How does motherhood affect self-employment performance?

In most Western economies, female self-employment rates are lower than those of men. As a result, many governments want to increase female self-employment through various targeted projects. Sweden is no exception. The rationale behind such policies is that self-employed individuals contribute to job creation and are important for economic growth. Women are assumed to be as good entrepreneurs as men, and they should be given equal opportunities to start a business. At the same time, there is a body of literature, primarily for Anglo-Saxon countries, that hypothesizes that self-employment is a way for women to facilitate the combination of work and family. A number of survey studies, both for Sweden and other countries, report that it is less common among women to become self-employed as a career choice and more common to choose self-employment for lifestyle reasons, particularly when the children are young, while men, to a larger extent, are driven by pecuniary motives.¹ Studies exploring this idea find in general that there is a positive correlation between the presence of young children in the household and self-employment probabilities among women (Boden 1996, 1999; Budig 2006; Carr 1996; Hundley 2000; Lombard 2001; Wellington 2006). Two recent papers for Norway and Sweden also find a positive correlation between the presence of young children in the household and self-employment entry (Rønsen 2014; Andersson Joona 2017)—this in spite of the existence of family-friendly policies that encourage labor force participation among mothers and fathers with young children. However, one difference found between Sweden and Anglo-Saxon countries is that self-employed women in Sweden spend more, or as much, time on market work as wage-earning women, while the opposite is true in many other countries (Gimenez-Nadal et al. 2012; Gurley-Calvez et al. 2009; Mångs 2013; Johansson Sevä and Öun 2015).

The relationship between self-employment and having children is complex, and most of the abovementioned studies do not pay particular attention to the direction of causality: the presence of children in the household could have an impact on self-employment entry, but self-employment could also have an impact on fertility decisions.² The direction of causality in this relationship has, however, been analyzed in a recent paper that uses an instrumental variable approach (Noseleit 2014). Here, it is found that family size appears to have a causal effect on self-employment but that self-employment does not affect family size.³

In the present paper, we ask novel questions to analyze the relationship between self-employment and fertility. The first question asked is related to the timing between self-employment entry and childbirth. Do women become self-employed before or after childbirth? The second question, which is also the main research question in the paper, is how the timing between self-employment entry and first child birth affects self-employment success. Becoming a successful entrepreneur in general requires much work, and much time needs to be invested in the business. This could be more difficult when having young children. At the same time, women who start their business while having young children might do it for other reasons than to become successful—it could be a way of facilitating the combination of work and family responsibilities.

It is a political goal to increase female self-employment, but not mainly for lifestyle reasons. Although this could be beneficial for the individual (e.g., easier to combine family and work), it would perhaps not be efficient for society if labor market productivity decreases when becoming self-employed. Therefore, it is important to learn more about whether self-employment performance differs depending on the circumstances under which the firm was started—did the (female) founder have children or did she start the firm before having children? We analyze six different measures of self-employment performance: income, being a top income earner, net revenues, corporate form, having employees, and the exit rate. The two last measures can be particularly important for the overall impact of self-employment on employment growth. By employing coworkers, the self-employed contribute to job creation, and the exit rate out of self-employment is related to the unemployment risk among the self-employed themselves.

We find, contrary to our expectations, that women who started their business while having children do not perform worse compared to women who became self-employed when they did not have children. In terms of annual earnings and revenues, we find that women who had children at the start perform better, while we find no difference in the exit rate. There is also some evidence that women who had children when entering self-employment have more employees. We suggest four different explanations for these results. Firstly, women who did not have children when entering self-employment postpone childbearing and have their children later. This then mechanically leads to this group of women, on average, having younger children during the years in which we observe self-employment outcomes. Moreover, the presence of young children has a negative impact on self-employment performance. Secondly, becoming a mother in general can mean that one becomes more efficient at work, independently of whether one is a wage-earner or self-employed. Thirdly, so-called mamatreneurs, a group of self-employed women, who, as a result of their motherhood, discover market opportunities that are related to children or motherhood itself, come up with better business ideas and as a result perform better. Fourthly, there is a selection of more motivated and more ambitious women who manage to start a business when they have (young) children. It is clear from this that we do not claim to have estimated the causal effect of having children at the time of self-employment entry on subsequent self-employment performance. To do this, we need some source of exogenous variation in the timing between self-employment entry and first child births.

