THE CASE OF SELF-EMPLOYMENT

Self-employment is a two-faced concept in labor economics. On the one hand, it is considered the basis of economic dynamism and the soul of capitalism. In line with Schumpeter’s theory, the entrepreneur is the prominent figure of capitalism: an individual capable of perceiving economic opportunities, combining various resources, and taking risks.

On the other hand, self-employment is presented today as a way of mitigating the human consequences of the economic crisis by diminishing the unemployment rate (even artificially). More generally, self-employment is considered a route out of poverty for unemployed or marginal populations. The flexibility allowed by self-employment is also seen as a convenient way to accommodate constraints on the labor market: it allows flexible hours and schedules, can often be done at home, and so offers an intermediate situation between inactivity and salaried work. It is therefore not surprising that the European plan in favor of entrepreneurship suggests that policy should aim “to promote specific actions for reaching out to groups that are [...] under-represented within the entrepreneurial population” (OECD 2013, p. 4): migrants, seniors, the unemployed, young people—and women.

Gender studies on self-employment lie at the crossroads between these two approaches. The gender gap in self-employment is analyzed in terms of the characteristics required to become an entrepreneur.

In this section, we present the statistics available on self-employment rates in OECD countries,^[778] showing that women are less likely to be self-employed than men in all OECD countries. Why do self-employment rates for women differ from those for men? A growing strand of research examines the role played by psychological characteristics (risk aversion, self-confidence, and perception of economic opportunities) in explaining this gap. Then we turn to studies of differences in the way men and women run businesses. Some studies examine the possibility that women face discrimination in the access to credit. But most analyses are centered on gender differences in family con­straints as the main reason for differing performance between men and women.

12.4.1 Stylized Facts

12.4.1.1 The Fuzzy Scope of Self-Employment

Self-employment covers a wide and heterogeneous range of economic activities: tradi­tional farmers, regulated professions in law or health, small business, start-ups, and so on. To measure it, the OECD has adopted a comprehensive definition: “Self-employment is a form of employment in which people work in their own business, farm or professional practice and receive some economic benefit for their work, such as wages, profits, in-kind benefits or family gain (for family workers).

As Ahmad and Seymour (2008) emphasized, entrepreneurship should not be con­fused with the above measure of self-employment. Entrepreneurial activity is defined as “the enterprising human action in pursuit of the generation of value, through the cre­ation or expansion of economic activity, by identifying and exploiting new products, processes or markets” (OECD, 2007). Consequently, entrepreneurship is not limited to small business units, but includes innovative large companies. In practice, studies often confuse self-employment and entrepreneurship. This has consequences on gender ana­lyses, which tend to confuse the relative dearth of self-employed women with the lack of female entrepreneurship.^[779]

In practice, comparable statistics on self-employment are based on national labor force surveys. The GlobalEntrepreneurship Monitor (GEM) measures entrepreneurship activ­ities through annual household surveys of the adult population in 54 countries and includes a set of questions on motivations and aspirations. GEM is also the main source of information on nascent entrepreneurship (businesses started less than 42 months and more than 3 months ago).

12.4.1.2 The Gender Gap in Self-Employment

In all developed countries the self-employment rate—the proportion of self-employed people relative to all employed people—is lower for women than for men, whatever the average level of self-employment.

In the 27 European Union countries in 2011, the self-employment rate was 9.7% for women and approximately the double for men (18.3%); the average rate was 15%, with great variations across countries (OECD, 2013). Some countries, such as Denmark, France, and Germany, have a much lower female self-employment rate (around 5%) than the EU average, whereas others, such as Italy and Greece, have much higher rates (around 20%).

The United States has a relatively low rate of self-employment, equal to 11% in 2009 (based on Bureau of Labor Statistics figures in Hipple, 2010) and stable since 2003.^[780] The female self-employment rate is, as expected, nearly half the male rate (8% and 14%, respectively), and the gap is particularly large in incorporated self-employment (2% and 5%, respectively).

