Barriers to Entrepreneurship

Institutions and Female Entrepreneurship*

Saul Estrina and Tomasz Mickiewiczb

Abstract: This paper compares the impact of institutions on individual decisions to become entrepreneurs in the form of new business start-ups by males and females across 55 developed and developing economies between 2001 and 2006. We hypothesise that women are less likely to undertake entrepreneurial activity in countries where the rule of law is weaker and where the state sector is larger. We use data from the Global Entrepreneurship Monitor survey (GEM) which covers at least 2,000 individuals annually in each of up to 55 countries, merged with country-level data, from the World Bank, Economic Intelligence Unit, Polity IV and Heritage Foundation. We find that women are less likely to undertake entrepreneurial activity in countries where the state sector is larger, but the rule of law is not generally found to have gender-specific effects. However, more detailed institutional components of discrimination against women, namely restrictions on freedom of movement away from home and lack of protection from violence against women, make female high aspiration entrepreneurship less likely, even if broadly defined entrepreneurial entry including self-employment is not affected. Keywords: female entrepreneurship, state sector, rule of law JEL classification codes: L26, J16, H11, O16, O17, P43

This version: September 11th 2010

* The authors are indebted to Paul Reynolds and Stephen Hunt for discussion of data and methodology. The paper has benefited from comments by Harry Barkema, Susan Hill, Fei Qin, Virginie Perotin, Paul Reynolds, Ute Stephan and the participants of the World Bank and University of Michigan conference on Female Entrepreneurship, June 2009. Any remaining errors are our own. a

Corresponding author: Department of Management, London School of Economics, Houghton St, London WC2A 2AE, UK; email: [email protected] b

SSEES, University College London; email: [email protected]

1

1. Introduction

Early work on entrepreneurship (e.g. Schumpeter, 1934; Kirzner, 1973) implicitly assumed that most entrepreneurs would be men, perhaps because most business creation at that time was a male preserve. However, the situation has changed markedly, with women now owning more than 40% of the privately held firms in the US (Brush, 2006). The phenomenon of female entrepreneurship is also increasingly global, with a recent Global Entrepreneurship Monitor (GEM) study estimating that companies owned by women comprise between 25% and 33% of formal sector business around the world (Minniti, Arenius and Langowitz, 2005a). These developments have led to the emergence of a large literature on female entrepreneurship, analysing for example personal attributes, access to finance and the role of social networks (see Greene et al, 2003). While male and female entrepreneurs are similar in terms of motivation, age and education, significant differences have been identified. Of these, the most important for female entrepreneurs concern the sectors chosen for new firm creation (service and retail rather than manufacturing (Hisrich and Brush, 1984)); access to resources (Verheul and Thurik, 2001)); and the use of less varied social networks (see Brush (2006) for a survey). In this paper, we do not consider the similarities and differences between male and female entrepreneurs as restricted to a particular institutional context. Rather, we seek to explain the cross-country variation in male/female entrepreneurship rates. While there is a growing body of work at the country level related to gender dimension of entrepreneurship, surprisingly little has yet been attempted to enhance our understanding of the impact of

2

alternative national business and cultural environments on female entrepreneurship 1. We develop hypotheses, to explain how institutional variation might impact on entrepreneurship, but differently on male and female choices to become entrepreneurs. In particular we hypothesise that women are less likely than men to undertake entrepreneurial activity in countries where the rule of law is weaker and the state sector is larger. We further consider how institutions affect the probabilities of men and women to aspire to higher growth projects, and analyse in detail the differential impact of gender specific aspects of the rule of law on male and female entrepreneurship and on higher aspiration growth projects. Our analysis is made possible by access to the GEM database, which covers 55 developed and developing economies between 2001 and 2006 and which includes all startups regardless of the legal form. The Global Enterprise Monitor survey (GEM) covers at least 2,000 individuals annually in each country. Our individual level data (475,000 usable observations) are generated through surveys creating stratified samples drawn from the whole working age population in each participating country. An empirical approach of this paper relies on merging of individual and country-specific institutional data sets, notably the World Bank, Economic Intelligence Unit, Polity IV, and Heritage Foundation data. Existing studies on entrepreneurship tend to focus either on self-employment, small enterprises requiring legal registration or on incorporated firms as a proxy for entrepreneurial activity. We are able to use data on the whole universe of potential entrepreneurs, rather than just of the existing business owners, and hence analyse the choice of entrepreneurial activity. This is particularly beneficial when analysing the different choices of male and female entrepreneurs. As suggested by Berglann et al. (2009), cross-gender differences are probably better captured by a wider definition of entrepreneurship.

1

A little more has been done on the related theme of female employment. In particular, see Giavazzi et al. (2009), Algan and Cahuc (2007).

3

Building on Aidis et al. (2010), we focus on two key institutional dimensions for entrepreneurial activity: the rule of law / property rights, and the size of the state sector. Utilising Williamson’s (2000) theoretical framework, we can view the first dimension as being at the constitutional level of institutions, while the second is located at the level of governance and proxies for the extent of government intervention in the economy. We find that a smaller state sector positively affects entrepreneurship rates for all, and this effect is significantly greater for females than for males. However, contrary to our hypotheses, we are unable to discern a significant differential impact of rule of law on entrepreneurial activity for women. We consider several reasons for this result, including that the rule of law needs to be disaggregated to reveal the impact of discrimination against women. We interpret our findings in terms of gender specific incentives and opportunities within a typical household context, as augmented by alternative institutions. The gender approach to economic issues can be justified on three different (and often complementary) grounds: equality, merit / efficiency, and need (Jacquette, 1990; Razavi, 1997). The starting point for the first approach is a moral and philosophical assumption about equality of rights between men and women. The second relies on the economic argument that any form of discrimination is economically inefficient. The third points out that women poverty and related issue of children poverty is an important social problem that calls for institutional reform restoring equilibrium between male and female economic positions. From all three perspectives, entrepreneurship appears as critical: the adequate role of women in new venture creation, ownership and growth may lead to their empowerment and may induce structural change that goes hand in hand with institutional reform. It is the broader objective of this paper to contribute to that agenda. The remainder of the paper is organised as follows. In Section 2 we provide a brief overview of the literature about female entrepreneurship and institutions and present our

4

hypotheses. The dataset, methodology and our approach to quantifying the institutional environment is outlined in the third section and the fourth reports our results. We present our conclusions and interpretations in Section 5.

2. Entrepreneurship, Female Entrepreneurship and Institutions

This section begins by considering the institutional literature and discussing how institutions may influence entrepreneurship in general, and female entrepreneurship in particular. We go on to present the hypotheses to be tested in the remainder of the paper.

2.1 Institutional theory According to North (1990), organizations set up by entrepreneurs adapt their activities and strategies moulding them to fit the opportunities and limitations provided through the formal and informal institutional framework. Though ideally, formal rules are designed to facilitate exchange reducing transaction costs (Williamson, 1985; 2000), they are also likely to affect individuals or groups in different ways. Since private interests may differ and individuals, who often have their narrow interests at heart, affect formal rules and institutions: the latter are not necessarily shaped in the interest of social well-being (North 1994; Olson, 2000). Where social groups are easily identified, as it is in the case of gender, opportunities for discrimination arise. Correspondingly, institutions can be maintained even if they are inefficient (DiMaggio & Powell 1983; North 1990). Formal institutions have a tenacious ability to survive because they become intertwined with habitual behaviour (i.e. culture) providing stability (Crawford and Ostrom, 1995). Institutions change infrequently due to the influence of path dependence (North 1990:10). In addition, lock-in can occur as a result of a symbiotic relationship between existing institutions and the organizations that

5

have evolved as a result of the incentive structure provided by those institutions (ibid. 1990:7; Dixit, 2004). Even when formal rules change, organizations which benefited from the outdated informal rules and which would lose their ‘comparative advantage’ if they adopted new informal practices complementary to formal rule changes will continue to participate in detrimental informal rule practices in order to retain their advantage. North and Thomas (1973), Williamson (1985, 2000), Barzel (1997), Rodrik (2000), Acemoglu and Johnson (2005), Aidis et al. (2010) and others have argued that the rule of law and its economic component, the property rights system are constitutional level institutions that form the backbone of the market economy. In essence, strong rule of law encompasses the dimension of economic property rights that are defined as “individual ability, in expected terms, to consume the good (or the services of the asset) directly or to consume it indirectly through exchange” (Barzel, 1997, p.3). Accordingly, in recent institutional research, the focus shifts from the assignment of rights and certification to the institutional environmental conditions that make execution of these rights, especially exchange and other legal contracts based on the property rights, effective. This has implications for the financing of projects; a well functioning property rights system creates the basis for low transaction cost financial contracts and a virtuous circle of the creation of assets and finance (De Soto, 2001). One important issue relates to the accessibility of these rights to the population as a whole, which may have critical implications for the extent and performance of the entrepreneurial sector (Sonin, 2003; De Soto, 2001). Moreover, in many countries around the world, universality of property rights is undermined by gender specific restrictions related to ownership rights, free transferability of assets, or even violations of basic personal rights like the freedom of movement or protection from violence (Jütting et al. 2006). These restrict the exercise of economic property rights and economic opportunities, including entrepreneurship, by women.

