due to the limitations imposed by lacking data, it was decided to drop two of the variables that are in the final selection, which are ‘procedures time and costs to build a warehouse’ and ‘registering property’ (OECD, 2014). These two factors were deemed the least relevant for our study, especially keeping in mind, that we are moving towards a knowledge society, in which warehouses are not a necessity anymore. Next, we will discuss each variable in turn, including the control variables, and the resulting model specifications.

After having gone over the dependent variables used in this research, we will now discuss the independent variables and their expected effect. As previously discussed, a list of six indicators compiled by the OECD was used as a basis. Statistical evidence for the model is lacking, which inspired this study in the hopes that statistical support can be found and the model can be supported and adjusted accordingly. Based on the literature review we clustered the independent variables into two groups, namely administrative complexity, which includes all variables relating to time and processes, and financial factors, which includes all variables that represent a financial burden. The variable definitions and descriptive statistics for all variables can be found in Appendix 2 and 3, respectively.

Administrative complexity

(14)

13 and Toth, 2015). Generally speaking, however, we would expect there to be a negative effect on entrepreneurship if the perceived carried burden by entrepreneurs is higher.

Number of days for starting a business (NUMDAY). Founding and registering a new business can be a lengthy process. This variable measures the average time that is required for all administrative procedures that are involved in a new enterprise start-up (OECD, 2014). As the above explanation suggests, the variable is measured in days and the data is taken from the World Bank Doing Business report. The lengthier the process, the more we might expect people to not finish all procedures or to not even start the procedures at all. Again, we can refer to Van Stel, Storey, and Thrurik who examined the relationship between this variable and entrepreneurship and did not find a significant relationship. Based on other literature, discussed earlier in this paper, we would expect the time-related factors to have an influence. This also holds true for the following variable.

Number of procedures for starting a business (NUMPRO). This variable is defined as “all generic procedures that are officially required for an entrepreneur to start an industrial or commercial business” (OECD, 2014). This variable goes hand in hand with the previous one, but they each look at the same issue from different angles. The number of procedures can be high, but the time it takes to complete them might be little. On the other hand, it might take months in other countries to only finish two procedures. In both cases, we would expect less entrepreneurship, but both factors, namely time and number, need to be taken into account. The data is taken from the World Bank and the variable is measured in a total number of the official procedures in place.

Time it takes to prepare, file, and pay the corporate income tax, VAT and social contributions (TAXETC). This variable focuses more on the financial side of the administrative factors. The data is again taken from the World Bank Doing Business report and is measured in hours per year (OECD, 2014). Since this is another burdensome task for new entrepreneurs to deal with, it can be expected that the more intensive it is, the less people are willing to do it. A negative impact of taxes on entrepreneurship (and FDI) has been established in the past (Djankov et al., 2010). However, in the light of the entire framework, this variable was also not significant in the research conducted by Van Stel, Storey, and Thurik (2007).

Financial factors

(15)

14 Minimum capital required for starting a business (MINCAP). This variable represents another financial aspect. Again there are studies showing a significant relationship, but also studies that do not find a significant relationship (Fonseca, Lopez-Garcia and Pissarides, 2001; Stel and Stunnenberg, 2006). However, Van Stel, Storey, and Thurik conducted a study using the same list of variables, in which they found that no variable in the OECD list actually had a significant impact, except for the minimum capital required. This makes this variable especially interesting. This is highlighted by other previous results that did not find any relationship for financially related factors in general. In many countries, there is an official minimum capital required if you wish to register a new business. This is a nonnegotiable, administrative requirement that could pose an impediment to some potential entrepreneurs and therefore hinder entrepreneurial behavior. Again, the data is taken from the World Bank (Doing Business) and is measured as a percentage of GDI per capita (OECD, 2014). As we can see from the variable descriptions above, we expect all of these administrative factors to have a negative effect. However, we expect differences in significance that we hope will illustrate which factors have the largest impact. Also, the results from previous research on the financial issues are mixed. Even though there has been evidence for a negative relationship, the literature discussed earlier on points more to a non-significant relationship. Besides these six main variables in the administrative category, we need to take into account certain control variables. These will be listed below.

