The estimated effect of business regulations on nascent entrepreneurship : what is the influence of establishment legislations on the rates of male and female early-stage entrepreneurs?

(1)

The estimated effect of business regulations on nascent entrepreneurship:

what is the influence of establishment legislations on the rates of male

and female early-stage entrepreneurs?

By Belle van Oostenbruggen, 10191194 Bachelor Thesis Finance & Organisation

Thesis supervisor: A. Rilović 30 January 2017

Abstract

This thesis examines the effect of establishment regulations on entrepreneurship rates across 32 countries, for the period 2005-2014. In addition, separate analyses on male and female

entrepreneurship rates investigate possible gender differences. The analyses have been done on combined panel data from the Global Entrepreneurship Monitor and World Bank databases. For the empirical analysis, the fixed effects regression method has been used that controls for all time-invariant country-specific characteristics. The main findings are as follows: First of all, no significant effects have been found of establishment legislations on total early-stage

entrepreneurship rates. Secondly, a significant difference was found between the estimated effect of cost on male versus female entrepreneurship rates.

(2)

2

Statement of Originality

This document is written by Student Belle van Oostenbruggen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

As stated in a survey prepared by Facebook in cooperation with the OECD and World Bank, “the proportion of firms less than three years old that expect to increase employment in the short term is higher than corresponding proportions for firms more than ten years old in nearly all countries” (OECD, 2016). According to their annual report, OECD states that start-ups boost labour productivity, overall employment and economic growth. Therefore, it should be favourable for governments to always aim to increase the rate of new business ventures.

On the one extreme, new business creation could solely depend on an entrepreneur’s inherent characteristics, such as her/his aspirations and overconfidence. In such a case, it can be quite difficult for policymakers and governments to find the right approaches to motivate these individuals and create a true ‘start-up ecosystem’. The other extreme is that

entrepreneurial intentions can only develop in countries where the right set of contextual, environmental characteristics are present, such as low interest rates and sophisticated educational systems. In these circumstances, policies and regulations could have a great influence on nascent entrepreneurship rates. If so, could it be the case that these contextual characteristics have a different influence on male versus female entrepreneurs? Moreover, are these differences significant to argue for more gender-specific regulations?

In previous studies, the focus has generally been on studying the different aspects of these two types of factors influencing entrepreneurship: inherent- and contextual factors. Hereby, it is especially interesting to examine the effects of those factors that can actually be adjusted for. One example of this are the entry ‘barriers’ to new business creation, which can be relatively easy altered by governments and policymakers. That is, barriers such as the time it takes to start a business, or the costs involved in the start-up process. However, what has not been done so far, is examining if these barriers have gender-specific effects on

entrepreneurship rates. In this way, one could find evidence to dismiss gender-neutral

establishment legislations1, in order to effectively increase early-stage entrepreneurship rates. Therefore, the aim of this thesis is to analyse the effect of establishment legislations on male and female nascent entrepreneurs2. To operationalize this question, relationships of

1

Term used by Audretsch, Grilo and Thurik (2007) to describe those regulations indicating certain requirements to start-up a business. 2

According to the Global Entrepreneurship Monitor (GEM, n.d.), a nascent entrepreneur is defined as “a person who is actively involved in setting up a business they will own or co-own; this business has not paid salaries, wages, or any other payments to the owners for more than three months”.

(5)

5

three start-up characteristics on early-stage entrepreneurship rates3 will be tested: the time

required to start a business, the cost to start a business and minimum paid-in capital

requirements. In this research, gender differences will be made observable by doing separate

analyses on male and female entrepreneurship rates. Thereafter, a comparison of the estimated coefficients will examine possible gender effects. The research will be done by using panel data from the Global Entrepreneurship Monitor (GEM) and World Bank database on 32 countries. By combining these two datasets, the aim of this research is to come up with new, unique results that can provide insights for future governmental policies.

This thesis will continue with the literature review (Section II), where previous and relevant research will be discussed. At the end of this section, the hypotheses will be stated. Section III will consist of the described methodology of the research, including the data used and estimation techniques. Thereafter, the results will be discussed in Section IV and the discussion will follow in Section V. At the end of this paper, the conclusion will describe recommendations for future research (Section VI).

3

According to the GEM (n.d.), early-stage entrepreneurship rates consist of the percentage of 18-64 population who are either a nascent entrepreneur or owner-manager of a new business, i.e., owning and managing a running business that has paid salaries, wages, or any other payments to the owners for more than three months, but not more than 42 months.

(6)

6

II. Literature review

a. Previous Literature

Relatively much research has been devoted to investigating the factors that contribute to entrepreneurial intentions. In order to understand which factors might contribute to the relationships examined in this paper, it is relevant to look at past literature on the entrepreneurial topic.

First of all, there are numerous meanings of the word ‘entrepreneur’. The most widely accepted definition is stated by Ahmad and Seymour (2008), explaining entrepreneurs to be “those persons (business owners) who seek to generate value, through the creation or

expansion of economic activity, by identifying and exploiting new products, processes or markets”.

Generally speaking, the factors influencing entrepreneurs can be divided into two categories: inherent- and contextual factors (Levesque & Minniti, 2006). Inherent factors are those characteristics that the entrepreneur possesses, such as his or her level of education and various personality traits. Contextual factors, on the other hand, are those characteristics from the entrepreneurs’ socio-economic environment, such as regulations and economic

development. Extensive research has been done on both types of factors. As inherent factors are expected to have a significant influence on the way contextual factors (i.e., establishment legislations) are perceived, relevant findings on both aspects will be discussed. Hereby, the emphasis will be on those factors influencing the decision to become an entrepreneur. During this process of decision making, we also expect establishment legislations to be of

considerable impact.