6

2.2. Institutions and Entrepreneurship Baumol (1990) argued that entrepreneurial activity could take productive, nonproductive and even destructive forms, depending on the institutional context. Institutions and the associated incentives and penalties for particular types of economic behaviour determine the balance between these three forms, with higher quality institutions motivating entrepreneurs to choose productive over value-reducing activities. This framework partially motivates our analysis. We commence with potential entrepreneurs who can be viewed as maximizing their expected return when making a decision to start new ventures (Casson, 1982). In contexts where institutions are functioning effectively, risks are primarily limited to factors associated with the nature of the ventures themselves and the characteristics of the individuals’ involved (Schumpeter, 1934; Kirzner, 1973). In a developing economy context, weaker institutions may increase net returns to non-productive or even criminal activity, reducing the incentives to enter (productive) entrepreneurship. Two specific aspects of the inter-relationship between entrepreneurial activity and institutions have been found to be among the most important in the literature (Johnson et al., 2002; Dermirguc-Kunt et al., 2006; Klapper et al., 2006; Aidis et al., 2010). The first is the rule of law / support for property rights. The second concerns the welfare and tax system, which influence both the opportunity cost and the net financial return to entrepreneurial activity (Aidis et al., 2010). A weak rule of law increases the transaction costs of entrepreneurship as well as the riskiness of entrepreneurial activity. Recent theories of entrepreneurship emphasise that “the institution of private property ... has an important psychological dimension that enhances our feelings of ... internal control and personal agency, and it thereby promotes entrepreneurial alertness” (Harper 2003, p. 74). The protection of property rights and the rule of law enable

7

the development of financial institutions, which influence entrepreneurs’ ability to access capital and therefore, for example, to start their businesses at a cost-minimising scale or to survive short-term negative shocks. De Soto (2001) argues that the lack of well-defined and efficient system of registering, protecting and trading of property rights may be the key obstacle that prevents the entrepreneurs from utilising and combining the potentially productive assets and turning them into real capital (see also Rodrik, 2000). Acemoglu and Johnson (2005) emphasise the critical role for the functioning of a market economy of constraints on the executive branch of the government, and this defines both elements of protection and stability of property rights. In their approach, that we adopt here, property rights are akin to the related, but wider concept of the rule of law, as the latter corresponds to a stable institutional framework restraining the erratic use of power by politicians and public administrators. The relationship between property rights / rule of law and entrepreneurship has been analysed extensively, but the results have been contradictory. Thus the property rights indicator was not found to be significant by Klapper et al. (2006), Desai et al. (2003) or Demirguc-Kunt et al. (2006), possibly because of their focus on incorporated companies, but, Johnson et al. (2002) suggest that the insecurity of property rights may be a key factor deterring investment in small manufacturing firms. There is also a literature about how the size of the state sector affects entrepreneurship; a larger state sector is argued to militate against entrepreneurial activity. Taxes and welfare provision may affect entrepreneurial entry via their direct impact on expected returns to entrepreneurial activity and its opportunity cost. High and increasing marginal level of taxes may weaken incentives for opportunity-driven entrepreneurship by reducing potential gains, while high levels of welfare support provide alternative sources of income and therefore by increasing the alternative wage may reduce the net expected return

8

to entrepreneurship. Taken together, this indicates that a larger state sector will crowd out entrepreneurial activity; a hypothesis confirmed in Aidis et al. (2000).

2.3 Institutions and Female Entrepreneurship

The literature, typically drawing on data from a single country, has shown that education and birth order have the same impact on female as male entrepreneurs (Schwartz, 1976; Hisrich and Brush, 1984; Greene et al., 2003), though entrepreneurial age profiles may differ between genders (Berglann et al., 2009). In addition, some authors have identified significant differences with respect to access to resources. Thus, though Verheul and Thurik (2001) and Coleman (2000) found no differences between the proportions of debt and equity used by male as against female entrepreneurs, Carter and Rosa (1998) identified gender differences in financing entrepreneurial activity. Thus, while men used more capital at start up, women were less likely to use formal financial instruments like bank loans or supplier credit. Bruce (1999) indicated there might be an important role for informal finance in female entrepreneurial activity by establishing that the presence of an employed husband led to inter-family capital transfers, helping women to become self-employed (see also Verheul et al., 2006; Greene et al., 2003). However, there is also considerable cross-country variation in the proportion of females undertaking entrepreneurial activity, though female entrepreneurs are everywhere in the minority. Thus Minniti et al. (2005a) report variation of female total entrepreneurial activity rate between 1.2% (Japan) and 39% (Peru). Even so, there has been little consideration of the role of institutions on female entrepreneurship with the notable exception of Verheul, van Stel and Thurik (2006) who explain differences in female and male entrepreneurship using a sample of 29 countries. One might argue that men tend to have stronger advantages in business entry in countries where property rights are weak and

9

institutional barriers to entrepreneurship are high. In such a situation, entrepreneurship depends especially on high levels of motivation, and, according to Shragg et al. (1992), negative self-perceptions in this context, are an important difference between male and female entrepreneurs. Moreover, weak property rights mean that entrepreneurs have to rely relatively more on informal networks and those networks tend to be male dominated (as exemplified by Russia (Aidis et al., 2008)); due to gender-defined social positioning, men may also be more effective dealing with government officials (Sabarwal et al., 2009). Brush (2006) concludes that male and female business networks are significantly different and that female entrepreneurs engage in less “weak-tie” networking than salaried males. In addition, the rule of law may have a differential positive impact on female entrepreneurship because it is closely correlated with eliminating (or at least alleviating) institutional traits that restrict the rights of women, including their economic rights. Those restrictions may include constraints imposed on female mobility (including leaving home), marriage decisions, inadequate protection of their physical integrity (that make it difficult for them to operate independently in a wider anonymous community), restrictions of their civil liberties and direct restrictions on their property rights (Jütting et al., 2006). We therefore predict that:

Hypothesis 1: Women are less likely than men to undertake entrepreneurial activity in countries where the rule of law is weaker.

We have argued that a larger state sector can limit the incentives for entrepreneurial entry both via incentive and opportunity cost factors (Verrheul et al., 2001; 2006). In addition, greater employment in the state sector may create alternative options for potential female entrepreneurs.

10

The size of the state also seems likely to have a relatively greater negative impact on women than men as entrepreneurs. This is because, consistent with the role investment theory (England, 1984; Bielby and Bielby, 1988; Lobel, 1991, Orser et al., 2006), women’s occupational decisions are often made within the social context of a household and it is in many countries a traditional pattern is that women’s’ activity rates are lower than men’s due to the additional burdens and responsibilities associated with domestic and caring labour. As a result, women’s decisions to enter either into employment or into entrepreneurship will be more sensitive to contextual factors because the perceived opportunity cost is higher: men may enter self-employment or employment first to satisfy income needs of a given household; a pecking order of entrepreneurial entry and entry into employment may become established. In this context, a large state sector both de-motivates women by offering security and by eliminating awards. The mechanisms may partly arise because if the state sector is small, the resulting weakness of social security may create incentives for women to become economically active 2. Moreover, when the state provides more services that affect household/families, the rewards from both employment and entrepreneurship are lower (say in terms of access to housing, education for children or healthcare). These factors may affect women more than men because of the tendency within the household to develop a genderspecific pecking order of entry into entrepreneurship and employment. In addition, as mentioned above, a larger state sector implies that markets are replaced by hierarchies and, due to socially defined gender positioning men may be in more advantageous positions while dealing with them (Saberwal et al., 2009). In contrast, markets, provided they remain competitive, will alleviate discrimination (Becker, 1971). 3 Therefore women’s economic

2

Childcare is an interesting special case to which we will return later on. Given the social context sketched above, lack of childcare provision could induce women to stay at home. Short-term effects like this appeared in 1990s in the ex-Soviet block. However, in equilibrium, state-sponsored childcare will also affect decisions to have children, so the net long-term effect on female activity rates is not obvious a priori. 3 This discussion concerns not the size of the state sector but government-sponsored legislation that may combat discrimination. The latter is associated with the rule of law, which implies universality and lack of

11

decisions may be more sensitive to the size of the state sector than men’s and we hypothesise that:

Hypothesis 2: Women are less likely than men to undertake entrepreneurial activity in countries where the state sector is larger.

3. Data and Methods In this section we consider the data used in our empirical work, including the variables describing the institutional environment (see also: Aidis et al., 2010). We go on to describe our estimation techniques.

3.1. Indicators Corresponding to Institutions and Macro Level Controls We utilise a set of various country-level measures of institutional environment and macroeconomic characteristics along with individual-level variables. Concerns about simultaneity bias are ameliorated in our use of aggregate explanatory variables because the individual decision of a potential entrepreneur should not affect country-level institutions or economic development. However, endogeneity may still arise because the mean countrylevel individual entry outcome may affect some of the macro variables, for instance GDP growth rate. We attempt to address this issue by lagging all our macroeconomics and institutional variables by one year. There is no universally accepted set of measures for our indicators of institutional quality. Scholars have largely relied on what is commonly referred to as institutional outcome variables (Glaeser et al. 2004) which include survey indicators provided by the International Country Risk Guide, the World Bank measures of Governance Effectiveness; discrimination, as previously discussed. However, there may be functional complementarity between the two: a larger state sector replaces competition with hierarchies and makes special legislation combating discrimination more urgent.

