در شکاف درآمدی مرد / زن و زنان خود اشتغال

This paper examines the gender differences in the choice to become self-employed. Of particular interest is the relationship between the male/female earnings gap in wage and self-employment, and the male/female differences in the average predicted probability of self-employment. It has been argued that earnings inequality in wage-employment lead women to choose self-employment. However, it may be the case that inequality in the form of consumer discrimination causes an earnings gap between males and females in self-employment. If inequality is higher in self-employment than in wage-employment, then there would be females in wage-employment who would be in self-employment in the absence of inequality in both sectors.

It is observed in the US that the male/female earnings gap is larger in self-employment than in wage-employment. Devine (1994) showed that in 1990, the ratio of median female to male hourly earnings in wage and self-employment was 0.75 and 0.48, respectively.1 It is also observed that the incidence of self-employment is much lower for females than males. Although males and females share many common determinants in their wage versus self-employment decision, it has been found that there are notable gender differences as well. Boden, 1996 and Boden, 1999a, Carr (1996), Connelly (1992), and Hundley (2000) find that the probability of self-employment for females is positively associated with marriage and having young children, whereas for males they find no significant relationship. Some of the above authors have also suggested that self-employment is seen by males as a way to capture the benefits of extra work effort. Boden (1999a) revealed that 11.9% of males claimed that the main reason they became self-employed was for the money or because the money was better, compared to 6.6% for females.2 Hundley uses some of these gender differences in the probability of self-employment to explain differences in the earnings between males and females in the self-employment sector. In his paper, he argues that females enter self-employment because it offers them the opportunity to choose their own hours, enabling them to perform their role as the primary worker at home and to perform market work. If there is a certain amount of time an individual must work before they produce any output, then hourly earnings of females in self-employment would tend to be lower than their male counterparts who spend more time in the market. Hundley hypothesizes that marriage and the number of young children increases the work load of females at home and thus indirectly lowers their productivity in market work. In his empirical work, he finds that the ratio of female to male hourly earnings in both employment sectors are almost identical and close to one when comparing similar males and females who are single with no children.3 If one is married with no children, then the ratio remains close to one in self-employment but falls to 0.88 in wage-employment. This is contrary to his hypothesis that females choose self-employment to accommodate increased home work because of marriage. However, in line with his hypothesis, an increasingly positive difference between the ratio in wage-employment and self-employment opens as one has more children. A drawback of Hundley's work is that he does not control for actual labor market experience. Since experience is probably negatively correlated with marriage and the number of children, the lower returns to marriage and children for females compared to males could be exaggerated in the wage and self-employment sectors. This paper synthesizes the research on the male/female differences in the probability of self-employment and the male/female differences in the earnings in wage and self-employment. Using the Canadian Survey of Labour and Income Dynamics (SLID) in 1994, the log hourly earnings gap in wage and self-employment are decomposed into the portion explained by differences in the endowment of productivity related variables and the residual—the portion explained by differences in the returns to these endowments. The residual portion of the earnings gap can be attributed to many things. As hypothesized by Hundley, differences in non-market productivity that cause one gender to specialize more in non-market work may be the cause. Alternatively, the residual gap could be explained by other unmeasured differences in productivity or differences in the reporting of earnings or hours worked between the sexes. This paper considers the effect of misreporting of hours worked and specialization in non-market activities in explaining the earnings gap in both employment sectors. Using the coefficients from the earnings equations, the expected earnings differential between self and wage-employment is then calculated and used in the estimation of structural probit equations for male and female self-employment choice. In this way, the direct effect of individual characteristics on self-employment choice can be distinguished from the indirect effect of these characteristics through expected earnings. The causes of the male/female difference in the average probability of self-employment are then investigated. Male/female differences in individual characteristics, in the coefficients of the structural probit, and in the coefficients of the earnings equations are all considered. The latter case is of particular interest because the portion of the earnings gap explained by the differences in the returns to endowments is often attributed to discrimination. Since the nature of the discrimination in each of the sectors is not the same – wage-employed females may face employer discrimination and self-employed may face consumer discrimination as described in Becker (1971) and Borjas and Bronars (1989) – the impact of discrimination on earnings in both sectors could potentially be quite different. Thus, the different levels of discrimination across sectors could explain the low female self-employment rate. If discrimination against females is higher in self-employment than in wage-employment, then there would be females in wage-employment who would be in self-employment in the absence of discrimination in both sectors. In this case, pay and employment equity legislation, while lessening discrimination in wage-employment, would increase the number of females misallocated to wage-employment.4 While discrimination in wage-employment is well documented, the meaning of discrimination in self-employment is less clear. As mentioned above, one possible source of discrimination in self-employment is consumer or buyer discrimination. Individuals may simply dislike buying goods or services from females. Although Borjas and Bronars (1989) suggest consumer discrimination in the context of racial differences, the United States Small Business Administration (1995) suggests that consumer discrimination based on gender differences exists as well. The credit market may also be a source of discrimination in self-employment as banks may be less willing to give loans to self-employed females. Less access to capital would limit the size and the success of female owned businesses. Empirical evidence on such discrimination is mixed. Lee-Gosselin and Grise (1990) find that 38% of female business owners in their study feel their gender was related to obstacles they faced in dealing with their customers, suppliers and banks. Furthermore, Riding and Swift (1990) find that there are gender differences in the terms of the loans made available by banks. On the other hand, Fabowala et al. (1995) show that after controlling for business related factors gender differences in the access to credit disappear. The effect of relative earnings between self and wage-employment on self-employment choice has been studied before by Rees and Shah (1986), Fujii and Hawley (1991), Bernhardt (1994) and Simpson and Sproule (1998). However, the first three concentrate solely on men and hence do not study the difference in the determinants of self-employment choice between genders. The latter does consider females and uses the SLID as well, but they use annual earnings in their regressions, do not consider the impact of the misreporting of hours worked and do not allow the presence of young children to affect earnings. In this paper, it is found that the male/female log hourly earnings gap in Canada is larger in self-employment than in wage-employment. After controlling for labor market experience, one-third to two-fifths of the earnings gap in self-employment can be attributed to differences in the returns to the presence of young children. Overall, the decomposition of the gaps into the explained and unexplained portion yields the result that in most cases the explained portion is larger in self-employment than in wage-employment. Therefore, if females were paid according to the male's earnings function in both wage and self-employment, female wage-employment earnings would rise more than female self-employment earnings and the average probability of female self-employment would decline. Since differences in individual characteristics account for only one-quarter to two-thirds of the gap in the average predicted probability of self-employment between males and females, it is the male/female differences in the way the probability of self-employment responds to changes in individual characteristics that is responsible for the gap.