3.2.3. The Control Variables

Besides the variables that we wish to test, we need to include certain control variables that have been shown in the past to have a definite influence on entrepreneurship. As was shown, the field of entrepreneurship is extremely vast with numerous factors influencing it. Therefore, the list of control variables is rather limited and tries to focus on the biggest and most obvious as well as in this case relevant indicators. The controls are clustered into economic factors, cultural factors, and lastly demographic factors.

Economic factors

GDP per capita (GDPPC). The first control variable that was included is a rather obvious choice that has an impact on many things in economic research, namely GDP per capita. The variable is measured in current US dollar. With this variable one could argue in two ways. On the one hand, if people are wealthier, they might be more willing to take the risk of starting their own business. On the other hand, and this is what we believe, the less money a person has, the more willing they are to become entrepreneurs. This is mainly due to necessity entrepreneurship, that is, the people have no other choice but to become entrepreneurs to make a living.

(16)

15 Market Dominance (DOM). The final control is market dominance which essentially measures whether markets in a country are dominated by many firms or a few. This helps us control for the easy of entry from an industry side, if it’s dominated by a few firms that probably means entry is difficult. It’s measured as a survey from 1-7 and the data is from the World Economic Forum.

Cultural factors

Uncertainty Avoidance (UA) and Individualism (IND). Again, we can refer back to chapter two, in which uncertainty avoidance was a much discussed factor in entrepreneurship. On top of that, we also decided to include individualism as a control variable. These two controls are two of Hofstede’s dimensions and in order to adjust for differences in cultures between countries we include it. Uncertainty Avoidance helps us understand whether it is in a country’s culture to take risks or only go for a sure thing. Countries that have higher UA are more likely not to take a risk. Individualism represents how focused on the individual a culture is and has as its opposite collectivism. The data as well this information is taken from the Hofstede website.

Demographic factors

Tertiary Education (EDU). As discussed in chapter two, previous researched has examined the relationship between education and entrepreneurship and has found significant results. Therefore, it was decided to include this variable as a control. Tertiary Education is measure of gross enrolment ratio for tertiary education for both sexes; it’s a percentage of age group which corresponds with that level of education. The data we use is from UNESCO. The reason for including this variable as a control is to adjust for higher level of education. We assume here that higher education will help people create and start good business. However if a population is higher educated that could also mean they are more likely to get employed and thus are not as necessity driven to be entrepreneurs. Unemployment (UNEMPLY). Unemployment is the percentage of unemployed of total labor force, data is taken from the World Bank. This variable serves a similar purpose to GDP per capita it helps us understand the amount of necessity driven entrepreneurship in a country. As previously discussed, past studies have included this variable as a control.

3.3 Model

Based on the variables discussed above, we end up with the following two models:

lnBRPOPit = β0 +β1CIDit + β2YEARit + β3lnBOGRit + β4lnBCSTARTit + β5lnMINCAPit + β6lnNUMDAYit + β7lnNUMPROit + β8lnTAXETCit + β9lnGDPPCit + β10lnEDUit + β11lnUAit + β12lnCORit + β13lnUNEMPLYit + β14lnDOMit + β15lnINDit + e

lnSELFit = β0 +β1CIDit + β2YEARit + β3lnBOGRit + β4lnBCSTARTit + β5lnMINCAPit + β6lnNUMDAYit + β7lnNUMPROit + β8lnTAXETCit + β9lnGDPPCit + β10lnEDUit + β11lnUAit + β12lnCORit + β13lnUNEMPLYit + β14lnDOMit + β15lnINDit + e

(17)