Inherent factors

When it comes to inherent factors, several aspects have been tested to determine their

contribution to one’s entrepreneurial choices. These factors include objective measures, such as age and education, as well as subjective measures, such as perceptual variables.

When it comes to objective measures, age and education are among the factors that are most examined. Levesque and Minniti (2006) found a negative relationship between age and entrepreneurial attitude. According to them, the willingness to become an entrepreneur peaks around a certain threshold age and declines when someone gets older. Almost a decade later, Kautonen, Down and Minniti (2014) went more deeply into the age topic by

(7)

7

considering self-employers versus owner-managers. They found an almost linearly positive relationship between entrepreneurial activity and age for individuals who prefer to be self-employers. For those aspiring to also hire other workers, the so-called owner-managers, a somewhat different relationship was found: the entrepreneurial activity increases up to a critical threshold age (late 40s). Thereafter, the relationship decreases.

When it comes to human capital, results are somewhat more contradictory. Block, Hoogerheide and Thurik (2013) used a dataset of 10,000 individuals from 28 countries and found a strong positive effect of education on the decision to become an entrepreneur. However, a year later, Mohamad, Lim, Yusof, Kassim and Abdullah (2014) tried to find similar results by questioning a total of 200 last-year students from the University Utara Malaysia. In their results, a negative relationship was found between academic achievement and the tendency to choose entrepreneurship as a career choice. Once people started their entrepreneurial careers, results look more consistent. Parker and Van Praag (2006) found a 7.2% rate of return on schooling when examining 461 entrepreneurs, supporting the idea that greater human capital increases entrepreneurial’ profits. Over six years later, Santarelli and Tran (2013) found a similar positive relationship when studying 1,398 Vietnamese ‘new-born firms’: in their sample, human capital would strongly predict the success of start-up firms.

However, as Langowitz and Minniti stated (2007), subjective perceptions should be taken into account at least as much as objective, measurable factors. Arenius and Minniti (2005) researched the effect of perceptual variables on entrepreneurship: opportunity perception, confidence in one’s skills, fear of failure and knowing other entrepreneurs. All four perceptual variables were shown to be strong predictors of the likelihood of being a nascent entrepreneur. Based on their regression results, they stated that entrepreneurs rely significantly on their own, subjective (and often biased) perceptions. Furthermore,

Koellinger, Minniti and Schade (2007) concluded that the strongest indicator for someone’s entrepreneurial propensity is his or her entrepreneurial confidence. However, they found the same (over)confidence to have a negative correlation with the approximate survival chances of nascent entrepreneurs.

The last subcategory of inherent factors that has been researched relatively extensive, is gender differences. In general, men have a higher tendency to choose entrepreneurship as a career compared to women (Mohamed et al., 2014). Van der Zwan, Verheul and Thurik (2012) researched ‘the entrepreneurial ladder’, which consists of the five levels of

engagements in the entrepreneurial process. They also concluded that men are, on average, twice as likely as women to consider an entrepreneurial career. However, further on in the

(8)

8

entrepreneurial process, when actual steps are being taken to start a business, differences between genders tend to fade away. The reason for this gender gap in the consideration-stage of the entrepreneurial process could have multiple explanations. As research findings suggest, it could be because of the higher risk aversion that women experience compared to men (Borghans, Heckman, Golsteyn, & Meijers, 2009). Also, Mueller (2004) found that men showed significantly higher risk-taking propensity than women. He analysed the gender gap in entrepreneurial traits in seventeen countries, for which he found similar results in all tested countries. On the contrary, the higher tendency towards overconfidence exhibited by male entrepreneurs could also be an explanation for this gender gap (Langowitz & Minniti, 2005). Thirdly, this gender gap could be based on the fact that women are ten percent more likely than their male counterparts to perceive finance as the only barrier to start their own business, according to findings of Kwong, Jones-Evans and Thompson (2012).

Now that the inherent factors have been discussed, this literature review will continue discussing the contextual factors.

Contextual factors

According to multiple papers, the entrepreneurial environment also plays a key role in the entrepreneurial decision process. Levesque and Minniti (2006) call these characteristics the contextual factors of an entrepreneurial ecosystem, and they consist of factors such as culture, legislation and finance opportunities. Also in this subcategory, objective and subjective characteristics have been examined.

When looking at subjective contextual factors, the effect of characteristics such as culture and economic development on nascent entrepreneurship rates have been examined. Verheul, Wennekers, Audretsch and Thurk (2001) stated in their ‘eclectic theory of

entrepreneurship’ that the national culture of a country can influence the rate of

entrepreneurship through the ‘supply’ and ‘demand’ side. On the supply side, culture can influence the individual preference for self-employment, because attitudes towards

entrepreneurship that are prevailing in a country can shape someone’s perspective. On the demand side, governments and politics are also influenced by these entrepreneurial values, to an extent that decisions on regulations and services for entrepreneurs can reflect these

cultural perceptions.

Also, the economic development of a country effects the entrepreneurial dynamics of a country. For example, Wennekers, Van Wennekers, Thurik and Reynolds (2005) found a U-shaped relationship between the rate of nascent entrepreneurship and the level of economic

(9)

9

development. Because of the complexity of the term ‘the level of economic development’, they used three different approaches to test their hypothesis. In their regression analysis, they used the variables (1) per capita income, (2) innovative capacity index and (3) various control variables, such as tax revenues and population growth. Based on the results of these three measures, they concluded that entrepreneurial dynamics depend on the stage of economic development that a country is experiencing. Acs, Desai and Hessels (2008) added to this theory that the global economy is divided into three different stages: the factor-driven stage, the efficiency-driven stage and the innovation-driven stage. These different stages require different governmental policies to foster entrepreneurial activities. Where countries in innovation-driven stage should invest in entrepreneurship education and training, factor-driven countries should mainly focus on achieving stable institutional and macro-economic environments. Policies of the countries in the efficiency-driven stage should focus on a combination of the latter two.

