Whose Business Is It Anyway
Whose Business Is It Anyway?
Reyes Aterido and Mary Hallward-Driemeier[1]
Research Department, World Bank
May 25, 2009
Abstract: The paper examines how important characteristics of the entrepreneur are in understanding productivity gaps between enterprises in Sub-Saharan Africa. Six dimensions are explored: the gender of the entrepreneur, the education of the entrepreneur, the prior experience of the entrepreneur, the motivation for becoming an entrepreneur, the means by which the enterprise was acquired and the management techniques implemented by the primary decision maker. While other papers have found little evidence of productivity gaps between male and female entrepreneurs in the formal sector in Sub-Saharan Africa, this paper finds that the definition used in classifying enterprises matters. Using a broad definition of female participation in ownership, there are indeed few significant differences by gender. However, narrowing the definition of gender based on decision making authority rather than (partial) ownership, uncovers a 15 percent productivity gap between female-led firms compared to male run enterprises. In terms of the other characteristics of the entrepreneurs, education and management quality are associated with higher productivity and growth. For prior experience and means of becoming an entrepreneur, the only significant effect is through family experience, either by having a father who was an entrepreneur or by joining a family business. The paper examines whether education, management quality and the broader family experience with entrepreneurship can help explain the result on gender. Although there is no additional evidence of non-linearities in the impacts of education and management quality by gender, the positive association of family enterprises on productivity does not hold for women-led enterprises. These results imply that improving women’s education – including business skills – could help raise the relative productivity of women entrepreneurs, with additional longer run benefits that additional female role models and members of networks can help future women entrepreneurs.
1. Introduction
With advancing women’s economic empowerment a Millennium Development Goal, understanding how constraints may vary across men and women entrepreneurs is an important question. The issue is particularly pressing in Sub-Saharan Africa, where women’s economic participation remains lower than men’s, poverty is high and where progress in achieving gender equality still lags in many spheres. Yet, previous papers that have looked for gender differences among formal firms in the region have found few significant differences. This paper reexamines the issue, using supplemental data that distinguishes ownership and decision making authority, and provides richer information on the background of the entrepreneur. It does find that a productivity gap exists between male and female-run firms using the tighter definition of control rather than (partial) ownership. Education and formal management techniques reduce, but do not eliminate, the effect. Rather, it appears that a family background in entrepreneurship and access to networks helps the sons, but not the daughters, in raising productivity and firm growth.
Bardasi, Blackden and Guzman (2007) examined the relative performance of male and female entrepreneurs in the formal sector across 28 countries in North and Sub-Saharan Africa surveyed from 2002-2006. They find few significant differences in either the reported constraints faced by the enterprises or in their productivity. Rather, the striking difference by gender was in the rates of participation, with far more men likely to own enterprises than women. More recent work by Bardasi and Sabarwal (2009) using 22 countries in Sub-Saharan Africa surveyed in 2006 and 2007 finds a similar result; although women may be slightly constrained in the size of their enterprises, they do not find evidence that they are in their productivity or growth.
This paper extends this work in two dimensions, finding that gender differences are present. First, it examines whether the definition of a female enterprise matters. In particular, it distinguishes between ownership and decision making authority. The papers that find little gender differences have used a measure of female participation in ownership. This broader definition could be including firms where the women in fact have little involvement in the running of the firm. Finding few differences with male run firms could thus be a result of including many male-run enterprises in the ‘female’ category. For sole-proprietorships, there is not much concern. Only a handful of such enterprises that have a single female owner are not actually run by women. However, for partnerships, we find that fully 77% do not involve the women owner in the decision making. While the ‘participation in ownership’ measure does not lead to differences in productivity between firms by gender, restricting ‘female led enterprises’ to ones where the women owner is the chief decision maker, does lead to a significant productivity gap of 15 per cent.