16 subscripts indicate that for all the variables, each country and each year was measured. Due to correlation problems within the model, several specifications were run for each dependent variable. The correlations table can be found in Appendix 4. In terms of estimation technique, the OLS pooled model was used. Apart from, the two main models above, an additional model was run to test the robustness of the two main models:

lnNBOit = β0 + β1CIDit + β2YEARit + β3lnBOGRit + β4lnBCSTARTit + β5lnMINCAPit + β6lnNUMDAYit + β7lnNUMPROit + β8lnTAXETCit + β9lnGDPPCit + β10lnEDUit + β11lnUAit + β12lnCORit + β13lnUNEMPLYit + β14lnDOMit + β15lnINDit + e

Since the dataset does not cover a long time-span and is relatively short in regards to the number of variables that are included in the model, the fixed and random effects are not applicable in all specifications. In order to be able to use the fixed and random effects estimation technique, an additional specification was used, for which those two techniques were applicable. The specification looks as follows:

lnBRPOPit = β0 + β1lnBOGRit + β2lnBCSTARTit + β3lnNUMDAYit + β4lnNUMPROit + β5lnGDPPCit + e lnSELFit = β0 +β1lnBOGRit + β2lnBCSTARTit + β3lnNUMDAYit + β4lnNUMPROit + β5lnGDPPCit + e

3.4 Methodology

In the next section, we will discuss the methodology applied in this research. First, panel data will be discussed and the different ways to measure panel data. Then, we will go into the assumption test and the results.

3.4.1 Panel Data

In this study, we are looking at a range of countries and observe each over a period of time. This makes the dataset an example of panel data, where “we can control for unobserved individual-specific characteristics” (Hill, Griffiths and Lim, 2012). To estimate the panel data models, the estimation techniques that were applied are the pooled OLS model, the fixed effects model, and the random effects model. The dataset for this research is short and wide which makes it unsuitable to use fixed or random effects for every specification and was therefore not used for each specification of the model (Hill, Griffiths and Lim, 2012). For future research, a more substantial dataset might be available that would allow using the fixed effects or the random effects model overall, which might lead to more valid and reliable results.

3.4.2 Assumptions

When estimating panel data, there are a number of assumptions that have to be fulfilled and adjusted for if necessary. We tested our variables for normality, multicollinearity, and heteroskedasticity. Each of these issues will be discussed in turn.

(18)

17 results of these test for the variable NBO are in Appendix 5). As we can see from both graphs, the assumption of normality is not fulfilled for either of the variables. The first variable, BRPOP, has no negative values, so the peak is to the very left and gradually decreases. The second variable, SELF, has a stronger tendency towards normality, but is positively skewed. To correct for the violated assumption, we take the log version of the variables, as shown in the models above.

Next, we need to test for multicollinearity using the VIF test (Hill, Griffiths and Lim, 2012). The results can again be found in Appendix 2. The results show that the highest number observed is at 5.71, which means that the results are of no concern. The assumption of non-multicollinearity is fulfilled. Even though there is no statistical evidence for multi-collinearity, we still expect there to be a certain overlap for some variables. For example, the number of procedures and the number of days it takes to register a new business could possibly overlap, which would influence our results. We avoid this problem (also related to correlation issues) by using several specifications of the model.

The last assumption we test for is that of heteroskedasticity. To test for this assumption, the Breusch-Pagan or Lagrange-Multiplier Test was applied (Hill, Griffiths and Lim, 2012). The results, which can be found in Appendix 2, indicate that the null hypothesis of homoscedasticity is fulfilled.

Overall, only the first of these three major assumptions is violated. This violation was corrected for in the model specification designs. Therefore, we can now move on the regression results of the study.

4 Results

In this section, we will go over the generated results for each dependent variable and the robustness test. First, the results relating to new businesses registered will be discussed, followed by a discussion of the self-employment results. Then, we will go over the results for the additional dependent variable to test the robustness of our models and finally the additional panel data model regarding fixed and random effects will be discussed. The results for new businesses registered and self-employment can be found in Table 1 and 2, respectively. The results and assumption test for NBO and the additional panel data models can be found in Table 3 and 4, respectively, with the Hausman Test results being in Appendix 6.