When considering more objective measures, unemployment rates and educational attainment-levels have been considered. According to Gennaioli, La Porta, Lopez-de-Silane and Schleifer (2013), the regional level of educational attainment is the most critical

determinant of regional development. Part of the reason is that a more educated workforce is positively correlated with the level of productiveness and innovativeness of a country’s economy. In addition, the entrepreneurs’ success is not only influenced by his/her own educational attainment, but also by the educational level of the population (Millan, Congregado, Roman, Van Praag, & van Stel, 2014). As their research findings suggest, a higher educated population provides more productive employees, which will contribute to the successfulness of start-ups. Moreover, higher educated consumers showed to have more differentiated consumer demands, with an emphasis on innovative products and services.

Findings from Audretsch, Carree and Thurik (2001) also suggest high unemployment rates to have a positive effect on entrepreneurship (the ‘refugee’ effect). The explanation of this result can be argued for as follows: when it comes to an individual entrepreneur, the distinction can grossly be made between two types. According to the GEM (n.d.), opportunity entrepreneurs are those taking advantage of a business opportunity. Necessity entrepreneurs, on the other hand, choose to become an entrepreneur because of the absence of better options. When countries face high unemployment rates, one could argue that entrepreneurship turns into a necessity, and therefore a positive relationship between unemployment- and

entrepreneurship rates can be found. On the contrary, the ‘Schumpeter effect’ (Audretsch et al., 2001) states that entrepreneurship also reduces unemployment rates.

(10)

10

Lastly, financial- and governmental policies are considered. Concerning finance related issues, research done by Brana (2013) suggest that women tend to have more difficulties than men when starting up a business. When analysing a portfolio of 3,640 microcredit applicants in France, Brana found that for the same project, women received support from fewer microcredit resources than men, on which they would also have to pay a significantly higher interest rate. Secondly, according to Verheul et al. (2001), a higher interest rate is expected to have a negative effect on entrepreneurship rates. This is because the high interest rate increases financial risk, due to the liability risks and interest payments.

Research previously done that is most similar to the research proposed in this paper comes from Van Stel, Storey and Thurik (2007). They examined the effect of several business regulations on nascent and young business entrepreneurship, using the same independent variables as this paper will. They found no significant effect of establishment factors, such as time-, cost- and number of procedures required to start a business, on the rate of entrepreneurship. The only significant effect found in their regression analysis was the negative correlation between the minimum paid-in capital requirements and entrepreneurship rates.

(11)

11

b. Hypotheses

After reviewing the relevant past literature about the topic of this thesis, the hypotheses can be stated. The hypotheses of this thesis are as follows:

- Hypothesis 1: The variable ‘minimum paid-in capital required to start a business’ has

a significant negative effect on early-stage entrepreneurship rates, both for men and women.

In line with the results found by Van Stel et al. (2007), a significant negative

relationship is expected between the independent variable ‘minimum paid-in capital required to start a business’ and nascent entrepreneurship, both on male and female rates. However, since more recent data will be used, new regressions will be carried out to confirm this expectation.

- Hypothesis 2: All tested establishment legislations have a greater effect on female

nascent entrepreneurship rates in comparison to male nascent entrepreneurship rates.

This prediction is in line with previous research on gender differences. The greater risk aversion exhibited by women (Borghans et al., 2009) compared with the lower chance for women to receive microcredit resources (Brana, 2013) could be part of the reasons women are more affected by establishment legislation.

- Hypothesis 3: Out of the three, the biggest gender gap is found in the explanatory

variable ‘cost associated with starting up a business’.

This hypothesis is in line with the findings of Kwong et al. (2012). As women are more likely to perceive finance as being the only obstacle to entrepreneurship, the relationship between female early-stage entrepreneurial activity and cost associated with starting up a business is expected to be more pronounced than the results found in the male regression results.

(12)

12

III. Methodology a. Data

To do the empirical research, data will be collected from two sources. Firstly, data will be used from the Global Entrepreneurship Monitor (GEM). This monitor tracks rates of entrepreneurship across multiple phases and assesses the characteristics, motivations and ambitions of entrepreneurs. Next to that, it analyses the national context in which these entrepreneurs operate and the way their environment has an impact on them. The publically available datasets from GEM range from the year 2001 until 2015. In these datasets, averages on entrepreneurial behaviour and attitudes can be extracted, that are based on the outcomes of a standardized questionnaire that was completed by at least 2000 adults in each country. Furthermore, interviews conducted with national experts in the entrepreneurial field have been used to identify relevant factors of the national context that play a crucial role in the entrepreneurial ecosystem. In addition, the World Bank Database can provide various collections of time series data, such as on Doing Business, World Development Indicators and Gender Statistics (The World Bank, 2016). Unfortunately, these two different data sources don’t provide full datasets on every country and/or year. Therefore, a selection will be made based on the availability of data (see Appendix, section Descriptive Statistics).

i. Dependent variables

In the GEM database, the dependent variable of this research can be found: the early-stage entrepreneurial activity for the working age population in a specific country. A distinction has been made between male and female rates. These two variables will be the dependent variables of the research model.

ii. Independent variables

The independent variables in the model can be obtained via the Doing Business World Bank database. This database “provides objective measures of business regulations for local firms in 190 economies” (Doing Business World Bank, 2016). In this database, the indicators on ‘starting up a business’ that are available are (1) cost required to start a business, (2) time required to start a business, (3) procedures required to start a business and (4) minimum paid-in capital required to start a buspaid-iness. Due to the relatively high correlation found between the second and third variables (see Appendix, Table 1), the variable procedures required to start

(13)

13

a business will not be used. For almost all listed countries, data on these variables are

available starting from 2005.