12

the World Bank’s Doing Business indicators; Polity IV measures of political institutions, Fraser Institute, and the Heritage Foundation / Wall Street Journal indices. In selecting variables, care should be taken to avoid multicollinearity, which may both render some coefficient insignificant and lead others to appear significant because of over-specification. Commencing with the size of the government, we use the Heritage Foundation’s variable, ‘Government Size’, which is simply based on the quadratic transformation of the ratio of government expense to GDP, with lower scores reflecting a larger government (see Beach and Kane 2008) 4. The Heritage Foundation / Wall Street Journal index also has wide country and time period coverage for its property rights indicator and this has been used by other scholars (e.g. Acemoglu and Johnson 2005; Aidis et al. 2010). However we face two problems here. First, the correlations with some other key dimensions are high (see Table 1 below). Second and even more importantly, the Heritage measure merges the two dimensions of property rights we discussed above, namely fundamental protection from arbitrary government and protection of private contracts. Following Acemoglu and Johnson (2005), we consider the former to be more important and therefore use the Polity IV measure of efficient constraints on the arbitrary power of the executive branch of the government as the measure of constitutional level indicator of rule of law and security of property. As an extension to our main results, we also substituted our two key institutional measures with the particular subcomponents that are likely to have a more specific impact on women. We replaced our general measure of rule of law by two more specific indicators: restrictions on women freedom of movement away from home, restrictions on movement, and inadequate protection combating violence against women, violence against women. Both of these indicators are reported by the OECD Development Centre. In turn, two more 4

The ratio of government expense to GDP is also reported by World Bank ‘World Development Indicators’, but Heritage data have the advantage of wider country and time period coverage.

13

specific measures of welfare that substitute for our general indicator of the size of the state are: maternity leave, a composite indicator that assesses length of maternity leave and benefits coverage. The second indicator, childcare, relates to availability, affordability and quality of childcare services, as well as the role of the extended family in providing childcare. Both are compiled by the Economist Intelligence Unit (EIU and WB, 2009). In addition to the institutional variables discussed above, we follow the literature in controlling for indicators of economic development and various personal characteristics of entrepreneurs that might affect entrepreneurial entry. Several studies have documented the existence of a relationship between entrepreneurial activity and economic development (Wennekers et al. 2005; Carree et al. 2002; Acs et al. 1994). We control by including a measure of per capita GDP at purchasing power parity (PPP). The literature has also been concerned with the link between the overall (cyclical) economic performance in a country and the incentives to entrepreneurial entry. Two conflicting effects may occur, and it is difficult to decide a priori which has the stronger impact. On the one hand, entrepreneurship may be ‘recession-push’, as the opportunity cost of entrepreneurial entry is lower when existing firms are not expanding, which reduces new job openings. On the other hand, there may be also ‘prosperity-pull’ effect; that is, a growing economy leads to larger expected gains from entrepreneurial activity (Parker, 2004). The rate of economic growth measured by GDP growth has been shown by Van Stel et al. (2007) to have a positive effect on the rate of opportunity entrepreneurship. In our models we include GDP growth as a control variable. We need to verify that the correlation between our independent institutional and macroeconomic variables in the model does not become excessive. For example, the Heritage Foundation measure of property rights is highly correlated with other institutional dimensions, as is the availability of finance (credit /GDP). Since the use of the latter in our regressions led to unstable results, and because as argued above the availability of finance is

14

strongly conditioned by the rule of law, we instead use only constraints on the executive in our equations. Table 1 below reports the simple correlation coefficients for the indicators we considered and to illustrate our choice of variables. Table 2 provides the definitions of variables and summary statistics.

{Tables 1 and 2}

3.2 Individual Level Data Our individual level data are drawn from GEM and are generated through surveys creating stratified samples of at least 2,000 individuals per country. The sample is drawn from the whole working age population in each participating country and therefore captures both entrepreneurs and non-entrepreneurs. While data on established businesses ownership is included, entrepreneurial activity is primarily viewed as new, nascent start-up activity. Nascent entrepreneurs are defined as those individuals between the ages of 18 – 64 years who have taken some real action toward creating a new business in the past year, and expect to own a share of those businesses they are starting, which must not have paid any wages or salaries for more than three months (Minniti et al., 2005b). In contrast, established entrepreneurs are defined as individuals who own or manage a company and have paid wages or salaries for more than 42 months (ibid.). Young businesses are located between these two categories (already not nascent, but younger than 42 months). We utilise all available GEM data from the 1998-2006 surveys including the 55 individual countries.

15

3.3 Gender and institutions: interactive effects Most research indicates that men have a higher probability of becoming entrepreneurs than women (Minniti et al. 2005a; Verheul et al. 2006); in an analysis based on GEM data, men were found to be about twice as likely to be involved in entrepreneurial activities than women (Reynolds et al. 2002). This we take as granted and to test our hypotheses we focus on interactive effects between female dummy and our institutional dimensions corresponding to the hypotheses as described above. Accordingly, we consider the interaction between the female dummy variable and the country-level variables; rule of law and the size of the state sector. In additional specifications we explore the effect of substituting each of those two dimensions in turn using more specific measures related first to women’s rights and next to female-specific welfare measures.

3.4 Control variables for individual characteristics The most common controls for personal characteristics in the literature are for age and human capital. The relationship between entrepreneurship and age is typically found to be inverse-U -shaped, with the maximum found at a relatively young age (Levesque and Minniti, 2006) and we use a quadratic control in our equations. Robinson and Sexton (1994) and Cooper and Dunkelberg (1987) find that the decision to become self-employed is influenced by education while the results of Delmar and Davidsson (2000) and Davidsson and Honig (2003) show an education effect for nascent entrepreneurs. More recent evidence compiled by Parker (2004) suggests that on average, entrepreneurs tend to be more educated than non-entrepreneurs and Brush (2006) confirms that these findings hold for female as well as male entrepreneurs. Wennekers et al. (2005) found a significant and positive relationship between the number of incumbent business owners and entrepreneurial start-ups. Role models may for

16

example help by providing information, which alleviates both uncertainty and the cost of starting the business (Minniti, 2005). We therefore also control for whether the potential nascent entrepreneur knows any other entrepreneurs. Finally, start-up rates may be influenced by whether the potential entrepreneur is employed while deciding to start his/her business. Because in our theoretical interpretation, we focused on the role of women within the households and on what we called a ‘pecking order of economic activity’, we control for whether women are within or outside employment using additional interactive effects. We know from the literature that the incentives for ‘offensive’ entry (from employment) and for ‘defensive’ entry (from outside employment) may be different for men and women (Berglann et al., 2009). 5

3.5

Dependent variables: start-up and high growth aspiration start-up When considering entrepreneurship, one needs to consider the broader economic

impact and this leads us to analyse high growth aspiration projects in addition to entrepreneurship as a whole. Thus, we estimate our models using the entire GEM sample (that is on the respondents drawn from the whole working age population) and utilize two nested dependent variables coded as dummies (the second is a subcategory of the previous one). Our first indicator of entrepreneurial entry relates to any start-up projects as in Aidis et al. (2010).The second indicator concerns entrepreneurs that at the time of entry aspire to create ten jobs and more over a period of five years time (18% of start-ups). In theory, one could instead use a continuous variable capturing the target employment of nascent entrepreneurs. However, there are a number of problems with this approach. We found the distribution of this variable differed significantly from normal because of a large share of expected zero employment and a considerable number of outliers located beyond both inner 5

We verified that no other relationships are sensitive to the omission of those effects; these alternative specifications are available on request.

17

and outer fence of the distribution (both as defined by inter-quartile range). Moreover, the larger values of target employment have considerable measurement error, which is strongly correlated with size, above some threshold level; all values above 10-20 employees are reported with rounded figures and their distribution becomes sparse at higher levels. Furthermore, to rely on a continuous employment variable would necessitate the use of a selection model, controlling for entry as a first stage but it would be difficult to identify such a model using this dataset. The nested nature of our dependent variables (each subsequent category is a subset of the previous one) enables us to estimate the determinants of each category of entrepreneurial entry separately, without concern for cross-equation correlation. This comes at a cost of efficiency. More specifically, separating the entry decision and the growth aspirations could lead to more significant estimates 6. Definitions and summary statistics for our dependent variables are presented in Table 3 below.

{Table 3}

3.6. More on methodology We control for annual time effects in all our estimations and experimented with the full set of country fixed effects, but the probit estimator collapses without producing credible Wald statistics. This problem arises because the GEM dataset is highly unbalanced with

6

We also estimated two stage probit-probit selection models but this is open to criticism on two grounds: First, the separation of the entry decision and growth expectations could be questioned, second, alternative choices of selection variables for the first (entry) equation is also problematic, as already indicated in the previous section. We therefore opt for the more conservative estimation strategy just outlined, though the two-stage model results were somewhat stronger. We also investigated the possibility of applying multiple probit, where both highaspiration entrepreneurship and low-aspiration entrepreneurship are pitched against inactivity (lack of entry). However, the results of Small-Hsiao tests rejected this model, indicating that there is strong interdependence in the three alternatives.