4.1 New Businesses Registered

(19)

18 whatsoever. The fourth variable measures the number of days it takes to start a business. This variable was included in specifications 3 and 5. It only showed significance in the first with a negative relationship. Next, we have the variable that measures the impact of the numbers of procedures. It was included in specifications 2 and 5, where, in both cases, it shows a strong significance and negative relationship. The final variable we considered relates to time taken to do taxes. It was used in specifications 4 and 5 and shows significance only in the former case and it has a negative relationship.

Table 1

Registered companies (BRPOP)

Specification 1 Specification 2 Specification 3 Specification 4 Specification 5

Number of observations 166 261 261 265 166 Burden of Government Regulation -.3626673 (.3211804) .2434846 (.2862434) .7216991** (.298454) .6744149* (.3951401) .2347364 (.3533651) Costs Required -.4337304*** (.0779709) -.1812664** (.0837771) Minimum Capital Required -.0503064 (.0540365) .0345171 (.0502017) Number of Days -.3370251*** (.0686682) .0640028 (.1028451) Number of Procedures -.9488611*** (.1316531) -.8317454*** (.2374493) Time spent on taxes0 -.2860376**

(.1377818) .1606294 (.1227542) GDP per capita .3240439*** (.0912719) .1991869*** (.0733783) .2376214*** (.0770977) .4479756*** (.1104483) Tertiary Education .1916024 (.1983807) .7262552*** (.1603964) .9225427*** (.165877) 1.59135*** (.1605325) .2002901 (.1820582) Uncertainty Avoidance .2890111 (.2602799) -.2246731 (.1453088) -.3001623** (.151714) -.4681962** (.1871917) .1961494 (.2499522) Perceived Corruption .1111538 (.1231995) .139274 (.1119375) .1045094 (.1172535) .0456156 (.1344514) .1281028 (.1093677) Unemployment Rate .3384346** (.1503375) .2193547** (.0997579) .2823976*** (.1044672) .1122927 (.1213982) .2098492 (.1411951) Market Dominance -.6483398 (.4041026) -2.010642*** (.3794456) Individualism .0164511 (.1661616) .4778347** (.1925803) Constant 211.688*** (75.27251) 128.4858** (62.47844) 129.887* (66.1747) 108.7236 (75.65669) 182.5381*** (69.64337) F-Value 20.89*** 30.87*** 25.47*** 17.92*** 21.21*** Adj R-Squared 0.5466 0.5084 0.4586 0.3906 0.6476

Standard errors are in parenthesis; *** indicates significance at the 99% level of confidence; ** indicates significance at 95%; * indicates significance at the 90%; All specifications are controlled for time and the country.

4.2 Self-Employment

(20)

19 first and the last specification. Both times, there is a high significance. Similarly to the first variable, however, we again see that the relationship is a different one than what was found for the other dependent variable. In this case, the relationship is positive. Our third variable showed no significance for the previous dependent variable. It does, however, show high levels of significance for the self-employment variable in both cases. It has a negative relationship with the dependent variable. The fourth variable regarding the number of days, showed only a slight significance in the final model. It has a negative relationship. Variable number 5 (number of procedures) shows only a slight significance and positive relationship in specification number 2. The last variable relating to taxes shows moderate levels of significance in both specifications that it was used in. It has a positive relationship.

Table 2

Self-employment (SELF)

Number of observations 166 261 261 265 166 Burden of Government Regulation .094961 (.1513607) -.2043452* (.1239379) -.2384904* (.1238715) -.5061876*** (.1415755) -.0693601 (.1731211) Costs Required .179423*** (.0367448) .1377781*** (.0410442) Minimum Capital Required -.0626519** (.0254654) -.0621991** (.0245949) Number of Days .0016883 (.0285003) -.0945652* (.050386) Number of Procedures .0986144* (.0570033) .1361169 (.1163315) Time spent on taxes .1381027***