iii. Control variables

Additionally, a few control variables will be used, based on previous literature and economic theory. As Baughn, Chua and Neupert (2006) stated in their paper, ‘comparative studies of entrepreneurial activity must take the level of economic development into account’. This is in line with findings previously discussed (Wennekers et al., 2005; Acs et al., 2008). Therefore, the natural logarithm of GDP per capita will be added as a control variable. The log function will be used because it will make the interpretation of the coefficient for this variable more straightforward. Furthermore, some other time-varying variables will be used, of which the rates of secondary and tertiary education, real interest (lending) and unemployment can be found in the World Bank Database. Firstly, education will be added because of previous findings discussed in the literature review (Parker & Van Praag, 2006; Santarelli & Tran, 2013). Secondly, interest rates will be added to control for the effect found by Verheul et al. (2001). Thirdly, male and female unemployment rates are added (see Audretsch et al., 2001). Hereby, rates on necessity entrepreneurs can be extracted from the GEM database to control partly for the ‘refugee’ effect.

(14)

14

Variable Code De scription Source

Dependent variables

Total early-stage Entrepreneurial Activity for Female Working Age Population

FEA Percentage of female 18-64 population who are either a nascent entrepreneur or

owner-manager of a 'new business' (i.e., owning and managing a running business that has paid salaries, wages, or any other payments to the owners for more than three months, but not more than 42 months).

GEM

Total early-stage Entrepreneurial Activity for Male Working Age Population

MEA Percentage of male 18-64 population who are either a nascent entrepreneur or

GEM

Total early-stage Entrepreneurial Activity for Working Age Population

TEA Percentage of 18-64 population who are either a nascent entrepreneur or

GEM

Independent variables

Cost to start a business COST Percentage of income per capita. WBDB

Time required to start a business TIME In days. WBDB

Minimum paid-in capital required to start a business

MINCAP Percentage of income per capita. WBDB

Control variables

Necessity-Driven Entrepreneurial Activity: Relative Prevalence

NECESS Percentage of those involved in TEA who are involved in entrepreneurship because they had no other option for work.

GEM

Natural logarithm of GDP per capita lnGDP In international dollars. GDP per capita based on purchasing power parity

(PPP). PPP GDP is gross domestic product converted to international dollars using purchasing power parity rates.

WBWDI

Table I

Variable s, de scription and source

GEM = Global Ent repreneurship Monit or, WBDB = World Bank dat abase - Doing Business, WBGS = World Bank dat abase - Gender St at ist ics, WBWDI = World Bank dat abase - World Development Indicat ors

(15)

15

Variable Code De scription Source

School enrollment, secondary, female FSECONDARY In gross ratio. Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown.

WBGS

School enrollment, secondary, male MSECONDARY In gross ratio. Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown.

WBGS

School enrollment, secondary, total SECONDARY In gross ratio. Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population of the age group that officially corresponds to the level of education shown.

WBGS

School enrollment, tertiary, male MTERTIARY _{In gross ratio. Gross enrollment ratio is the ratio of total enrollment, regardless}

of age, to the population of the age group that officially corresponds to the level of education shown.

WBGS

School enrollment, tertiary, total TERTIARY _{In gross ratio. Gross enrollment ratio is the ratio of total enrollment, regardless}

of age, to the population of the age group that officially corresponds to the level of education shown.

WBGS

Real interest rate INTEREST _{In percentage. Real interest rate is the lending interest rate adjusted for inflation}

as measured by the GDP deflator. The terms and conditions attached to lending rates differ by country, however, limiting their comparability.

WBWDI

Unemployment, female FUNEMPLOY _{In percentage of female labor force. Unemployment refers to the share of the}

labor force that is without work but available for and seeking employment.

WBWDI

Unemployment, male MUNEMPLOY _{In percentage of male labor force. Unemployment refers to the share of the}

WBWDI

Unemployment, total UNEMPLOY In percentage of total labor force. Unemployment refers to the share of the

WBWDI

GEM = Global Ent repreneurship Monit or, WBDB = World Bank dat abase - Doing Business, WBGS = World Bank dat abase - Gender St at ist ics, WBWDI = World Bank dat abase - World Development Indicat ors

Table I Continue d

(16)

16

b. Estimation Technique

If we would only make use of data for one year, of which the largest sample can be found in the years 2013 and 2014 (as both years include data on 70 countries), the regression could suffer from endogeneity bias in the form of omitted variable bias. As discussed in the

literature review, there are many factors found to have significant effects on entrepreneurship rates. However, many of these factors can be characterized as time-invariant factors, such as a country’s culture and sophistication. In addition, expressing these factors in numerical values can be quite complex. Therefore, such factors cannot be included in our regression model, but their effects will end up in our model’s error term. By using panel data (i.e., cross-country time series data), this problem can be avoided.

When dealing with panel data, one can choose between a fixed- and random effects regression. To choose the correct regression, one can run a Hausman test. What this test basically does, is testing if the ‘between country’ effects are significantly different from the ‘within country’ effects found in the panel data. However, since previous literature already stated that time-invariant factors contribute to entrepreneurs’ decision-making, it is expected that unobservable heterogeneity is the case in the panel data. To confirm this, the output of the Hausman test also points to the fixed effects model being the more appropriate choice (see Appendix, Table 3 and 4, for more detailed results of the test). By doing this type of regression, we can control for observable and, more importantly, unobservable predictors that can affect the estimated model. Hereby, a fixed effects regression removes the effects of these time-invariant characteristics. As Stock and Watson (2003) said: “the key insight is that if the unobserved variable does not change over time, then any changes in the dependent variable must be due to influences other than these fixed characteristics”. By using this method, the focus will be laid on the variations within countries.