18

many countries appearing just once or twice over time. Moreover, the introduction of fixed country effects while retaining institutional measures is highly problematic, as the cross sectional variation of the latter is washed out leaving only time-variation which for institutional indicators consists predominantly of measurement error (as changes in institutions are often registered with random time lag and imprecisely). Moreover, it is precisely the most fundamental institutions that vary least over time, and these are the institutions that we expect to have most impact, i.e. property rights. To confirm this intuition, we undertook a simple experiment. We took the full set of institutional and policy measures as reported by Heritage Foundation /Wall Street Journal for all countries over the entire available data period, 1995-2010. Next we assigned a value of zero to all observations which did not change from the previous year and a value of one to those which did change. The property rights index changed only in 14% of observations while at the other end of the spectrum the monetary policy indicator, changed for 91% of observations. 7 As a result, instead of using fixed effects, we cluster our standard errors on countryyear groups allowing for within-group correlation. We consider clustering on country-years rather than on countries as more appropriate (producing higher, more conservative estimates of standard errors), because it corresponds to the logic of GEM sampling and corrects for sample specific issues 8.

7

A less sophisticated experiment is just to compare in between and within variation for each indicator. The resulting ordering is similar. 8 Without this correction, all our significance levels will appear artificially much better. This also applies to models with country fixed effects: driven by overall huge sample size, using country fixed effects instead of clustering standard errors generates some selective highly significant results, but those are misleading and are very sensitive to any change in specification; see out discussion above. There is a big difference in significance levels, when we erroneously focus on variation of several hundred thousand individual observations, instead of variation in fifty countries. We also utilise information in the GEM dataset which attributes population-based weighting to each observation. All our estimations apply this weighting. We thank Paul Reynolds for drawing our attention to this issue. However similarly to Ardagna and Lusardi (2010), we found little difference in estimation results with and without weights.

19

3.7 Framework for analysis In the next section, we start with a simple description of the data based on correlations at a country level. These concern the relationship between the country-level ratios of male to female start-up rates and the institutional coefficients corresponding to the dimensions captured by our hypotheses. We next move to multivariate estimations that utilise all sample information. Our core model is constructed as follows. If i denotes individuals, j denotes countries and t denotes time, we estimate an equation of the form: Prob(Entry) ijt = f(Rule of Law jt , Size of the State jt , GDP/capita jt , GDP growth rate jt , Individual Level Controls jit , Female ijt , Interactive effects between “institutions” and Female jit ), where Entry is a dummy variable denoting whether or not an individual in a particular country at a particular date is engaged in nascent start up activity. We use Probit as our estimator, reporting robust standard errors and marginal effects instead of coefficients to make interpretation easier. We allow for the possibility that the observations are not independent for each country-year sample in our dataset and this is reflected in the estimated variance-covariance matrix and reported coefficients, which is based on clustering. In addition, we explore the robustness of our results and develop our interpretation of the findings by focusing in greater depth on two issues for female entrepreneurship: high aspiration entrepreneurship versus all-encompassing measure of entry and replacing our general institutional measures with more specific dimensions of institutions and policies that are likely to have differential impact on women. The first extension focuses on the role of female entrepreneurs in high aspiration start-up; that is a start up for which respondent expects to create 10 or more jobs in five years time. To explore this, we estimate the

20

likelihood of the ‘high aspiration’ entry. In the second extension, we replace in turn our two core institutional indicator with female-specific indicators of institutions and policies.

4. Discussion of Results

4.1. Country level correlations Table 4 present country-level averages for all key variables, and Figures 1-2 present country level correlations between the ratios of female to male start-up rates and our core institutional indicators. We present ordinary least square lines with 5% confidence intervals, where we found these significant. Consistent with the logic of the regression models applied below, we also control for the level of development. Therefore, the vertical axes in both figures correspond to the residuals of these institutional measures from regressions on GDP per capita at purchasing power parity.

{Table 4}

There is no clear pattern linking the rule of law to the gender composition of entrepreneurship (Figure 1). The horizontal spectrum spans between Singapore, a rich country with significant scope for government arbitrariness, and countries like India, Jamaica, Peru and South Africa, where the quality of institutions is higher than one might expect for their level of development. While there is relatively low share of women in entrepreneurship in Singapore, one finds high female entrepreneurship rates in the other four countries where institutional quality is relatively high controlling. The Philippines is an outlier, having both relatively strong institutions and a high share of female

21

entrepreneurship 9. However, this positive relationship is counterbalanced by a number of countries with a low share of women in entrepreneurship, Greece being a most notable example. On the other hand, a more limited state sector seems to correlate with a higher share of women in entrepreneurship (Figures 2). Emerging markets countries tend to have more a limited state sector and more female entrepreneurship (with Philippines, Thailand, Indonesia, Malaysia and Peru representing the most characteristic cases). In contrast, several European countries – e.g. Sweden, Denmark France, Croatia - have large state sector and relatively limited rates of female entrepreneurship. United Arab Emirates seems to be an outlier with a limited state but very low rates of female entrepreneurship; interpretation of this case not simple because two other countries with similar cultural and religious traits – Malaysia and Indonesia - look very different.

{Figures 1-2}

4.2. Multivariate estimations: probit and bivariate probit models results Our estimations, which are all based on probit models, are presented in Tables 5-7. The benchmark results are reported in Table 5 columns 1 and 2, where the first model corresponds to the general measure of start-up taken as a dichotomous dependent variable, and the second one to high growth aspiration entrepreneurship only. All subsequent specifications are presented for these two dependent variables. Considering that the estimation is based on both country-level and individual heterogeneity, the regression has a good fit (with a slightly better fit for the high aspiration entrepreneurship models and the signs and significance of the controls are consistent with the findings in the literature. Thus, 9

As our subsequent regression models could be sensitive to inclusion/exclusion of Philippines, we run all our models eliminating observations from Philippines. However, the results were not affected: if anything they appeared slightly more significant thanks to an improved fit.

22

we confirm that the probability of becoming an entrepreneur is lower for women, and older people (after some threshold level) and enhanced by education (but with the incremental effect of higher education being insignificant), by social interaction with entrepreneurs, by being a business angel in the past, by employment status (being in employment makes entrepreneurial entry more likely) and by lack of fear of failure. Turning to the cross country controls, entrepreneurship is found not to be related to the business cycle (model 1), but the relationship between high aspiration entry and lagged GDP growth is marginally significant. Moreover, the change in sign is interesting; perhaps while broadly defined entry is countercyclical (inversely related to GDP growth, in model 1) indicating some push effects, high aspiration entrepreneurship is procyclical. Similarly, one can discern interesting differences related to GDP per capita. While there is a negative and significant association with the likelihood of (broadly defined) entrepreneurial entry, for high aspiration entry the effect has the opposite sign and is insignificant. This is consistent with there being a higher proportion of higher value added projects amongst start-ups in developed countries.

Our hypotheses concern property rights and the size of the government. Commencing with the effects on entrepreneurship as a whole, these two core institutional effects are statistically significant for broadly defined and high aspiration entrepreneurship: Thus as predicted a larger government is associated with a lower probability of entrepreneurship (note the reverse sign, following Heritage Foundation coding) and effective constraints on executive, our measure of rule of law, is associated with both broad and higher aspiration entry. Of those two variables, the former (size of the state) has somewhat more significant effect that the latter (rule of law), consistent with Aidis et al. (2010).

23

To test our hypotheses concerning differential effects by gender, we introduce in models 3 and 4 interactive terms between the institutional indicators and the dummy representing a female respondent. Our second hypothesis is found to be confirmed: the signs of the interactive coefficients are as expected, and the interactive effects with government size are significant, both for broad and for high aspiration entry. However our first hypothesis is not confirmed in these regressions; the coefficient on the interactive term is not even weakly significant in either column 3 or 4. As noted previously, this may be because our measure of rule of law is too general to capture gender-specific effects. To explore this possibility, we ran models (5)-(8), in which our previous measure of the rule of law is replaced by two more gender relevant features, namely restrictions imposed on the freedom of movement of women away from home, and the OECD measure of inadequate protection of women against violence. Those experiments reveal interesting results. While not affecting broadly defined entry, discrimination against women is found to influence their likelihood of high aspiration entry. Thus, restrictions on movements by women have no significant effects on broadly defined entrepreneurship, perhaps because many forms of self employment are perfectly compatible with being physically constrained to the close neighbourhood. However, their effect on high aspiration entrepreneurship is negative and highly significant (at 0.001 level). Moreover, rules imposing constraints on women affect both female and male entrepreneurship equally strongly, while the differential effect (related to interactions term) is insignificant. The effects of violence against women are directionally similar but weaker in terms of statistical significance. This may indicate some displacement effects, where women fall back on self-employment if participation in other forms of employment is risky. However, the effect on high aspiration female entrepreneurship is again negative, albeit this time it is marginally insignificant (at the 12% level or 10.8% for a model with the Philippines taken out).

24

We performed an additional set of experiments replacing the size of the government measure with two more specific indicators of government welfare arrangements, namely maternity rights and childcare provision. In this case, we find that institutional and policy measures specifically affecting women have a significant impact on both female and male entry, and also a differential effect on female entrepreneurship. In particular, maternitywelfare related measures decrease the likelihood of (broadly defined) entry, and the effect on women is marginally (but significantly) stronger. The effect is also significant for high aspiration entry, albeit here the differences between man and women are no longer significant (models 9 and 10). This suggests that optimising decisions are typically taken jointly within the households and many welfare measures affect both sexes, even if targeting women. A similar pattern relates to the availability of childcare. Its effect is negative, both in case of broadly defined entry and in case of high aspiration entry. However, in this case there is also a significant gender effect on high aspiration female entrepreneurship (models 11-12).