(.0493661) .1444056** (.0601399) GDP per capita -.194537*** (.0430131) -.3408361*** (.0317714) -.3549462*** (.0319989) -.0882497 (.054111) Tertiary Education -.2703478*** (.0934896) -.2480971*** (.0694486) -.2684993*** (.0688463) -.4416364*** (.0575175) -.2295108** (.0891942) Uncertainty Avoidance .0689102 (.1226605) .0940473 (.0629159) .1120169* (.062968) -.0341056 (.0670693) -.1498284 (.1224569) Perceived Corruption -.0169588 (.0580594) -.028571 (.0484668) -.0204054 (.0486654) -.0635638 (.0481729) -.0387848 (.0535816) Unemployment Rate -.01615 (.0708486) -.1402958*** (.0431933) -.1455586*** (.0433585) -.0593199 (.043496) .0562123 (.0691745) Market Dominance .0253084 (.1447867) .4701031** (.1858985) Individualism -.50867*** (.0595344) -.4644728*** (.0943492) Constant -31.63318 (35.4732) -44.04356 (27.05197) -38.11292 (27.46541) -19.69049 (27.10719) -17.25968 (34.11978) F-Value 32.49*** 53.46*** 52.50*** 48.81*** 28.08*** Adj R-Squared 0.6562 0.6449 0.6407 0.6443 0.7111

4.3 Robustness Tests

(21)

20 For the first independent variable, burden of government regulation, the NBO supports the results that were generated with the self-employment variable. In all models, this variable is significant and has a positive relationship. The second variable (costs required) is only significant in the first specification and shows a positive relationship. Again, this outcome seems to support the results of the self-employment variable. In the overall model, the variable shows no significance. In the case of the third variable (minimum capital required) the robustness test agrees with the results of the first dependent variable. All results are insignificant. The robustness test of the fourth variable shows significance only in the last specification and confirms the outcome of a negative relationship that was shown previously. Next, the number of procedures it takes to start a business has a significant and positive relationship with the dependent variable. The robustness test for the final variable (taxes) renders no significant results.

Table 3

New business ownership rate

Number of observations 104 167 167 168 104 Burden of Government Regulation .7656811*** ( .2606655) .6530441*** (.2048613) .5445953** (.2133326) .6445674** (.2466726) 1.298821*** (.2859665) Costs Required .1581209** (.063612) .0028047 (.0715479) Minimum Capital Required .0462224 ( .0539988) .0308875 (.05223) Number of Days .0792521 (.0563491) -.2267517** (.0911539) Number of Procedures .3095566*** (.1047302) .7634832*** (.2011597) Time spent on taxes .1624842

(.0987235) .1405197 (.1302692) GDP per capita -.4220343*** ( .091928) -.4221518*** (.0608327) -.4388791*** (.0635162) -.3462139*** (.1236568) Tertiary Education .3317641 (.2039406) .1810735 (.1601596) .0794615 (.1640877) -.2308824 (.1525306) .1441218 (.2062477) Uncertainty Avoidance -.6373753** ( .2430728) -.2514331** (.1036348) -.2173818** (.1054807) -.217739* (.1186542) -.346764 (.2740699) Perceived Corruption -.0990821 ( .0981599) -.112842 (.0795777) -.0955092 (.080965) -.1063766 (.0844887) -.043176 (.0939135) Unemployment Rate -.1661809 ( .124671) -.1523664** (.076565) -.1711248** (.0785064) -.1114161 (.0809874) -.0157265 (.1277185) Market Dominance -.9395628*** (.2670752) -.3384718 (.3763128) Individualism -.2152314* (.1117391) .4456448** (.2017789) Constant -132.5121** (57.25848) -96.26109** (44.13473) -88.74342* (45.5392) -74.01529 (46.91749) -108.6064* (56.75071) F-Value 4.80*** 12.55*** 11.32*** 8.43*** 5.06*** Adj R-Squared 0.2694 0.3850 0.3588 0.3078 0.3714

(22)