Next, the data has been checked for both heteroskedasticity and autocorrelation. The output of both tests can be found in the Appendix, Table 5 and 6. As the Wooldridge test indicates, the data is not characterized by autocorrelation. However, the Cook-Weisberg test does indicate the case of heteroskedasticity in the error term. That is, the variance of the error term is not constant. Therefore, robust standard errors will be used when running the

regressions.

All used independent variables will be estimated (i) independently, (ii) jointly and (iii) including all control variables. This will be done in order to obtain more detailed descriptive statistics.

(17)

17

Two fixed effects regressions will be done4, one on male early-stage entrepreneurship rates (𝑀EA_it) and one on similar rates on female entrepreneurs (𝐹EA_it). The two fixed effects models will look as follows:

𝑀𝐸𝐴𝑖𝑡 = 𝛽1𝐶𝑂𝑆𝑇𝑖𝑡+ 𝛽2𝑇𝐼𝑀𝐸𝑖𝑡 + 𝛽3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡+ 𝛽4𝑁𝐸𝐶𝐸𝑆𝑆 + 𝛽5𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛽6𝑀𝑆𝐸𝐶𝑂𝑁𝐷𝐴𝑅𝑌 + 𝛽7𝑀𝑇𝐸𝑅𝑇𝐼𝐴𝑅𝑌𝑖𝑡+ 𝛽8𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝑖𝑡 + 𝛽9𝑀𝑈𝑁𝐸𝑀𝑃𝐿𝑂𝑌𝑖𝑡+ 𝛼𝑡+ 𝜀𝑖𝑡 𝐹𝐸𝐴𝑖𝑡 = 𝛿1𝐶𝑂𝑆𝑇𝑖𝑡+ 𝛿2𝑇𝐼𝑀𝐸𝑖𝑡+ 𝛿3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡+ 𝛿4𝑁𝐸𝐶𝐸𝑆𝑆 + 𝛿5𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛿6𝐹𝑆𝐸𝐶𝑂𝑁𝐷𝐴𝑅𝑌𝑖𝑡+ 𝛿7𝐹𝑇𝐸𝑅𝑇𝐼𝐴𝑅𝑌𝑖𝑡+ 𝛿8𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝑖𝑡 + 𝛿9𝐹𝑈𝑁𝐸𝑀𝑃𝐿𝑂𝑌𝑖𝑡 + 𝛾𝑡+ 𝜃𝑖𝑡

In which MEA_it and 𝐹𝐸𝐴_𝑖𝑡 correspond to the dependent variables (where 𝑖 = country and 𝑡 = year). The variables COST_it, TIME_it and MINCAP_it are the discussed independent variables pertaining to establishment legislation. The control variables are represented by the fourth until the ninth variable in the equation. The fixed time effects will capture any countrywide trend, as denoted by the year intercepts 𝛼_𝑡 and 𝛾_𝑡. At the end of the equation, the error term can be found ( 𝜀_𝑖𝑡 and 𝜃_𝑖𝑡). See Table I for detailed descriptions of the variables used.

After the regressions have been done, the coefficients will be compared. Hereby, possible differences between the observed male and female estimators will be tested. This can be done by testing if the difference between the values of the ‘male regression’ and ‘female regression’ are significantly different from zero. Hereby, the p-value will show if the null hypotheses can be rejected. If this is the case, a significant gender difference can be found. The hypotheses that will be tested look as follows:

𝐻0: 𝛽1𝐶𝑂𝑆𝑇𝑖𝑡 = 𝛿1𝐶𝑂𝑆𝑇𝑖𝑡 𝐻1: 𝛽1𝐶𝑂𝑆𝑇𝑖𝑡 ≠ 𝛿1𝐶𝑂𝑆𝑇𝑖𝑡

𝐻0: 𝛽2𝑇𝐼𝑀𝐸𝑖𝑡 = 𝛿2𝑇𝐼𝑀𝐸𝑖𝑡 𝐻1: 𝛽2𝑇𝐼𝑀𝐸𝑖𝑡 ≠ 𝛿2𝑇𝐼𝑀𝐸𝑖𝑡

𝐻0: 𝛽3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡 = 𝛿3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡 𝐻1: 𝛽3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡 ≠ 𝛿3𝑀𝐼𝑁𝐶𝐴𝑃𝑖𝑡

A t-test can be used to determine if the coefficients are significantly different and thereby test for a possible effect of gender.

(18)

18

IV. Results

The results will be presented using the fixed regression estimators. Two separate regression analyses have been done on male and female early-stage entrepreneurship rates. The

empirical strategy as outlined in Section III has been followed.

In the next section, the estimated coefficients of the regression on male

entrepreneurship rates will be discussed. This will be done by looking at the five different regression outputs, shown in Table II. Thereafter, the estimators on female nascent

entrepreneurship rates (Table III) will be reviewed. Finally, male and female output

coefficients will be compared to determine if they are significantly different from each other (Table IV).

a. Male estimates

The estimated effects on male early-stage entrepreneurship rates are measured using the three establishment legislation characteristics (for clarification, see Table IV). Firstly, the three independent variables have been regressed independently. The results of these

regressions suggest two significant relationships: Firstly, a 10-percent significant negative effect of the cost required to start a business on early-stage male entrepreneurship rates (-0.096). Secondly, a 5-percent significant negative effect of the minimum paid-in capital

required to start a business-variable (-0.014). To elaborate, a one percent increase in cost

required to start a business decreases male entrepreneurship rates by 0.096. These negative relationships are in line with the expectations drawn from previous literature. Notice that controlling for other time-variant factors leads to the insignificance of all establishment legislations on male entrepreneurship rates.