In this paper, the gender differences in the choice to become self-employed in Canada are investigated. Previous papers (Boden, 1996, Boden, 1999a, Carr, 1996 and Connelly, 1992) have studied gender differences in the probability of self-employment. Others (Rees and Shah, 1986, Fujii and Hawley, 1991 and Bernhardt, 1994) have examined the effect of relative earnings between self and wage-employment on self-employment choice for men. Also, Boden (1999b) has examined gender inequality in wage earnings and female self-employment choice. The contribution of this paper is to look at the effect of gender inequality in both wage and self-employment earnings on the male/female gap in self-employment rates. Although the male/female earnings gap is larger in self-employment than in wage-employment, it is found that a larger fraction of the earnings gap in self-employment can be explained by productivity differences than in wage-employment. Therefore, differences in the returns to productivity related explanatory variables cannot account for the gap in the average predicted probability of self-employment. If differences in the return to characteristics between genders are interpreted as discrimination, this implies females face less discrimination in self-employment than in wage-employment. Also, it is found that although family related characteristics do not explain the entire earnings gap in self-employment as in Hundley (2000), unobserved productivity differences controlled by these family related characteristics are important in explaining the earnings gap in both wage and self-employment. Without these controls, in some cases, the fraction of the earnings gap explained in wage-employment could be larger than in self-employment. This could lead to the erroneous prediction that the average predicted probability of self-employment for females would increase when females are compensated according to the male's earnings functions. Since the earnings gap failed to explain the male/female gap in the average predicted self-employment probability, differences in observed characteristics and differences in the response of the probability of self-employment to these characteristics were considered. Differences in observed characteristics could only explain between 25 and 33% of the gap in the average predicted probability of self-employment, so differences in response to these characteristics are the main source of the gap. The findings in this paper agree with Boden, 1996 and Boden, 1999a, Carr (1996) and Connelly (1992) in that women are more likely to enter self-employment for the purpose of caring for a young child. However, this factor does not explain why fewer women than men are in self-employment. Instead, understanding the reasons behind the fact that experience, higher levels of education and marriage affect the probability of self-employment probability of males more than females, and why being an immigrant and having foreign mother tongue has a positive effect on male self-employment but a negative effect on female self-employment, are key to explaining the gap between male and female self-employment rates. Liquidity constraints are one possible explanation for the difference. Liquidity constraints are likely to be less binding on married males with more experience and education and it might be the case that females have less access to capital than males. Different social attitudes specific to immigrants and individuals with a foreign mother tongue is offered as another possibility. While public policy is likely ineffectual in changing the social attitudes of newly arrived Canadians, it could improve women's access to external capital. Improved access need not necessarily come in the form of direct government sponsored loans. Instead, programs offering greater access to business skills training and mentoring might provide the same benefit. These would improve the quality of business plans presented to banks and would encourage the development of important contacts.