In the empirical analysis, we use Swedish register data for the cohorts of women born between 1970 and 1975. We construct a balanced panel and observe self-employment entry and childbirths for each year in the period 1995 to 2013. We exclude women who are self-employed in 1995 and women who already had a child in 1995. Since the sample is fairly young in 1995 (20–25 years), we believe that we are likely to observe the first entry into self-employment and the birth of the first child. Having data on six full cohorts of women and being able to follow them over a period of 19 years gives us a unique opportunity to study self-employment entry and fertility choices and how they are related to each other.

Even though the focus of the paper is on self-employment outcomes, we also look at the impact of having young children on self-employment entry. Previous literature has found a positive correlation between having young children and entering self-employment, and this is also what we find in our sample. In addition, we see that nearly 70% of women in these cohorts who were self-employed at some point between 1996 and 2013 entered self-employment while being a mother, and hence, approximately 30% entered self-employment when they did not have any children.

The remainder of the paper is structured as follows: in Sect. 2, the pros and cons of entering self-employment before and after childbirth are discussed, and we discuss how we expect this to affect outcomes; in Sect. 3, the data and descriptive statistics are presented; in Sect. 4, the results on self-employment entry are presented; in Sect. 5, the results on self-employment performance are presented; and in Sect. 6, the results are summarized, and some conclusions are drawn and their implications discussed.

After graduating and being on the verge of entering the labor market, an individual has to make a number of choices. In this paper, we are interested in the choice of self-employment entry and fertility choices. A woman could enter self-employment for the first time before or after having children. If and when each of these events will take place will be influenced by a number of factors, and we will not go through previous research on factors affecting self-employment entry and fertility choices in general.⁴

Before we turn to the results on self-employment performance, we look at the results for the self-employment entry equation. Among those in the sample who enter self-employment at some point during the observation period, approximately 32.4% enter self-employment when they do not have children, and 67.6% enter self-employment after childbirth. It is hence much more common to become self-employed after having had a child. There are several explanations for this pattern. First, as is argued in several previous papers, the demand for being self-employed might increase when there are children in the household. Second, already having had a child, or possibly all children one plans to have, would mean that one no longer has to worry about how parental leave benefits would depend on self-employment incomes. Third, women who have aspirations to become self-employed might also choose to postpone self-employment entry due to the fact that it might be difficult to combine business ownership and take care of a newborn child. If the mother wants to be on leave full-time from work just after the child is born, she might choose not to be self-employed during this period. This factor might also go in the other direction. Most wage-earning mothers in Sweden take a fairly long parental leave, around 1 year, and few combine parental leave with market work. If one, however, has a preference for a shorter period of parental leave or for combining parental leave and market work in a more flexible way, this would be more easily accomplished if one is self-employed than if one is organizationally employed. A previous Swedish study finds that self-employed women, on average, taker fewer days of parental leave than female wage-earners (Anxo and Ericson 2015). They suggest that this is explained by a portion of women with high performance-related income choosing self-employment over wage-employment.

Here, covariates are measured the same year as self-employment entry (marital status (married/single/other), education (seven categories), region of birth, county of residence), while labor market status 1 year prior to self-employment entry (wage-earner, in education, unemployed, parental leave, and other) is measured at time t − 1.

Our initial hypothesis was that women who become self-employed after childbirth, i.e., those who become self-employed while being a mother of a young child, will not perform as well as women who are childless when they become self-employed. Women who do not have children can potentially spend more time on their business during the start-up phase, something that can be important for the success of the firm. Previous research also points towards differences in performance depending on the motives for becoming self-employed.