21

4.4 Additional panel data models: FE and RE

As previously discussed, there are three different ways of estimating our model. So far, our results only related to the pooled OLS approach. However, the fixed effects and random effects models are two slightly more desirable estimating techniques. In order to use these two techniques, the given sample needs to not only be large, but also cover a long period of time. That is, a short and wide dataset does not allow us to run the fixed and random effects estimations (Hill, Griffiths and Lim, 2012). Unfortunately, this is the case for this paper. As discussed before, there are several specifications of the model that were run due to correlation issues. One of the specifications that was used was suitable for the fixed and random effects estimation techniques. Table 4 shows the results for the fixed and random effects estimation. All four models are significant. The Hausman Test was applied to compare the random and fixed effects model (see Appendix 6). It was found that the fixed effects model is a better estimator.

Table 4

Random and Fixed effects

Registered Companies Self Employment

FE RE FE RE Number of observations 388 388 388 388 Burden of Government Regulation -1.518997 (2.72043) .4311702* (.2332407) -2.213839 (1.392447) -.0178951 (.0934845) Costs Required .2500592 (.2335023) -.1840319*** (.0575972) .5777272*** (.1195177) .1089891*** (.0233509) Number of Days .3565927 (.4534967) -.0805512 (.0805246) -.469617 (.2321214) -.0862885*** (.0322701) Number of Procedures -2.715909* (1.007099) -.5735862*** (.1730471) -.4181726 (.5154815) .0768318 (.0708309) GDP per capita .3707683 (.4241998) .2404984*** (.0536555) -.3502797 (.2171258) -.3429618*** (.0215254) Constant -4.099428 (8.334874) -7.176671*** (.6934025) 9.982968* (4.266189) 6.235925*** (.2771493) F-Value/ Wald Chi2 15.35** 292.70*** 17.46*** 664.64*** R2_(overall) _0.2498 _0.4049 _0.4064 _0.6404

R2_(within) _0.9505 _0.8288 _0.9562 _0.6780

R2_(between) _0.2312 _0.3914 _0.3991 _0.6374

(23)

22 For the second variable, namely self-employment, the results differ. Again, there is only one variable that is significant. In this case, it is the variable measuring the costs that are required to start a business and it is significant at a 1%-level. Paradoxically, the relationship is positive. That is, the higher the administrative costs to start a business, the more people are self-employed. The overall r-squared for this model is at 40.64%.

5 Discussion

After having gone over the individual results, let us now go on to the essential discussion of the outcomes. We will first go over the variables and attempt to explain and rationalize the results. This will then lead us to the policy implications and future research propositions. Finally, we will go over the limitations of this study.

5.1 Discussion of Results

5.1.1 New businesses registered

First of all, let us discuss the results that were generated for the first independent variable, namely new businesses registered. Going back to the theory discussion, we started out with dividing the variables into two general clusters, namely procedure focused factors and financial factors. It was hypothesized that the financial factors do not have a significant impact. The results at this point show that the costs required to start a business do have a negative relationship with the rate of new businesses registered. This is in line with some pieces of previous research, which established negative relationship between entrepreneurship and flack of financial support (Grilo & Irigoyen, 2006; Verheul et al., 2010; Van der Zwan et al., 2013). Another financial variable that was considered is the minimum capital required. For this variable, we find no significant results, which is in line with our expectations. We can only partially confirm our hypothesis.

The second cluster focused on procedural aspects of administrative burdens on entrepreneurship. The perceived burden of government regulation shows mixed results but does not seem to have a significant impact. This contradicts our expectations. However, as mentioned before, previous research suggested that policy makers focus on closing the gap between the actual and the perceived burden of government regulation. It could be that this has already taken place to an extent that the perception of this complexity is no longer an issue. These results are supported by the outcomes for two other procedural variables, namely the number of days it takes to register a business and the time and effort it takes to pay taxes. Both of these have a significant negative impact on the rate of new businesses registered. Again, the results are convincing enough to say that the hypothesis has been supported.