5. Conclusions Aidis et al. (2010) confirmed the theoretical argument that the rule of law would play an important role in the institutional environment conducive to entrepreneurial activity. Our results take this further, indicating that the impact extends to high growth aspiration entry. We had expected that a stronger rule of law would be associated with enhanced female specific rights that would combat gender inequality (Jütting et al. 2006). However, there we were not able to identify a differential impact between male and female entrepreneurs when we use a general indicator of rule of law in our models 3 and 4. Our findings are consistent with the view that the key element in the rule of law for entrepreneurs concerns the reduction of transactions costs in doing business, and that these affect male and female entrepreneurs equally. If gender inequality in access to rights is associated with a weaker rule of law, the

25

former may still affect both female and male entrepreneurship: male entrepreneurship may also to some extent suffer as a result of constraints inflicted on women via the joint effects on households. It is possible that the gender inequality which restricts female mobility and social interactions and therefore leads to lower start-up rates is a factor that is not measured well by general institutional quality. Hence we need to look into more specific indicators; restrictions imposed on freedom of movement away from home by women and protection from violence against women. We find an interesting pattern. Discrimination against women either has either no effect on broadly defined entry or even some weak positive effects (in case of violence, via self-employment, which may be seen as a second best labour market alternative). In contrast, the impact of gender discrimination is significant and negative on the likelihood of high aspiration entrepreneurship (in the case of restriction on movement of women). We have robust and clear-cut findings on the links between the size of the state sector and female entrepreneurship.. The size of the state sector has a significant and negative impact on both entry and female entry, and this applies to high aspiration entrepreneurship as well. We interpret this result as consistent with prevailing social structures where women are more likely to take more caring and domestic responsibilities within households. It follows that women will be more responsive to trade-offs between domestic work and outside involvement, and so will respond more strongly to higher incentives to gain additional income (when level of state provision of services such as education and health are lower) and to disincentives associated with a larger state in terms of higher taxes. Moreover, since a smaller state enhances not just female entrepreneurship, but high aspiration female entry, the effect cannot be caused solely by better access to welfare support. If this were true, the

26

‘push’ factors would dominate. In contrast, we see that ‘offensive’ entrepreneurship by women suffers as well, and this is likely to be due to incentive effects caused by higher taxes. We also investigated models where general measures of government size are replaced by welfare variables which specifically target women. We found that these significantly affect the likelihood of both female and male entrepreneurship. We interpret this as an evidence that optimising decisions are usually taken in the context of a household where factors affecting women have also implications for the behaviour of men.

27

References Acemoglu, Daron and Simon Johnson. 2005. “Unbundling Institutions” Journal of Political Economy 113 (5): 949-995. Acemoglu, Daron and Thierry Verdier. 2000. “The Choice between Market Failures and Corruption” American Economic Review 90 (1): 194-211. Acs, Zoltan, David Audretsch and David Evans. 1994. “Why does the self-employment rate vary across countries and over time?” Discussion Paper 871. London: CEPR. Aidis, Ruta, Saul Estrin and Tomasz Mickiewicz. 2008. “Institutions and Entrepreneurship Development in Russia: A Comparative Perspective” Journal of Business Venturing, 23 (6): 656-672. ______. 2010. “Size Matters: Entrepreneurial Entry and Government”. Small Business Economics. DOI 10.1007/s11187-010-9299-y. Algan, Yann and Pierre Cahuc. 2007. “The Roots of Low European Employment: Family Culture?” In Christopher Pissarides and Jeffrey Frenkel, eds., NBER International Seminar on Macroeconomics 2005. Cambridge MA: MIT Press, 65-109. Barzel, Yoram. 1997. The Economic Analysis of Property Rights. Cambridge, UK: Cambridge University Press. Baumol, William. 1990. “Entrepreneurship: productive, unproductive, and destructive”. Journal of Political Economy 98 (5): 893-921. Beach, William and Tim Kane. 2007. Methodology: Measuring the 10 Economic Freedoms. Washington DC: The Heritage Foundation. Retrieved March 5, 2009 from http://www.heritagefoudation.org Becker, Gary. 1971. The Economics of Discrimination. Chicago IL: University of Chicago Press. Berglann, Helge, Espen Moen, Knut Røed and Jens Skogstrøm. 2009. “Entrepreneurship: Origins and Returns”. Discussion Paper 4250. Bonn: IZA. Bielby, Denise and William Bielby. 1988. “She Works Hard for the Money: Household Responsibilities and the Allocation of Work Effort” American Journal of Sociology 93 (5): 1031-1059. Botero, Juan, Simeon Djankov, Rafael La Porta, Florencio Lopez-de-Silvanes and Andrei Shleifer. 2004. “The Regulation of Labor”. Quarterly Journal of Economics 119 (4): 13391382. Bowen, Harry and Dirk de Clercq. 2008. “Institutional Context and Allocation of Entrepreneurial Effort”. Journal of International Business Studies 39 (4): 747-67.

28

Bruce, Donald. 1999. “Do Husbands Matter? Married women entering self-employment” Small Business Economics 13 (4): 317-29. Brush, Candida. 1990. Women and enterprise creation: Barriers and opportunities. Paris: Organization for Economic Cooperation and Development. Brush, Candida. 2006. “Women Entrepreneurs: A Research Overview”. In: Mark Casson, Bernard Yeung, Anuradha Basu, and Nigel Wadeson, eds., Oxford Handbook of Entrepreneurship. Oxford UK: Oxford University Press. Carree, Martin, André van Stel, Roy Thurik and Sander Wennekers. 2002. £Economic Development and Business Ownership: An analysis using data of 23 OECD countries in the period 1976 – 1996”. Small Business Economics 19 (4): 271 – 290 Carter, Sara and Peter Rosa. 1998. “The financing of male- and female-owned businesses”. Entrepreneurship and Regional Development 10 (3): 225-242. Casson, Martin. 1982. The Entrepreneur. An Economic Theory. Oxford UK: Martin Robertson. Coleman, Susan. 2000. “Access to Capital and Terms of Credit: A Comparison of Men- and Women-Owned Small Businesses”. Journal of Small Business Management 38 (3): 37-52. Cooper, Arnold C. and William Dunkelberg (1987). „Entrepreneurial Research: Old Questions, New Answers and Methodological Issues”. American Journal of Small Business 11 (3): 11-23. Costello, Anna and Jason Osborne. 2005. “Best Practices in Exploratory Factor Analysis”. Practical Assessment, Research & Evaluation 10 (7): 1-9. Crawford, Sue and Elinor Ostrom. 1995. “A Grammar of Institutions”. American Political Science Review 89 (3): 582-600. Davidsson, Per and Benson Honig. 2003. “The Role of Social and Human Capital among Nascent Entrepreneurs”. Journal of Business Venturing 18 (3): 301-331 Delmar, Frédéric and Per Davidsson. 2000. “Where do they come from? Prevalence and characteristics of nascent entrepreneurs”. Entrepreneurship and Regional Development 12 (1): 1-23. Demirguc-Kunt, Asli, Inessa Love and Vojislav Maksimovic. 2006. “Business Environment and the Incorporation Decision”. Journal of Banking and Finance 30 (11): 2967-2993. Desai, Mihir, Paul Gompers and Josh Lerner. 2003. „Institutions, Capital Constraints and Entrepreneurial Firm Dynamics: Evidence from Europe”. Harvard Negotiation, Organizations and Markets Research Papers 03-59. De Soto, Hernando. 2001. The Mystery of Capital. (London: Black Swan.

29

DiMaggio, Paul and Walter Powell. 1983. “The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields”. American Sociological Review 47 (1): 147-160. Dixit, Avinash. 2004. Lawlessness and Economics. Princeton: Princeton University Press. Economic Intelligence Unit and World Bank (2009). Women’s economic opportunity. A new pilot index and global ranking. London and Washington D.C.: Economic Intelligence Unit and World Bank. England, Paula. 1984. “Wage Appreciation and Depreciation: A Test of Neoclassical Economic Explenations of Occupational Sex Segregation”. Social Forces 62 (3): 726-800. Evans, David and Boyan Jovanovic. 1989. “An Estimated Model of Entrepreneurial Choice Under Liquidity Constraints”. Journal of Political Economy 97 (4): 808 – 827. Evans, David and Leighton L. (1989). “Some Empirical Aspects of Entrepreneurship. American Economic Review” 79 (3): 519 – 535. Fabrigar, Leondre, Duane Wegener, Robert MacCallum and Erin Strahan. 1999. Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods 4 (3), 272-299. Giavazzi, Francesco, Fabio Schiantarelli and Michel Serafinelli. 2009. “Culture, Policies and Labour Market Outcomes”. NBER Working Paper 15417. Greene, Patricia, Candida Brush, Myra Hart and Patrick Saparito. 2001. “Patterns of venture capital funding: is gender a factor?” Venture Capital 3 (1): 63-83. Greene, Patricia, Myra Hart, Elizabeth Gatewood, Candida Brush and Nancy Carter. 2003. “Women Entrepreneurs: Moving Front and Center: An Overview of Research and Theory”. Coleman White Paper Series. Gros, Daniel and Alfred Steinherr. 2004. Economic Transition in Central and Eastern Europe. Cambridge UK: Cambridge University Press. Harper, David. 2003. Foundations of Entrepreneurship and Economic Development. Abingdon: Routledge. Hellman, Joel, Geraint Jones and Daniel Kaufmann. 2003. “Seize the State, Seize the Day: State Capture and Influence in Transition Economies”, Journal of Comparative Economics, 31 (4): 751-773. Hisrich, Robert D. and Candida G. Brush. 1984. “The Woman Entrepreneur: Management Skills and business problems”, Journal of Small Business Management, 22 (1): 30-37. Hurst, Erik and Annamaria Lusardi. 2004. Liquidity Constraints, Household Wealth and Entrepreneurship. Journal of Political Economy 112 (2): 319-347.