In all countries, gender segregation by industry in self-employment seems similar to the segregation in salaried employment. Women are relatively absent in manufacturing and building, which account for approximately 25% of male self-employment. Self­employed women are more likely to work in consumer-orientated services (health, social work, arts, education). These sectors represent around 40% of female self-employment (OECD, 2013).

12.4.2 Why So Few Women in Self-Employment?

The lower propensity of women to start businesses is well documented (see Allen and Langowitz, 2013; Koellinger et al., 2013; Langowitz and Minniti, 2007; OECD 2013, Chapter 2; for the most recent references based on the GEM database; Anderson and Wadensjo, 2008 on the Swedish case; Furdas and Kohn, 2010, on the German case). Once a business is started, in general there are no significant differences in survival rates across genders when economic characteristics are taken into account (for the United States, see, e.g., Perry, 2002).

87

Why do women start fewer businesses than men? Gender differences in observed human capital are not sufficient to explain the gender gap in self-employment (Minniti and Nardone, 2007; Wagner, 2007).

A large strand of literature has sought to explain the lack of self-employed women in terms of psychological and social factors, which can be classified in three broad categories: risk aversion, perception of economic opportunities, and preferences for self­employment. According to the model developed by Kihlstrom and Laffont (1979), in which individuals make their occupational decision by comparing the risky returns of entrepreneurship with the nonrisky wage in the competitive labor market,^[781] gender dif­ferences in risk aversion are a natural candidate for explaining the gender gap in self­employment. Risk aversion may also play a role by discouraging access to financial resources to start a business; using the English Household Survey of Entrepreneurship 2003, Sena et al. (2012) found that women are more reluctant than men to borrow from banks, and this gender difference in the search for external funding reduces their prob­ability of becoming self-employed. In addition to risk aversion, perception of opportu­nities (see Kirzner, 1979), self-confidence, and knowing other entrepreneurs are crucial characteristics for starting a business. Consequently, the lack of self-confidence among women, the gender difference in perception of the economic environment (women tend to perceive themselves and the entrepreneurial environment in a less favorable light than men do), and the higher fear of failure lower the propensity to start a business (Langowitz and Minniti, 2007; Minniti and Nardone, 2007; Koellinger et al., 2013; Wagner, 2007).

However, some recent studies have argued that the importance of social and psycho­logical factors and gender stereotypes should not be overestimated, and that the decision to be self-employed also depends on economic opportunities for both women and men. A macroeconometric analysis of longitudinal UK data, estimating the extent to which women are influenced by economic factors, shows that economic factors (gross domestic product [GDP], interest rates, house prices, employed and self-employed incomes) influ­ence equally the self-employment choices of both women and men (Saridakis et al., 2013). The impact of economic factors on the decision to be self-employed also has been studied using microanalysis by comparing the behavior of individuals with similar human capital. Generally, the behavioral difference between men and women in the propensity to become self-employed may be strongly mitigated when the population is homoge­neous. Analyses restricted to professional and managerial self-employment indicate that women choose self-employment to pursue a careerist model and to obtain work auton­omy as men do (for the United States see Budig, 2006a). Leoni and Falk (2010), who analyzed the propensity to be self-employed among Austrian university graduates, find that age and the field of studies explain two-thirds of the gender gap observed in self­employment. When the group studied is limited to medical graduates, they no longer observe any gap between men and women in the propensity to be self-employed.⁸ But it could be argued that medical graduates are a self-selected group that attracts women prepared to exercise as licensed professionals.

12.4.3 Self-Employed Women: Family Constraints and Gaps in Working Hours and Earnings

Given the lower propensity of women to start businesses, what factors might encourage women to become entrepreneurs? One motivation for women to become self-employed could be to avoid the gender wage discrimination they face on the labor market^{[782] [783] and to be rewarded according to their productive characteristics. Evidence, however, is not very convincing: empirical work tends to find a small unexplained part in the gender earnings gap of self-employed workers, once productive characteristics and work hours have been controlled for (for the United States see Clain, 2000; for Canada see Leung, 2006).[784] But despite these results, which suggest that women have higher returns on their human cap­ital when they are self-employed, the role of gender wage discrimination in the transition toward self-employment is not firmly proved (Leung, 2006; see Williams, 2012, for a set of European countries).}