As we look at the results, adding more variables to the model affects estimates of all tested independent variables. This can be due to the fact that the variables are correlated with a control variable that has been added, which is known as multicollinearity. However,

looking at the correlation matrix (Appendix, Table 10), this does not appear to be the case. Another explanation could be that adding control variables reduces the degrees of freedom without reducing the sum of squares of the error.

The only significant effect found on male entrepreneurship rates is the natural

logarithm of GDP (see Regression 5). Because of the transformation into the log function, the effect of GDP can be stated as follows: a 1% change in GDP per capita is associated with a 0.114 increase in male entrepreneurship rate.

(19)

19

Somewhat awkward findings can be found in the fourth and fifth regression when looking at the Cost variable. Hereby, a non-significant positive effect of the cost required to

start a business on early-stage male entrepreneurship rates can be found. This is not in line

with previous literature and somewhat difficult to explain. Therefore, this finding will be discussed in the Discussion, Section V.

b. Female estimates

Table V shows the results on female early-stage entrepreneurship rates in the 32 countries that are included in the regression. Hereby, the results are shown to be somewhat different. As similar to the estimated effects on male entrepreneurship rates, all independent variables show negative coefficients when regressed independently. However, only one

Variables 1 2 3 4 5

Cost -0.096 * 0.025 0.063

(.086) (.070) (.100)

T ime -0.054 -0.055 * -0.004

(.035) (.032) (.036)

Minimum paid-in capit al -0.014 ** -0.010 -0.001

(.007) (.008) (.017)

Necessit y rat e of ent repreneurship -0.031

(.052)

Nat ural logarit hm of GDP 11.409 **

(5.085)

Secondary enrollment (male) 0.044

(.064)

T ert iary enrollment (male) 0.052

(.147)

Int erest rat e _-0.001

(.071)

Unemployment rat e (male) 0.256

(.187)

Int ercept _12.059 _12.530 _11.588 _12.548 _-111.371

R square 0.827 0.832 0.826 0.833 0.844

Standard errors are in parentheses and in italic

* Significant at the 10-percent level. ** Significant at the 5-percent level. *** Significant at the 1-percent level.

Table II

Estimate s of the e ffe cts of e stablishme nt le gislation on e arly-stage e ntre pre ne urial activity De pe nde nt variable : Total e arly-stage Male Entre pre ne urial Activity (MEA)

(20)

20

variable turned out to be slightly (on a 10-percent level) significant: namely, the minimum

paid-in capital required to start a business-variable. After adding control variables to the

regression model (see Regression 5), no significant linear dependence of the independent variables can be observed.

Compared to the male regression output in Table II, the results in Regression 5 are similar. The GDP variable has been found to be the only significant factor influencing female early-stage entrepreneurship rates: a 1% change in GDP per capita is associated with a 0.099 increase in female entrepreneurship rate.

In addition, similar to findings on male entrepreneurship rates, a positive estimated coefficient of the Cost variable can be observed in Table III, Regression 5.

Variables 1 2 3 4 5

Cost -0.034 0.108 0.127

(.094) (.087) (.102)

T ime -0.043 -0.059 ** -0.017

(.036) (.029) (.031)

Minimum paid-in capit al -0.015 *** -0.016 ** -0.009

(.004) (.007) (.008)

Necessit y rat e of ent repreneurship 0.003

(.036)

Nat ural logarit hm of GDP 9.907 ***

(2.633)

Secondary enrollment (female) 0.016

(.037)

T ert iary enrollment (female) 0.034

(.051)

Int erest rat e 0.007

(.050)

Unemployment rat e (female) 0.274

(.180)

Int ercept 7.475 8.171 7.501 8.013 -98.853

R square 0.838 0.845 0.842 0.849 0.880

Table III

Estimate s of the e ffe cts of e stablishme nt le gislation on e arly-stage e ntre pre ne urial activity De pe nde nt variable : Total e arly-stage Fe male Entre pre ne urial Activity (FEA)

(21)

21

Now that the estimated coefficients of the independent variables on both male and female entrepreneurship rates have been reviewed, the first hypothesis can be rejected (‘The

variable ‘minimum paid-in capital required to start a business’ has a significant negative effect on early-stage entrepreneurship rates, both for men and women’.). As mentioned, all

coefficients of the independent variables have been found to be of insignificant effect on both male and female entrepreneurship rates.

c. Differences

As the coefficients of the establishment legislation variables on male- and female early-stage entrepreneurship rates have been discussed, these results will now be compared. Hereby, the second and third hypothesis will be tested. The results of the t-test, which is used to test for significant differences between the estimated

coefficients (see Regression 5 in Table II and Table III), can be found in Table IV.

As none of the estimated coefficients of the independent variables have been found to be significant, no comparison can be made in terms of smaller/larger effects. Therefore, the second hypothesis (‘Establishment legislations have a greater effect on

female nascent entrepreneurship rates in comparison to male nascent entrepreneurship rates’.) will be rejected.

Given the results of the t-test, the only

independent variable that has a (at a 10-percent level) significantly different coefficient, is the variable cost

required to start a business. Therefore, one could argue to have found evidence to not reject

the third hypothesis: Out of the three, the biggest gender gap is found in the explanatory

variable ‘cost associated with starting up a business’. However, due to the positive estimated

coefficient of this variable of interest, these findings are not in line with the rationale behind this hypothesis. chi2( 1) = 0.70 Prob > chi2 = 0.4030 (1) [FEA]CAPIT - [MEA]CAPIT = 0 chi2( 1) = 1.10 Prob > chi2 = 0.2944 Table IV T-te st output (1) [FEA]COST - [MEA]COST = 0 chi2( 1) = 3.04 Prob > chi2 = 0.0814*

(22)

22

V. Discussion

The results of this thesis show no significant effects of establishment regulations on early-stage entrepreneurship rates. Examining the differences between male and female entrepreneurship rates did find some interesting results confirming expectations, but most results turned out to be insignificant. In addition, some unanticipated results have been found.