In the regressions, we use information on all years in which an individual is self-employed during our observation period of 1996–2013. Depending on the duration of self-employment and on the number of self-employment spells, individuals will appear a different number of times. In all regressions, we include a control for the total number of years in self-employment and year. Standard errors are robust and clustered at the individual level.

Self-employment earnings are measured in logs; hence, we exclude those with negative incomes from self-employment. It is not evident that corporate form is a measure of success. However, up until the beginning of 2010, there was a requirement of a capital investment of 100,000 SEK (approx. 10,000 euro) to start an incorporated firm. On April 1, 2010, this limit was reduced to 50,000 SEK (approx. 5000 euro). In our sample, approximately 27% started an incorporated firm; hence, it is more common to start an unincorporated firm. The size of the firm measured in terms of number of employees is often used as a measure of self-employment performance.¹² The distribution of this variable is skewed, with approximately 49% of the self-employed having no employees. Therefore, we use a Tobit model in the regression analysis. Another measure of the economy of the firm is the net revenues. This is also measured in logs, and hence, those with negative net revenues are excluded.

The last measure of self-employment success is the probability of exiting self-employment. The specification of this model differs slightly from the other outcomes, since we need to condition it on whether an individual is self-employed in period t − 1. All covariates are also measured in period t − 1. Exit is defined as a dummy variable equal to one if someone was self-employed in period t − 1 but was no longer self-employed in period t and zero if someone was self-employed in period t − 1 and continued to be self-employed in period t.

The results for the different measures of self-employment success are somewhat mixed. For self-employment earnings, the probability of being a top income earner and net revenues, we find that women who entered self-employment after childbirth perform better compared to women who entered self-employment before childbirth. In the first earnings regression, which does not include any covariates, we find that self-employed who had a child when entering self-employment on average have lower incomes. The estimate, however, changes sign when controls for children and marital status are included. Having young children is correlated with lower earnings, and since those who were childless when entering self-employment postpone childbearing, this means that they have younger children, on average, when we measure outcomes. In the final regression, where the full set of covariates is included, women who entered self-employment after childbirth have, on average, approximately 23% higher income. For the probability of belonging to the top 10 percentiles in income, we basically find the same pattern. In the first three specifications of the model, it appears that women entering self-employment after childbirth have a significantly lower chance of reaching the highest income percentile, but when controls for children and marital status are included, they have a significantly higher probability.

For net revenues, we find a positive estimate in all specifications. It is large in the first specification but decreases in size as more controls are added. In the last specification, however, it is of the same size as in the first specification. It should be noted that the R ² value is large already in the second specification, when controls for corporate form are included; hence, this appears to explain a large part of the variation in revenues. If the full specification of the model is estimated only for those with unincorporated firms, the estimate for “child at start” is very similar to the corresponding estimate for the whole model (β = 0.253), while the R ² value is only 0.1618.

It can be questioned whether corporate firm is actually a measure of performance, but having an incorporated firm is positively correlated with success in general and is included as a control variable in all the other regressions. In most cases, it is a very strong predictor of self-employment success (Appendix 1 Table 10). We find that women who entered self-employment after childbirth are more likely to have a firm registered as an incorporated firm. The estimates are positive and significant in all specifications, but they decrease in size as more controls are included.