On the contrary, there is a general consensus that familial responsibilities play a dom­inant role in women’s choice of self-employment and that this status is often chosen by (married) women as a way to achieve a better work-life balance, given their family con­straints (Carr, 1996). The family status and the presence of young children is therefore a key variable explaining the choice of self-employment because the flexibility in working hours allows parents to reduce the cost of childcare (Connelly, 1992). Empirical studies have regularly found a positive influence of fertility on women’s selection into self­employment (Boden, 1999; Macpherson, 1988). Women with young children are more likely than men to declare flexibility of schedule and child constraints as main reasons for becoming self-employed (for the United States, see Boden, 1999). Therefore self­employment seems to be a close substitute for part-time work and labor market inactivity, especially in countries where there are few public childcare services (see Georgellis and Wall, 2004 on the US case). To be covered by their husband’s health insurance is also a positive factor of women’s self-employment, especially in the US case (Devine, 1994). Lombard (2001) presents a careful econometric study of the choice of self-employment over wage/salary employment by married women in the United States. The women’s requirements in terms of monetary (wages) and nonmonetary job attributes (flexibility and a nonstandard work week) were estimated in a first stage, and these estimates, together with husbands’ health insurance, were used to explain the self-employment decision. The conclusion is that these three factors—the relative earnings potential when self-employed, the demand for flexibility and a nonstandard work week, and the hus­band’s health insurance—positively affect the probability of women being self-employed.

Another way to demonstrate the key role played by work-family time in self­employment is to compare the use of time by self-employed people and employees. According to a study based on the Australian Time Use survey (Craig et al., 2012), self-employed mothers spend more time on domestic work and childcare and their paid work hours are shorter than those of salaried mothers, whereas the fathers’ time does not vary much across employment types. Some studies also have found positive nonmonetary outcomes for women (mothers) who have opted for self-employment: self-employed married women report greater job satisfaction, less occupational burnout, and less neg­ative spillover from work to home (based on a 1997 US survey; Hundley, 2001).

Given these differences in work commitment and working hours, it is therefore not surprising that self-employed women are regularly found to perform worse than men. The raw gender earnings gap (defined as the difference between male and female average self-employment incomes divided by the male average self-employment income) is around 35% for OECD countries as a whole, with large variations between countries.

Portugal and Poland (countries where farms are still important) have the largest gap (60%), Nordic countries (Iceland, Sweden, Denmark) the smallest (10%), and the US and most European countries are at around 40% (OECD, 2013, gender data portal).⁹ But, as mentioned above, the unexplained part of this gap is small: in most countries this gender difference in earnings generally vanishes quite entirely when structural differences (human capital, financial capital, sectors of activity), and particularly women’s reduced working hours and their commitment to children and housework, are taken into account (for the United States see Budig, 2006b; Fairlie and Robb, 2009; Marshall and Flaig, 2014; see Du Rietz and Henrekson, 2000, for Sweden; see Young and Wallace, 2009, for Canadian lawyers). Hundley (2000) analyzes this male/female earnings gap in self-employment in terms of their different reasons for being self-employed (compared with salaried work). The starting point is that self-employment offers a broader range of options of work organization than salaried work: self-employment is not bounded by minimum work hours (and corresponding minimum pay), nor by maximum hours, con­trary to salaried work. Combinedwith the gender division of labor (see Section 12.6), this implies that self-employed married women will do more hours of housework than sal­aried women (and less market work), and conversely self-employed men will do less hours of housework than salaried men (and more market work). As a result, the gender earnings gap is greater in self-employment than in salaried work. Empirical analysis based on the US Panel Study of Income Dynamics survey confirms that marriage, family size, and hours of housework have a negative effect on self-employed women’s earnings and a positive effect on those of self-employed men. Hundley (2000) also suggests that this pat­tern may operate as a barrier to entry into business by ambitious women, as they may fear that self-employment will oblige them to do the lion’s share of the housework.