The biggest issue concerning this research was the size of data available. Ideally, we would have wanted to use complete datasets on as many countries possible. However, the availability of data on this subject was quite restricted. Firstly, the World Databank was the only publically available database that provided numerical indicators of business regulations on a country-level basis. Because this database only goes back to 2005, the panel data was already quite restricted. In addition, there was not one database that could provide complete datasets on the control variables too. Lastly, the Global Entrepreneurship Monitor was one of the few databases including gender specific entrepreneurship rates, in particular those on early-stage entrepreneurs. Due to the fact that they did not collect data on the same sample of countries each year, a selection of countries had to be made.

In addition, by choosing the fixed effects regression method to analyse the data, all countries with data on only one year had to be dropped. This is due to the fact that this method requires datasets including at least two observations per country. The choice had also been made to drop all countries with less than 8 years (in the range of years 2005 until 2014) of data available. This substantially decreases the number of observations and, therefore, the power of the estimates.

Next to the data size, the results also imply unobserved omitted variable bias. This is due to the fact that after adding control variables, the 𝑅2 of the regressions increases with relatively small portions. Because of the use of a fixed-effects model, there has only been controlled for time-invariant characteristics. Entrepreneurship rates are influenced by many observable and unobservable factors. Therefore, it is most likely that the dataset does not include all time-variant factors influencing the dependent variable of the model. Furthermore, there are a lot of possible influences that are difficult to obtain but do influence the chance for someone to become an entrepreneur. However, the number of factors contributing to the rate of entrepreneurship is extensive, and therefore, a selection will always have to be made.

Another concern is the used fixed effects method to analyse the data. Although this method solves the problem of omitted variable bias to a large extent, it also adjusts for between-countries differences. Therefore, it reduces the variation in the variables of interest,

(23)

23

which makes it less likely to get significant estimates of the relationships examined. Furthermore, this procedure maximizes the error-in-the-variables bias.

The last concern that should be accounted for, is the estimated coefficient of the cost

required to start a business on both male and female early-stage entrepreneurship rates. As

the coefficients are not statistically significant, no conclusions can be drawn from these results. However, notice that the effect seems to be positive. This is an awkward finding which is hard to explain. Therefore, this relationship should be examined more extensively.

(24)

24

VI. Conclusion

This thesis investigated the effects of establishment legislations on early-stage entrepreneurship rates. In contrast with previous research on the topic, the effect has been estimated on both male and female entrepreneurship rates. Thereby, testing on differences between these models has provided new insights on the topics gender and entrepreneurship.

The main conclusion of this thesis is that no significant effects have been found of the three different types of establishment legislations (cost-, time- and minimum paid-in capital

required to start a business) on male and female early-stage entrepreneurship rates. The

results are based on the empirical model, which controls for country fixed effects. Because of the insignificance of the t-values, it is not appropriate to draw any conclusions about the estimated effect of establishment legislations on entrepreneurship rates. However, when comparing the two model estimates, a significant difference has been found between the estimates of the cost required to start a business on male and female entrepreneurship rates.

Overall, although most of the results were insignificant, these findings did add a new dimension to the field of research done on start-up establishment regulations. By using a combined dataset that consisted of data on 32 countries over 8-10 years, the findings did provide credible estimates on the matter. Therefore, the results of this thesis suggest that governments and policymakers should focus on gender-neutral establishment legislations. Thereby, alteration of these regulations should not be of considerable impact on early-stage entrepreneurship rates.

For future research, there are some aspects of this thesis that should be examined more thoroughly. Firstly, the insignificant, but positive effect of the cost required to start a

business on entrepreneurship rates is yet to be explained. Therefore, future research should

examine this relationship more thoroughly. Secondly, when available, larger datasets should be used in order to investigate rates on more countries and/or years. Hereby, it could be interesting to examine different clusters of countries. Thirdly, new research should be done that tests the effects of a broader scope of governmental policies on nascent entrepreneurship. For example, this could be done by adding variables indicating the level of difficulty of getting credit and/or the amount of training available for early-stage entrepreneurs.

(25)

25

References

Acs, Z. J., Desai, S., & Hessels, J. (2008). Entrepreneurship, economic development and institutions. Small Business Economics, 31(3), 219-234.

Ahmad, N., & Seymour, R. (2008). Defining entrepreneurial activity: Definitions supporting frameworks for data collection.

Arenius, P., & Minniti, M. (2005). Perceptual Variables and Nascent Entrepreneurship. Small

Business Economics, 24(3), 233-247.

Audretsch, D., Carree, M., & Thurik, R. (2001). Does entrepreneurship reduce

unemployment? (No. TI 01-074/3).

Audretsch, D. B., Grilo, I., & Thurik, A. R. (2007). Explaining entrepreneurship and the role of policy: a framework. The handbook of research on entrepreneurship policy, 1-17.

Baughn, C. C., Chua, B. L., & Neupert, K. E. (2006). The normative context for women's participation in entrepreneurship: A multicountry study. Entrepreneurship Theory and

Practice, 30(5), 687-708.

Borghans, L., Heckman, J. J., Golsteyn, B. H., & Meijers, H. (2009). Gender differences in risk aversion and ambiguity aversion. Journal of the European Economic Association,

7(2-3), 649-658.

Block, J. H., Hoogerheide, L., & Thurik, R. (2013). Education and entrepreneurial choice: An instrumental variables analysis. International Small Business Journal, 31(1), 23-33.

Brana, S. (2013). Microcredit: an answer to the gender problem in funding?. Small Business

Economics, 40(1), 87-100.