30

Jacquette, Jane. 1990. “Gender and Justice in Economic Development”. In Irene Tinker, ed., Persistent Inequalities. Oxford UK: Oxford University Press): 54-69. Johnson, Simon, John McMillan and Christopher Woodruff. 2002. “Property Rights and Finance” American Economic Review 95 (5): 1335-1356. Jütting, Johannes, Christian Morrisson, Jeff Dayton-Johnson and Denise Drechsler. 2006. “Measuring Gender Inequality”, OECD Development Centre Working Paper No. 247. Kirzner, Israel. 1973. Competition and Entrepreneurship. Chicago IL: University of Chicago. Klapper, Leora, Luc Laeven and Raghuram Rajan. 2006. Entry Regulation as a Barrier to Entrepreneurship. Journal of Financial Economics, 82 (3): 591-629 Korosteleva, Julia and Tomasz Mickiewicz. 2008. “Finance and Entrepreneurial Entry”. Centre for the Study of Economic and Social Change in Europe Working Paper No 96. University College London. La Porta, Rafael, Florencio Lopez-de-Silanes and Andrei Shleifer. 2008. “The Economic Consequences of Legal Origins”. Journal of Economic Literature 46 (2): 285-332. Levesque, Moren and Maria Minniti. 2006. “The Effect of Aging on Entrepreneurial Behavior”. Journal of Business Venturing 21 (2): 177-194. Lobel, Sharon. 1991. “Allocation of Investment in Work and Family Roles”. Academy of Management Review 16 (2): 507-521. Marlow, Susan and Dean Patton. 2005. “All Credit to Men? Entrepreneurship, Finance and Gender”. Entrepreneurship Theory and Practice 29 (6): 717 – 735. Mickiewicz, Tomasz. 2005. Economic Transition in Central Europe and the Commonwealth of Independent States. Houndmills UK: Palgrave Macmillan. Mickiewicz, Tomasz. 2009. “Hierarchy of Governance Institutions and the Pecking Order of Privatisation: Central-Eastern Europe and Central Asia Reconsidered”, Post-Communist Economies 21 (4): 399-423. Minniti, Maria. 2005. “Entrepreneurship and Network Externalities”. Journal of Economic Behavior and Organization 57 (1): 1-27. Minniti, Maria, Pia Arenius and Nan Langowitz. 2005a. GEM 2004 Report on Women and Entrepreneurship. Retrieved October 19, 2007 from http://www.gemconsortium.org Minniti, Maria, William Bygrave and Erkko Autio. 2005b. Global Entrepreneurship Monitor. 2005 Executive Report. Babson College, MA, US and London Business School, UK. Retrieved September 8, 2007 from http://www.gemconsortium.org Muravyev, Alexander, Oleksandr Talavera and Dorothea Schäfer. 2009. “Entrepreneurs’ Gender and Financial Constraints: Evidence from International Data”. Journal of Comparative Economics, 37 (2): 270-286. 31

North, Douglass. 1990. Institutions, Institutional Change and Economic Performance. New York NY: Cambridge University Press. North, Douglass. 1994. ‘Economic performance over time’. American Economic Review, 84 (2): 359 – 368. North, Douglass and Robert Thomas. 1973. The Rise of the Western Word: A New Economic History. Cambridge UK: Cambridge University Press. Olson, Mancur. 2000. Power and Prosperity. New York: Basic Books. Orser, Barbara, Allan Riding and Kathryn Manley. 2006. “Women Entrepreneurs and Financial Capital”. Entrepreneurship Theory and Practice 30 (5): 643-665. Parker, Simon. 2004. The Economics of Self-Employment and Entrepreneurship. Cambridge UK: Cambridge University Press. Pett, Marjorie A., Nancy R. Lackey and John J. Sullivan. 2003. Making Sense of Factor Analysis. Thousand Oaks, CA: Sage Publications. Razavi, Shahra. 1997. “Fitting Gender into Development Institutions” World Development 25 (7): 1111-1125. Reynolds, Paul, William Bygrave, Erkko Autio, Larry Cox and Michael Hay. 2002. Global Entrepreneurship Monitor: 2002 Executive Report. Retrieved September 8, 3007, from http://www.gemconsortium.org Robinson, Peter and Edwin Sexton. 1994. “The Effect of Education and Experience on SelfEmployment Success”. Journal of Business Venturing 9 (2): 141-156. Rodrik, Dani. 2000. “Institutions for High-Quality Growth: What They Are and How to Acquire Them”. Studies in Comparative International Development 35 (3): 3 – 31 Russell, Daniel. 2002. “In Search of Underlying Dimensions: The Use (and Abuse) of Factor Analysis”. Personality and Social Psychology Bulletin 28 (12): 1629-1646. Sabarwal, Shwetlena, Katherine Terrell and Elena Bardasi. 2009. “How do Female Entrepreneurs Perform? Evidence from Three Developing Regions”. Paper presented at the World Bank conference on Female Entrepreneurship, mimeo. Schreiner, M. (2001). “Informal Finance and the Design of Microfinance”. Development in Practice, 11 (5): 637-640. Schumpeter, Joseph. 1934. The Theory of Economic Development. Cambridge, MA: Harvard University Press. Schwartz, Eleanor. 1976. “Entrepreneurship: A New Female Frontier”. Journal of Comtemporary Business 5 (1): 47-76.

32

Shragg, P., L. Yacuk and A. Glass. 1992. “Study of Barriers Facing Albertan Women in Business”. Journal of Small Business and Entrepreneurship 9 (4): 40-49. Sonin, Konstantin. 2003. “Why the Rich may Favor Poor Protection of Property Rights”. Journal of Comparative Economics 31 (4): 715-731 Van Stel, André, David Storey and Roy Thurik. 2007. “The Effect of Business Regulations on Nascent and Young Business Entrepreneurship”. Small Business Economics 28 (2-3): 171 – 186. Verheul, Ingrid and Roy Thurik. 2001. “Start-up Capital: Does Gender Matter?”. Small Business Economics 16 (4): 329-341. Verheul, Ingrid, Sander Wennekers, David Audretsch and Roy Thurik. 2001. “An Eclectic Theory of Entrepreneurship”. Tinbergen Institute Discussion Papers No 030/3. Verheul, Ingrid, André van Stel and Roy Thurik. 2006. “Explaining female and male entrepreneurship across 29 countries”. Entrepreneurship and Regional Development, 18 (2): 151-183. Vonderlack, R. & Schreiner, Mark (2002). “Women, Microfinance, and Savings: Lessons and Proposals”. Development in Practice, 12 (5): 602-612. Wennekers, Sander, André van Stel, Roy Thurik and Paul Reynolds. 2005. “Nascent Entrepreneurship and the Level of Economic Development”. Small Business Economics 24 (3): 293 – 309. Williamson, Oliver. 1985. The Economic Institutions of Capitalism. Firms, Markets, Relational Contracting. New York: The Free Press. Williamson, Oliver. 2000. “The New Institutional Economics: Taking Stock, Looking Ahead”. Journal of Economic Literature 38 (3): 595-613.

33

Table 1: Correlation matrix for institutional and macroeconomic variables, country-level mean data Variables

Executive constraints (Polity IV)

Property Government Rights Size (Heritage) (Heritage)

Credit to private sector / GDP

GDP growth

Executive constraints (Polity IV) Property Rights (Heritage) Government size (Heritage)

.29

1

-.43

-.40

1

Credit to private sector / GDP

.21

.57

-.06

1

GDP Growth

-.36

-.41

.32

-.25

1

GDP pc ppp

.24

.81

-.39

.66

-.35

GDP pc ppp

1

1

Source: GEM 2001-2006. Lagged values. Government size: larger value implies less spending.