5.1.2 Self-Employment

(24)

23 company, when the administrative costs are so high. The hypothesis relating to the financial issues is not supported.

Two out of the four procedural variables have a negative effect, namely the burden of government regulation as well as the number of days it takes to start a business. This is in line with our hypothesis. However, the time spent on taxes has a positive impact, which is an unexpected result. This indicates that when measured as self-employment, entrepreneurship goes up as the time needed to pay taxes goes up. Overall, the hypothesis relating to the procedural aspects is supported. 5.1.3 Comparison

We have a number of contradicting results between the two dependent variables. There are three initial explanations for this. First of all, it could be that the results were affected by a sample bias. If a larger and more longitudinal sample could have been applied, it might have altered this outcome. A second explanation lies in the definition of entrepreneurship. Maybe, self-employment is a better measure for entrepreneurship than new registered businesses, or the other way around. Thirdly, we could attempt to justify both results by looking more closely at what they imply. Perhaps, registering a company inherently requires more administrative work and has more regulatory burdens imposed. So, whereas self-employment suffers from more regulations, actually registering a business automatically requires the government to impose more regulations.

First of all, let us discuss the outcome for the financial cluster.

The variable, which relates to the costs of starting a business, shows contradicting results. It appears that higher costs lead to less companies being registered, which makes sense. The more people have to pay, the less willing they seem to be engage in this type of activity. Self-employment, however, seems to increase the higher the costs are. This result is supported by the robustness test. A possible explanation for this could be that registering a business (the variable itself) deals with registration of a firm that self-employment does not fall under. So, when someone wants to engage in entrepreneurial activity they have the option to register a business or be self-employed. If costs are higher to register then being self-employed is the cheaper venture. Again, it could also be that due to the small sample size, the outcome is altered. Future research in this area would be interesting to bring more light to this problem.

(25)

24 results, and again we have to consider the possibility that the results were influenced by the sample and estimation bias.

Overall, it is hard to say whether our hypothesis that the financial aspects have no influence has been proven. The results are contradictory in parts; however, based on the discussion above, we conclude that hypothesis H2 has not been proven.

Moving on to the second set of variables, we will now discuss the cluster of procedural variables. One of the variables included measures of the number of days that it takes to start a business. It shows significance once for BRPOP and once for SELF. In both cases, it had a negative relationship. This means that the longer it takes to get started the less people are willing to engage in entrepreneurship. This is line with what we had previously expected.

Another variable, which measured the number of procedures, has slightly contradictory results. It is highly significant for BRPOP and shows a negative relationship. This implies that entrepreneurial activity is impacted negatively by many procedures. In one case (specification 2) it shows a positive relationship with SELF. This positive relationship is supported by the robustness test. Overall, a positive impact is surprising. But again, a similar explanation to the one mentioned above could be given. Self-employment may rise when the effort for registering a company is too big.

The last independent variable we wished to measure relates to time taken to do taxes. Again, the results are very contradictory and make a final conclusion difficult. It seems that for registering new companies, taxes have a negative effect, whereas positive effects can be noted for self-employment. For BRPOP it is in line with what we expected. However, we cannot explain the positive impact on SELF, unless the taxes do not relate to self-employment but only registering a firm. In that case, we can again argue that due to the effort that goes into taxes when owning a company pushes people towards self-employment.

Overall, the results are again partially contradictory, but point in a clear direction. It appears that the results support hypothesis H1.

All in all, we can see that the two dependent variables render contradictory results almost all of the time. This could be due to a number of things. First of all, one of them might be a better measure of entrepreneurship than the other. Also, they could be affected differently, for instance, due to the time it takes to record the data, a lagged model might have been more appropriate. Based on the results, we believe that the first variable (new registered limited liability firms) serves as a better measure of entrepreneurship in this case. However, the large amount of limitations in this research has to be considered.

5.2 Limitations

(26)

25 The most obvious and impactful limitation of this study is the lack of data. The sample of countries and years had to be majorly reduced due to missing data. Collecting the data would have been out of the scope of this paper, but is a necessity for future research.