12.4.4 Are Women Discriminated Against in Their Access to Credit?

Another well-documented stylized fact on gender differences in entrepreneurship is that self-employed women make less use of bank loans than men do. Are women discrimi­nated against by banks, experiencing more credit denials or higher interest rates than men? Or do they simply apply for less credit? This issue has been largely studied in the United States (Blanchard et al., 2008; Blanchflower et al., 2003); it is regularly found that women do not face discrimination in terms of access to loans, unlike ethnic or racial groups. Asiedu et al. (2012) confirmed these results using the latest Survey of Small Busi­ness Finances in 2003. The 2003 database allows them to extend the analysis to loan renewals (and not only new loans). Loan renewals, compared with new loans, are

⁹² There are specific methodological difficulties in collecting data on self-employment earnings: how to dis­tinguish between the return on capital and the payment for hours worked, how to treat negative earnings, the bias due to incomplete declarations, etc. (see Section 12.2). However, these biases are assumedto affect men’s and women’s earnings equally.

interesting because there is less information asymmetry, so unexplained differences in denial rates (with white men as the reference group) can be attributed to discrimination with more certainty. Aftercontrollingforselection bias and nonlinearity (i.e., the possible different impact of some of the control variables according to the level of applicants’ cre­dentials), Asiedu et al. found discrimination against ethnic minorities, but not against (white) women; they even found that women tend to pay a slightly lower interest rate than white males.

Then the limited use of banking credit may stem from gender differences in behavior. Although there may be no gender differences in the granting of bank loans, differences in the perception and expectation of their banking relationship may negatively affect women’s demand for credit. Saparito et al. (2013) studied gender differences in the per­ception of banking relationships; their research was based on a survey of 696 matched pairs of business owner/managers and bank managers and that included qualitative vari­ables. They found that gender influences the quality of the relation between business owner and banker: the male pairs (a male firm owner and a male bank manager) have the highest level of trust and satisfaction with credit access, and the female pairs have the lowest level of these indicators. Interestingly, an influence of gender is also observed in women’s access to angel capital. In general, women are less likely to seek angel financ­ing, although they have an equal probability of obtaining funding from this source; they are also more likely to seek and obtain financing from women angels (Becker-Blease and Sohl, 2007).

However, the role of behavioral variables should not be overestimated in explaining gender differences in access to credit; the differences in structural characteristics explain a large part of this gap. Using Survey of Small Business Finances over two decades, Cole and Mehran (2009) summarized the major characteristics of female-owned firms as fol­lows. These firms are smaller, younger, more likely to be in retail trade and business ser­vices, and more likely to be organized as proprietorships rather than corporations. Female owners are younger, less educated, and less experienced than male owners. Econometric analysis indicates that these structural differences explain why female-owned firms are less likely to obtain credit than male-owned firms.

The above findings on the absence of gender banking discrimination are for the US case. There is no certainty that the result is valid in other OECD countries. In fact, Alesina et al. (2013) offer a counterexample in the case of Italy, where the share of self-employment is high and a lot of microfirms are owned by women. They find robust evidence that women pay more for credit than men, even after taking into account a large number of characteristics relating to the type of business, the borrower, and the structure of the credit market. The disadvantage is higher in male-dominated industries. The authors suggest two explanations for this result: a taste-based discrimination against women, as Italian society still has very traditional values concerning the place of women in society, and asking for a loan could be considered unsuitable, and a lower ability to negotiate good deals with banks because women are more averse to competition and more reluctant to ask.

To sum up, some features already highlighted in the analysis of the gender wage gap also seem to be relevant to gender differences in self-employment, namely the choice of field of study, industrial segregation, and (once again) family constraints and the need for flexible work hours to reconcile childcare and professional activity. The possibility of combining work with staying at home seems to be contradictory to the relative scarcity of self-employed women. Gender differences in risk aversion seem to provide a solid basis for explaining this paradox, but their importance should not be overestimated. Finally, self-employment is a very heterogeneous field, with huge differences in types of business (how can one compare the behavior of lawyers and farmers?) and across countries, and for which data are still too limited (compared with those for wages) to provide a compre­hensive picture of gender inequalities.

12.5.