34

Table 2: Descriptive statistics and definitions of explanatory variables used in estimations Variable Definition and source Institutional variables and macroeconomic controls: Constraints on Polity IV ‘Executive Constraints’; scores executive from 1=”unlimited authority” to 7=”executive parity”; higher denotes less arbitrariness Size of government Heritage Foundation index based on government expenses (including consumption and transfers) / GDP, higher value → lower spending, range: 0 to 100 GDP per capita GDP per capita at purchasing power parity, constant at 2005 $USD (WB WDI 04 2009) GDP growth Annual GDP growth rate (WB WDI 04 2009) Restricted freedom Restricted freedom of movement of movement (opportunity to move freely outside the house) for women (OECD Development Centre), time invariant; range 0-1 Violence against Violence against women, lack of relevant women legislation (OECD Development Centre), time invariant; range 0-1 Maternity leave Composite indicator that assesses length of maternity leave and benefits coverage (Economic Intelligence Unit), time invariant, range 0-3.1 Childcare Availability, affordability and quality of childcare services, as well as the role of the extended family (Economic Intelligence Unit), time invariant, range 1-5. Personal characteristics: Age The exact age of the respondent Age squared Age squared Female 1=female, 0 otherwise Employment 1=respondent is either in full or part time employment, 0 if not Post-secondary & 1=respondent has a post secondary or higher education higher education attainment, 0 otherwise Higher education 1=respondent has a higher education attainment Business angel 1=business angel in past three years, 0 otherwise Current owner of 1=current owner/manager of business, 0 business otherwise Knows other 1=personally knows entrepreneurs, in last entrepreneurs two years, zero if not Fear of failure 1=respondent believes that the fear of failure would not prevent him/her from starting a business

Mean

S.D.

6.71

.8786

46.5

24.1

26240

10114

3.05

2.28

.03

.12

.26

.17

2.09

.88

3.60

.94

43.13 2096.1 .5187 .6209

15.35 1429.0 .4996 .4852

.3421

.4744

.1352

.3419

.0330

.1786

.0883

.2837

.3619

.4806

.6510

.4766

35

Table 3: Descriptive statistics and definitions of dependent variables Categories of entrepreneurial entry: Start-up activity excluding self-employment

Start-up 10 jobs or more

Definition

1=respondent is engaged in startup activity expecting to employ other people in 5 years time or earlier, zero otherwise 1= respondent is engaged in startup activity and expects to create 10 or more jobs in 5 years time, zero otherwise

Mean

.02884

.1674

.0086

.0921

Notes to Tables 2 and 3:

1. Source: GEM 2001-2006 unless specified otherwise. 2. All statistics for institutional and macroeconomic variables are reported for their lagged values. 3. All statistics are based on the set of observations actually used in estimations (i.e. eliminating joint effect of missingness in all variables).

36

Table 4. Institutional and macro variables, country averages Country Start-up rate HGA start-up rate Argentina 8.0% 2.9% Australia 4.9% 1.2% Austria 2.8% 0.9% Belgium 2.1% 0.5% Brazil 5.1% 0.9% Canada 4.6% 1.7% Chile 7.8% 3.5% China 5.5% 2.5% Colombia 10.6% 5.4% Croatia 2.8% 1.2% Czech Rep 5.0% 2.6% Denmark 2.7% 1.1% Ecuador 16.7% 1.6% Finland 2.3% 0.5% France 2.2% 0.3% Germany 3.0% 1.1% Greece 4.3% 0.3% Hong Kong 1.4% 0.8% Hungary 2.9% 0.7% Iceland 6.6% 2.0% India 7.1% 0.7% Indonesia 9.6% 0.9% Ireland 4.4% 1.4% Israel 2.3% 0.8% Italy 2.4% 0.7% Jamaica 11.2% 0.9%

Gov. size 77.8 61.7 22.6 24.8 71.8 50.1 84.2 88.4 68.3 25.2 36.8 7.9 76.8 25.9 14.5 32.4 53.0 90.0 27.7 37.7 76.7 86.0 67.0 24.1 30.0 57.4

Exec constr. 6.0 7.0 7.0 7.0 6.0 7.0 7.0 3.0 6.0 7.0 7.0 7.0 6.0 7.0 6.0 7.0 7.0 7.0 7.0 6.0 7.0 7.0 7.0 7.0

GDP growth GDP pc (ppp) Restr. on move 3.4 10006 0 3.2 30719 0 2.0 33616 0 1.9 31424 0 3.4 8265 0 2.7 34399 0 4.1 11679 0 10.3 3829 0 6.9 7635 0 4.7 14612 6.8 21674 0 2.6 33364 0 8.0 6426 0 2.9 29714 0 1.7 30200 0 0.7 31111 0 4.4 24624 0 4.4 31563 0 4.2 16143 0 4.4 32956 0 5.9 1976 0.5 5.5 3335 0.5 5.5 36685 0 1.3 22199 0 1.1 28207 0 2.1 7125

Violence 0.25 0.5 0.166667 0.17 0.58 0.25 0.42 0.58 0.33 0.416667 0.25 0.17 0.5 0.25 0.17 0.33 0 0.5 0.5 0.33 0.67 0.17 0.08 0.42

Maternity Childcare 1.4 3 0 5 1.864 4 2.7 5 2.5 3 2 3 2.3 4 0.4 4 2.455 3 2.784 2.9 1.3 3.1 2.9 1.4 1.7 0.5 2.9 2.4 0.4 0.293 2.9 2.5 2.6

4 5 3 5 5 3 2 5 4 5 3 3 2 5 3

37

Table 4. Continued Country Japan Jordan Korea Latvia Malaysia Mexico Netherlands New Zealand Norway Peru Philippines Poland Portugal Russia Singapore Slovenia South Africa Spain Sweden Switzerland Taiwan Thailand Turkey Uganda UK United Arab Em. United States Uruguay Venezuela

Start-up rate HGA start-up rate Gov. size Exec constr. GDP growth GDP pc (ppp) Restr. on move Violence Maternity Childcare 56.8 7.0 1.5 29542 0 1.0% 0.67 1.7 3 0.5% 66.3 3.0 8.6 4149 0.5 10.5% 0.25 0.3 2 1.4% 81.2 6.0 5.6 19966 0 5.0% 0.17 1.3 4 1.6% 51.9 7.0 11.4 13877 3.2% 1.4% 75.1 4.0 5.8 12205 0.5 4.8% 0.42 0.381 2 0.5% 81.6 6.0 2.4 12420 0 7.6% 0.17 2 3 1.2% 32.0 7.0 1.5 34440 0 2.1% 0.17 2.8 5 0.6% 52.4 7.0 3.9 23672 0 7.7% 0.33 2.8 4 1.9% 35.6 7.0 2.4 45864 0 3.4% 0.25 3 5 1.0% 81.6 7.0 6.4 6362 0 29.5% 0.42 1.8 3 7.3% 88.9 6.0 5.3 3027 0 5.5% 0.17 1.254 4 0.2% 43.1 7.0 2.7 12465 0 3.9% 0.33 1.9 4 1.0% 40.0 7.0 1.8 20639 0 2.0% 0.25 2.838 3 0.1% 65.0 5.0 5.7 10402 0 2.1% 0.25 2 2 0.9% 90.1 3.0 6.2 41530 0 3.2% 0.5 0.5 5 1.5% 42.9 7.0 4.4 22913 2.5% 1.0% 79.5 7.0 4.1 8107 0.5 3.1% 0.42 1.7 3 0.6% 51.9 7.0 3.6 27370 0 2.4% 0.25 2.6 3 0.4% 0.6 7.0 3.7 31289 0 1.5% 0 3 5 0.6% 58.9 7.0 1.4 35313 0 2.8% 0.25 1.9 4 0.9% 82.7 6.0 4.3 18000 1.2% 0.4% 90.9 5.4 5.0 6359 0 7.4% 0.33 1.4 4 1.8% 68.1 7.0 6.9 11584 0 1.9% 0.416667 1.254 3 1.6% 84.9 3.0 6.7 866 0.5 15.0% 0.75 0.4 3 1.4% 48.5 7.0 2.7 31806 0 2.3% 0.08 2.5 3 0.8% 76.1 9.4 51586 0.5 1.6% 0.75 0.126 2 1.0% 63.9 7.0 2.3 40243 0 6.1% 0.33 0 3 2.3% 45.8 7.0 4.6 10106 0 6.7% 0.42 1.356 3 2.3% 83.3 5.0 1.3 8899 0 17.8% 0.42 1.6 3 5.3%

38

Table 5 Estimation Results

Dependent

Age Age squared Female In employment Post-secondary & higher educ. Higher education Business angel in last 3y Current owner of business Knows other entrepreneurs Fear of failure not prevent start-up Size of government (reverse) Constraints on executive

(1) startup

(2) startup, expects >10 jobs

(3) startup

(4) startup, expects >10 jobs

0.00212*** (0.000306) -3.02e-05*** (3.99e-06) -0.0125*** (0.00113) 0.0141*** (0.00210) 0.00623*** (0.00154) 0.00147 (0.00285) 0.0397*** (0.00310) -0.00784*** (0.00186) 0.0365*** (0.00234) 0.0208*** (0.00110) 0.000562*** (8.16e-05) 0.00360* (0.00178)

0.000189*** (5.73e-05) -3.37e-06*** (7.47e-07) -0.00395*** (0.000358) 0.00163*** (0.000342) 0.00226*** (0.000478) 0.000625 (0.000682) 0.00807*** (0.00108) -0.000330 (0.000371) 0.00757*** (0.000645) 0.00360*** (0.000264) 7.36e-05*** (1.41e-05) 0.000491+ (0.000273)

-0.000446 (0.000633) -4.23e-07** (1.37e-07)

0.000215+ (0.000131) 1.17e-08 (2.59e-08)