Table 8 presents the results for the number of employees. The distribution of the number of this variable is skewed, with nearly 50% of the self-employed having no employees. Following Burke et al. (2002), we investigate the difference in firm size between the groups and the impact of other covariates by estimating a Tobit model. The same specifications as before are estimated. In the first three specifications, we find a positive correlation between having entered self-employment after childbirth and the number of employees. The interpretation of the estimate in the third specification is that the expected number of employees is 28.6 persons higher in firms started by women who had children. In the fourth and fifth specifications, where controls for having young children, the total number of children in the household and marital status are included, the estimate is no longer significantly different from zero. However, in the last specification, the estimate becomes significantly different from zero again and indicates that women who had a child at the start have, on average, approximately 26 more employees in their firms. In the last specification, we have added four additional covariates: industry, labor market status prior to self-employment entry, earnings prior to self-employment entry, and the regional unemployment rate. Adding these variables one at the time, we see that it is the inclusion of earnings the year before entering self-employment that leads to a significant estimate of child at start. This is positively correlated with the number of employees and could be an indication of financial resources when starting the firm.

The last measure of performance is exits from self-employment. The results from the same specifications as before are presented in Table 9. The presence of young children seems to play an important role for the exit rate as well. When this is not controlled for in the first three specifications, we find that women who entered self-employment after childbirth are significantly more likely to exit self-employment. In the fourth specification, when we include controls for the presence of children and marital status 1 year prior to self-employment exit, the estimate changes sign and is no longer significantly different from zero. We also find that that having young children in itself increases the probability of exiting self-employment. This is consistent with the results found in Lin et al. (2000). Hence, the presence of young children appears to be positively correlated with both entries into and exits from self-employment.

Comparing all outcomes, we see that the results are somewhat mixed. One the one hand, we find that women, who already had children when they became self-employed have higher incomes, are more likely to belong to the top income percentile and have higher revenues. On the other hand, we find no difference in the exit rate. For firm size, the results from the Tobit model suggest that women who had a child at start employ, on average, more employees.¹³ There are, hence, no indications of that women who were mothers when entering self-employment perform worse compared to women who entered self-employment while being childless.

It is interesting to note that the outcomes for which we find a positive effect of motherhood at self-employment entry are short-term measures of self-employment performance. There could potentially be large variations in income and revenues over time, and income and revenues could be affected by fertility rates during this particular year. The inclusion of controls for the presence of young children and the total number of children in the same years as when the outcomes are measured has a large impact on the estimated difference between women who enter self-employment before and after childbirth. More long-term measures of self-employment performance, such as the exit rate, might depend less on the activities or the family situation of the self-employed herself during a single year.

Both labor force participation and fertility rates are high among Swedish women, and women participate in the labor force to a high extent also when the children are young. Even though there is publically provided high-quality child care that makes it possible for mothers, and fathers, with young children to participate in the labor market, having children still means a lot of work.

Anecdotal evidence indicates that it is not uncommon for women in fertile ages to leave a wage-earning job to become self-employed, not primarily because they aim at becoming successful entrepreneurs and earn a lot of money, but because self-employment offers a way of being able to have it all, a career and a family, without being pushed to the limit of one’s capabilities.

Our initial hypothesis was that women who already had children when entering self-employment perform less well compared to women who did not have children when becoming self-employed. One reason to expect that women who are childless when becoming self-employed perform better is that they can potentially spend more time on market work compared to women who have children during the start-up phase of the firm.

Our results do not lend support to this hypothesis. We find that women who had a child at the time of self-employment entry have higher incomes and higher revenues and are employing more people in their firms, while we find no difference in exit rate out from self-employment between the groups. The variables that seem to be most important to control for are the presence of young children and total number of children the same year as outcomes are measured. Having young children is generally negatively correlated with all outcomes, and women who start their firm before they have children are likely to postpone the birth of their first child and, hence, are more likely to have younger children when outcomes are measured.