Linked to this issue of a small and biased sample is the estimation method. Since the sample was to small and included too many factors, running a fixed or random effects model was not possible for all model specifications. We were limited to using the pooled OLS model for most specifications, which might not have been the best estimation technique for this study.

Even though the number of independent variables was too large for a fixed or random effects analysis overall, it was already a much shortened and limited list of factors. The subject of entrepreneurship is so broad and so many factors that impact it, that one has to make harsh decisions on which variables to include. As explained in the beginning chapters of this paper, we went from a very broad level to more and more detailed levels. At the lowest level, we even decided to narrow it down further and only include the (to us) most important variables. This in itself limited the results that could be generated. But at the same time, it was more than the dataset was able to work with.

Overall, these are the most prominent and impactful limitations. The large goals we set for ourselves were probably too high and not manageable at this point in time. We managed to generate some useful insights for policy makers and researchers which will be discussed next.

5.3 Policy and Research Implications

5.3.1 Policy Implications

We found conclusive evidence that time and process based factors have a negative impact; however, the role of financial factors is still unclear. Based on this observation, governments might want to focus on reducing the aspects that limit the time that needs to be invested rather than on reducing the costs. For instance, by speeding up certain processes and providing more help to entrepreneurs. The specific actions that need to be taken will vary from country to country and would require some more in-depth analysis.

5.3.2 Research Implications

(27)

(28)

27

6 References

Acs, Z. (2006). How Is Entrepreneurship Good for Economic Growth?. Innovations: Technology, Governance, Globalization, 1(1), pp.97-107.

Ács, Z., Autio, E. and Szerb, L. (2014). National Systems of Entrepreneurship: Measurement issues and policy implications. Research Policy, 43(3), pp.476-494.

Ács, Z., Szerb, L., Ortega-Argilés, R., AIdis, R. and Coduras, A. (2014). The Regional Application of the Global Entrepreneurship and Development Index (GEDI): The Case of Spain. Regional Studies, pp.1-18.

Adler, P. (2001). Market, Hierarchy, and Trust: The Knowledge Economy and the Future of Capitalism. Organization Science, 12(2), pp.215-234.

Ahmad, N. and Hoffmann, A. (n.d.). A Framework for Addressing and Measuring Entrepreneurship. SSRN Journal.

Anokhin, S. and Schulze, W. (2009). Entrepreneurship, innovation, and corruption. Journal of Business Venturing, 24(5), pp.465-476.

Audretsch, D., Grilo, I. and Thurik, A. (2007). Handbook of research on entrepreneurship policy. Cheltenham, UK: Edward Elgar.

Baptista, R., Karaöz, M. and Mendonça, J. (2013). The impact of human capital on the early success of necessity versus opportunity-based entrepreneurs. Small Bus Econ, 42(4), pp.831-847.

Baumol, W. and Strom, R. (2007). Entrepreneurship and economic growth. Strat.Entrepreneurship J., 1(3-4), pp.233-237.

Blanchflower, D. (2000). Self-employment in OECD countries. Labour Economics, 7(5), pp.471-505. Bornefalk, A. and Du Rietz, A. (2009). Entrepreneurship Policies in Denmark and Sweden Targets and Indicators. Available at: http://www.snee.org/filer/papers/518.pdf [Accessed 10 Jan. 2015].

Braunerhjelm, P. and Eklund, J. (2014). Taxes, tax administrative burdens and new firm formation. Kyklos, 67(1), pp.1-11.

Cunningham, J. and Lischeron, J. (1991). Defining Entrepreneurship. Journal of Small Business Management, 29(1), p.45.

Danish Business Authority, (2015). Initiatives Targeting Entrepreneurship - Danish Business Authority. [online] Danishbusinessauthority.dk. Available at: http://danishbusinessauthority.dk/initiatives-targeting-entrepreneurship [Accessed 10 Jan. 2015].