0.00208*** (0.000303) -2.98e-05*** (3.95e-06) -0.0312*** (0.00763) 0.0143*** (0.00208) 0.00622*** (0.00152) 0.00152 (0.00283) 0.0393*** (0.00307) -0.00783*** (0.00183) 0.0362*** (0.00231) 0.0206*** (0.00110) 0.000472*** (7.98e-05) 0.00320+ (0.00167) 0.000224*** (4.45e-05) 0.000962 (0.000876) -0.000435 (0.000626) -4.16e-07** (1.35e-07)

0.000184** (5.65e-05) -3.29e-06*** (7.37e-07) -0.00626** (0.00210) 0.00164*** (0.000336) 0.00223*** (0.000470) 0.000628 (0.000672) 0.00793*** (0.00107) -0.000333 (0.000363) 0.00743*** (0.000633) 0.00354*** (0.000260) 6.33e-05*** (1.36e-05) 0.000472 (0.000287) 3.27e-05** (1.02e-05) 5.15e-05 (0.000220) 0.000211 (0.000129) 1.18e-08 (2.53e-08)

474745 4104*** -80087 0.11

468434 2730*** -20973 0.12

474745 4277*** -80036 0.11

468434 2928*** -20963 0.12

Gov. size (rev.) x Female Constraints on exec x Female GDP growth rate GDP per capita (ppp)

Observations Wald Chi squared Log pseudolikelihood Pseudo R squared

39

Table 6. Estimation results. Dependent Age Age squared Female In employment Post-secondary & higher educ. Higher education Business angel in last 3y Current owner of business Knows other entrepreneurs Fear of failure not prevent startup Size of government (reverse) Restricted freedom of movement

(5) startup 0.00215*** (0.000306) -3.03e-05*** (3.97e-06) -0.0226*** (0.00223) 0.0146*** (0.00217) 0.00632*** (0.00156) 0.00196 (0.00274) 0.0405*** (0.00324) -0.00776*** (0.00188) 0.0365*** (0.00235) 0.0204*** (0.00118) 0.000385*** (7.32e-05) -0.000602 (0.0129)

(6) startup, expects>10jobs 0.000198*** (5.56e-05) -3.45e-06*** (7.25e-07) -0.00560*** (0.000617) 0.00159*** (0.000339) 0.00215*** (0.000466) 0.000715 (0.000679) 0.00832*** (0.00108) -0.000296 (0.000370) 0.00743*** (0.000635) 0.00355*** (0.000268) 5.98e-05*** (1.15e-05) -0.00740*** (0.00195)

Violence against women Gov. size (rev.) x Female Freedom of move x Female

0.000187*** (4.04e-05) 0.00462 (0.00587)

2.85e-05*** (8.32e-06) 0.000418 (0.00219)

Violence x Female GDP growth rate GDP per capita (ppp) Observations Wald Chi squared Log pseudolikelihood Pseudo R squared

-0.000199 (0.000608) -3.69e-07* (1.77e-07) 466739 4100*** -76110 0.10

0.000153 (0.000111) -1.00e-08 (2.79e-08) 460534 2394*** -20010 0.12

(7) startup 0.00213*** (0.000307) -3.01e-05*** (3.97e-06) -0.0238*** (0.00229) 0.0148*** (0.00214) 0.00619*** (0.00152) 0.00225 (0.00272) 0.0401*** (0.00308) -0.00782*** (0.00186) 0.0361*** (0.00230) 0.0203*** (0.00119) 0.000361*** (6.85e-05)

(8) startup, expects>10jobs 0.000198*** (5.59e-05) -3.44e-06*** (7.27e-07) -0.00546*** (0.000632) 0.00172*** (0.000339) 0.00219*** (0.000478) 0.000740 (0.000664) 0.00818*** (0.00109) -0.000325 (0.000375) 0.00751*** (0.000648) 0.00354*** (0.000270) 5.64e-05*** (1.21e-05)

0.0111 (0.00803) 0.000167*** (4.11e-05)

-0.000118 (0.00136) 3.66e-05*** (9.76e-06)

0.00841+ (0.00476) -0.000366 (0.000632) -3.29e-07* (1.53e-07) 466739 4430*** -76588 0.10

-0.00197 (0.00125) 0.000154 (0.000121) 1.09e-08 (2.64e-08) 460534 2328*** -20075 0.11

40

Table 7. Estimation results. Dependent Age Age squared Female In employment Post-secondary & higher education Higher education Business angel in last 3y Current owner of business Knows other entrepreneurs Fear of failure not prevent start-up Constraints on executive Maternity leave

(9) startup 0.00214*** (0.000319) -3.05e-05*** (4.16e-06) -0.00524 (0.00511) 0.0142*** (0.00251) 0.00685*** (0.00163) 0.00258 (0.00261) 0.0385*** (0.00320) -0.00723*** (0.00197) 0.0352*** (0.00259) 0.0206*** (0.00136) 0.00465* (0.00189) -0.0100*** (0.00135)

(10) startup, expects>10jobs 0.000197** (6.24e-05) -3.49e-06*** (8.16e-07) -0.00208+ (0.00126) 0.00169*** (0.000400) 0.00233*** (0.000508) 0.000794 (0.000677) 0.00803*** (0.00106) -0.000313 (0.000406) 0.00740*** (0.000704) 0.00366*** (0.000330) 0.000637+ (0.000361) -0.00143*** (0.000277)

Childcare Constraints on exec x Female Maternity leave x Female

-0.000446 (0.000894) -0.00202+ (0.00106)

-0.000339+ (0.000187) 0.000250 (0.000194)

Childcare x Female GDP growth rate GDP per capita (ppp) Observations Wald Chi squared Log pseudolikelihood Pseudo R squared

0.000697 (0.000771) -7.59e-07*** (1.74e-07) 454410 4923*** -78576 0.10

0.000318+ (0.000166) -3.64e-08 (3.29e-08) 448586 3170*** -20650 0.12

(11) startup 0.00216*** (0.000334) -3.09e-05*** (4.37e-06) 0.000296 (0.00588) 0.0139*** (0.00242) 0.00876*** (0.00197) 0.000987 (0.00301) 0.0413*** (0.00344) -0.00651** (0.00200) 0.0363*** (0.00277) 0.0212*** (0.00135) -1.62e-05 (0.00192)

(12) startup, expects>10jobs 0.000194** (6.33e-05) -3.48e-06*** (8.29e-07) 0.000216 (0.00131) 0.00169*** (0.000374) 0.00257*** (0.000534) 0.000595 (0.000695) 0.00850*** (0.00114) -0.000223 (0.000407) 0.00756*** (0.000727) 0.00374*** (0.000311) -5.20e-05 (0.000366)

-0.00536** (0.00190) -0.00123 (0.000778)

-0.000583+ (0.000327) -0.000276 (0.000181)

-0.00120 (0.000944) 0.000803 (0.000730) -6.80e-07** (2.07e-07) 454410 4917*** -78534 0.10

-0.000634** (0.000205) 0.000321* (0.000161) -2.06e-08 (3.80e-08) 448586 3086*** -20694 0.12

41

Notes to Tables 5-7: 1. All equations estimated using probit. 2. Marginal effects reported instead of coefficients. 3. All macro level variables lagged one year, except for: restrictions of freedom of movement, violence against women, maternity, and childcare which are all time invariant. 4. A set of year dummies included but not reported due to shortage of space. 5. Standard errors clustered on country-years. 6. *** significant at 0.001, ** significant at 0.01, * significant at 0.05, + significant at 0.10.

42

2.5

Figure 1. Rule of law and ratio of female start-up rate to male start-up rate

Singapore

Uganda China Jordan

Indonesia Peru Mexico Afr Ecuador South Jamaic Brazil New Zealand India Australia Finland Hungary Argentina Taiwan Chile Switzerla Italy Colombia United St Latvia Spain Korea Uruguay Poland Canada Austria France Germany Denmark UK Portugal Japan Sweden Turkey Netherlands Russia Ireland IsraelCroatia CzechBelgium Rep Slovenia Norway Greece

Venezuela Malaysia

Thailand

0

Ratio of female to male startup rates .5 1 1.5 2

Philippines

-4

-3 -2 -1 0 1 Residuals from Executive Constraints on Size on GDP pc ppp

Note: Countries with rule of law (controlled for GDP pc) are located on the right. Source: authors calculations based on GEM 2001-2006 country averages and Polity IV database.

43

2.5

Figure 2. The size of the state sector and ratio of female start-up rate to male start-up rate

Ratio of female to male startup rates .5 1 1.5 2

Philippines

Indonesia Peru Thailand Malaysia Mexico Ecuador Venezuela Uganda Jamaica India Brazil South Afr China Finland Australia New Zealand Hungary Chile Jordan Italy Switzerla Colombia Latvia United St France Spain Denmark Uruguay Poland Argentina Germany UK Canada Taiwan Austria Iceland Japan Korea Ireland Belgium Portugal Netherlands Turkey Sweden Russia Croatia Israel Czech Rep Slovenia Greece Norway

Singapore Hong Kong

0

United Arab Em

-50

0 Residuals from HF Government Size on GDP pc ppp

50

Note: countries with larger state sector (controlled for GDP pc) are located on the left. Source: authors calculations based on GEM 2001-2006 country averages and Heritage Foundation /Wall Street Journal

44