In a sensitivity analysis, we separately looked at self-employed women who had not yet had a child at the end of our observation period, but this group appears to be the least successful (results not shown). This leads us to the conclusion that female business owners who trade family formation and childbearing for a successful career are few. In our sample, many of the women who are successful as entrepreneurs started their firm while being a mother and while the children were young. If we look at some simple descriptive statistics for women who belong to the top income percentile in the most recent year in which we observe them in self-employment and with positive incomes from self-employment, we see that nearly 75% of women in the top of the income distribution had children when they entered self-employment. At the bottom (P10) and at the median (P50), the share is almost 10% points lower. See Appendix 2 Table 11. A larger share of women at the top have two children, while a lower share have no young children in the household. Women at the bottom of the income distribution stand out in that a larger share have no children, but a larger share also have large families with four or more children. Part of the difference in fertility rates can be explained by differences in age, where both present age and age at the time of self-employment entry is positively correlated with location in the income distribution.

What are possible explanations for these results? First, we observed that the majority of women who did not have children at self-employment entry postpone childbearing and have their children later. Mechanically, this means that this group of women, on average, has younger children when we measure self-employment outcomes, and we have seen that the presence of young children has a negative impact on self-employment performance. Second, becoming a mother in general can mean that one becomes more efficient at work, independently of whether one is a wage-earner or self-employed. If this is the case, the estimated correlations can be interpreted as a causal effect of motherhood on self-employment performance. We do not suggest that this is the only interpretation of our results. Third, so-called mamatreneurs come up with better business ideas and, as a result, perform better. “Mamatreneurs” are defined as a group of self-employed women who, because of their motherhood, discover market opportunities that are related to children or motherhood itself. Fourth, there is a selection of more motivated and more ambitious women who manage to start a business when they have (young) children. If this is the case, it means that we have not estimated the causal effect.

An additional explanation can be borrowed from Lazear’s theory that entrepreneurs are jacks-of-all trades (Lazear 2004). He argues that individuals with a more balanced set of skills are more likely to be self-employed. Becoming a parent means that one learns, for most adults, a totally new set of skills. Some are directly related to taking care of a child, but some are related to the organization of a family in general. It could be things such as having meals ready at a certain hour or having routines at bedtime and in the morning when the family needs to get ready for kindergarten, school, and work. Learning such skills could also be a factor that leads an individual to the decision of becoming self-employed, which might also have a positive impact on self-employment performance.

The entrepreneurship literature often makes a distinction between necessity- and opportunity-driven entrepreneurship. Previous literature has found differences in outcomes depending on type of entrepreneurship. However, in the present paper, it is difficult to arrive at a hypothesis about the correlation between type and having a child at the time of self-employment entry. It would be possible to find entrepreneurs of both types, necessity and opportunity, not only among women who enter self-employment when having children but also among women who do not have children when entering self-employment. In the regression analysis, we control for previous labor market status, which is most likely the information we have that would be our best proxy for whether entrepreneurship is necessity or opportunity driven. Labor market status prior to self-employment is similar between the groups, and the inclusion of this variable in the regressions has a rather small impact on our key variable of interest.

There is an outspoken goal to increase self-employment among women. The primary reason is that female entrepreneurs, just as male entrepreneurs, are believed to contribute to economic growth and job creation. For some individuals, self-employment could also be a solution to a situation of economic inactivity or unemployment. At the same time, several studies indicate that women are less likely to choose self-employment for economic reasons and more likely to choose it for family-related reasons. In this paper, we compare the self-employment outcomes of two groups of women who started their business during different phases of their life cycles. Our results do not indicate that female self-employed who had young children at the time of self-employment entry perform less well than women who are childless when becoming self-employed. One interpretation of these results is that there does not need to be a contradiction between becoming self-employed when the children are young and being a successful entrepreneur.

In future studies, it would be very interesting to see whether similar results can be found in countries with another institutional setting, i.e., where labor force participation among mothers with young children is lower and where many women choose to leave a wage-earning job when the children are young. Also, as is indicated by the analysis of time-use data, self-employed women in Sweden spend as much, or more, time on market work than wage-earning women, while the opposite pattern has been found in many other countries. This could be an indication that Swedish mothers who choose self-employment when the children are young have strong preferences for market work, which also might have an impact on subsequent self-employment success.