Doing Business World Bank (2016). About Doing Business. Retrieved from http://www.doingbusiness.org/about-us

GEM Global Entrepreneurship Monitor. (n.d.). Retrieved from http://www.gemconsortium.org/wiki/1150

Gennaioli, N., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2013). Human capital and regional development. The Quarterly Journal of Economics, 128(1), 105-164.

Kautonen, T., Down, S., & Minniti, M. (2014). Ageing and entrepreneurial preferences.

Small Business Economics, 42(3), 579-594.

Koellinger, P., Minniti, M., & Schade, C. (2007). “I think I can, I think I can”:

Overconfidence and entrepreneurial behavior. Journal of economic psychology, 28(4), 502-527.

(26)

26

Kwong, C., Jones-Evans, D., & Thompson, P. (2012). Differences in perceptions of access to finance between potential male and female entrepreneurs: Evidence from the UK.

International Journal of Entrepreneurial Behavior & Research, 18(1), 75-97.

Langowitz, N., & Minniti, M. (2007). The entrepreneurial propensity of women.

Entrepreneurship theory and practice, 31(3), 341-364.

Levesque, M., & Minniti, M. (2006). The effect of aging on entrepreneurial behavior.

Journal of Business Venturing, 21(2), 177-194.

Millan, J. M., Congregado, E., Roman, C., Van Praag, M., & van Stel, A. (2014). The value of an educated population for an individual's entrepreneurship success. Journal of

business venturing, 29(5), 612-632.

Mohamad, N., Lim, H. E., Yusof, N., Kassim, M., & Abdullah, H. (2014). Estimating the choice of entrepreneurship as a career: The case of Universiti Utara

Malaysia. International Journal of Business and Society, 15(1), 65.

Mueller, S. L. (2004). Gender gaps in potential for entrepreneurship across countries and cultures. Journal of developmental entrepreneurship, 9(3), 199.

OECD (2016), Entrepreneurship at a Glance 2016, OECD Publishing, Paris. http://dx.doi.org/10.1787/entrepreneur_aag-2016-en

Parker, S. C., & Van Praag, C. M. (2006). Schooling, capital constraints, and entrepreneurial performance: The endogenous triangle. Journal of Business and Economic Statistics,

24(4), 416-431.

Robichaud, Y., LeBrasseur, R., & Nagarajan, K. V. (2010). Necessity and opportunity-driven entrepreneurs in Canada: An investigation into their characteristics and an appraisal of the role of gender. The Journal of Applied Business and Economics,

11(1), 59.

Stock, J. H., & Watson, M. W. (2003). Introduction to econometrics (Vol. 104). Boston: Addison Wesley.

Van der Zwan, P., Verheul, I., & Thurik, A. R. (2012). The entrepreneurial ladder, gender, and regional development. Small Business Economics, 39(3), 627-643.

Van Stel, A., Storey, D. J., & Thurik, A. R. (2007). The effect of business regulations on nascent and young business entrepreneurship. Small Business Economics, 28(2-3), 171-186.

Verheul, I., Wennekers, S., Audretsch, D., & Thurik, R. (2001). An eclectic theory of

(27)

27

Wennekers, S., Van Wennekers, A., Thurik, R., & Reynolds, P. (2005). Nascent

entrepreneurship and the level of economic development. Small Business Economics,

(28)

28

Appendix

a. Descriptive Statistics

In this section, it will shortly be explained how the data was modified. Relevant data has been found on 596 observations, consisting of 107 countries. However, because the GEM does not gather data on the same lists of countries over the past 17 years, the dataset is strongly

unbalanced. In addition, also the World Bank database does not provide complete datasets on the different independent and control variables either.

- Missing data

Firstly, a selection has been made of those countries with complete datasets on at least eight out of ten years. As some of the control variables are not available for the years 2001-2004 and 2015-2016, a selection has been made of data available on the range of years from 2005 until 2014. The countries that are included in this regression are those of which data was available on at least eight out of ten years. (297 observations deleted)

- Independent variable selection

Thereafter, a selection has been made on the available establishment legislation-variables. Based on the high correlation found between variables TIME and PROCED, the indicator

procedures required to start a business will not be used. (dropped variable PROCED)

Table 1

Correlation matrix independent variables

COST TIME PROCED CAPIT

COST 1

TIME 0,27 1

PROCED 0,49 0,65 1

CAPIT 0,19 0,05 0,28 1

Table 2

Descriptives independent variables

Mean Std. Dev Min. Max.

COST 7,68 8,32 0 46

TIME 22,42 25,00 4 156

PROCED 7,24 3,23 2 17

(29)

29

b. Graphs and tables

Graph 1

Female Entrepreneurial Activity (FEA) plotted over time

Graph 2

(30)

30

Graph 3

Female Entrepreneurial Activity (FEA), mean and variance

Graph 4

(31)

(32)

(33)

(34)

(35)

(36)

(37)

37 Variables 1 2 3 4 5 Cost -0.063 0.069 0.098 (.088) (.076) (.094) Time -0.048 -0.057 * -0.031 (.035) (.030) (.033)

Minimum paid-in capital -0.014 ** -0.013 0.011

(.006) (.008) (.012)

Necessity rate of entrepreneurship 0.006

(.043)

Natural logarithm of GDP 9.433 **

(3.462)

Secondary enrollment (total) -0.014

(.060)

Tertiary enrollment (total) 0.013

(.089)

Interest rate -0.009

(.060)

Unemployment rate (total) 0.227

(.176)

Intercept 9.735 10.325 9.524 10.248 -85.917

R square 0.841 0.847 0.842 0.849 0.856

Table 11

Estimates of the effects of establishment legislation on early-stage entrepreneurial activity Dependent variable: Total early-stage Total Entrepreneurial Activity (TEA)

The estimated effect of business regulations on nascent entrepreneurship : what is the influence of establishment legislations on the rates of male and female early-stage entrepreneurs?