Second, the paper then looks whether other characteristics of the entrepreneur also matter for performance and whether there are non-linear effects of these characteristics, particularly by gender. The education of the manager is the most common characteristic that is examined in the literature, with not-surprising finding that more educated entrepreneurs are more likely to be working in more productive enterprises. Of particular interest here four additional sets of characteristics that try to address more specifically dimensions of entrepreneurship: the prior experience of the manager; the reason why the entrepreneur opened the business; how the enterprise was acquired, and management techniques used in the running of the business.
Prior experience could reflect important skills that are being transferred to the enterprise, and distinguishing between the types of enterprises where this experience was gained (e.g. in the formal or informal sector) could provide insights into the importance of job mobility and the potential for the informal sector to contribute to the creation of formal sector firms. However, the results show little evidence that prior work experience matters for current performance. Rather, what can matter is having had exposure to entrepreneurial activities through having a father[2] who was an entrepreneur.
What is also interesting is that the only method of acquiring a business that is associated with higher productivity is joining a family business. Starting your own business, purchasing an existing business and even inheriting a business are not significantly associated with current performance. Some of the effect could reflect a selection effect, individuals are able and more likely to join a family business if it is doing well. But the results could also be understood by the importance of having a role model or mentor in the actually running of a business. It could also be an indicator of networks and the importance of being able to build on existing relationships and contacts in boosting firm performance.
Interestingly, the motivation for becoming an entrepreneur – in pursuit of a market opportunity, the desire for additional income, the desire for greater flexibility or a lack of alternative opportunities – could help explain why some entrepreneurs may put greater effort into their businesses. One hypothesis is that those seeking to exploit an opportunity are the more entrepreneurial and might be expected to be more successful than those who are turning to it out of a lack of alternative opportunities. In fact, the one significant result is that those who would otherwise be unemployed are the ones now operating firms that are relatively more productive. For these entrepreneurs, their business is their source of income and the prospect of failure may not be an option.
The last issue relates to the importance not of more general education or experience, but the use of specific management techniques that have been shown to be associated with better performance in Western Europe, China and India (Bloom et al.). They include having specific performance measures, employee performance incentives, processes for implementing changes, and participatory decision-making procedures. The results show these measures of management quality are robust to the inclusion of the manager’s education and experience and additional firm characteristics. This holds promise for the contribution of additional training in business skills and management training of entrepreneurs in improving firm performance.
Looking at which entrepreneurs are more likely to use these beneficial management techniques, education is again a positive predictor. Prior experience, however, is either not significant or actually negatively associated with setting objectives and monitoring performance.
One further question is whether these other characteristics of the entrepreneur can help explain away the finding that the gender of the entrepreneur matters. As women entrepreneurs are less likely to use a number of the formal management techniques, this could account for their lower productivity. The paper tests where there could be additional non-linearities by gender in the returns to education and management, but finds little such evidence. Including these additional controls does reduce the significance of the productivity gap for women entrepreneurs, although the size of the effect does not necessarily diminish.
The area where gender non-linearities are significant are in the impact of having an existing family business, either with a father having run an enterprise or that a family business existed to join. In both cases, there is a strong positive effect for male-run firms, but no overall effect for female-run firms. These interactions reduce the size and significance of any direct effect of gender on performance. If having the role model or being part of an existing network is important for performance, measures that increase the number of women entrepreneurs now will have longer benefits over time as they can facilitate more women entrepreneurs in the future.
The paper is structured as follows. After a review of the literature, the paper discusses the new survey data that this paper utilizes to refine the hypothesis testing. Section 4 then tests whether the characterization of ‘women’s enterprises’ by ownership or decision making authority matters. Section 5 then examines how experience, education, motivation and management could matter. Section 6 then looks at the types of entrepreneurs that are likely to have had different types of prior experience and motivations to become an entrepreneur. It also looks at who is using the stronger management techniques. Section 7 then discusses the tests of whether there are gender differences – not just directly – but through the impact of education, experience and management. Section 8 concludes.
2. Literature:
This paper looks at two literatures, one examining whether there is a gap in enterprise performance by gender, and one that looks more broadly at whether characteristics of the entrepreneur explain performance. This paper looks at both sets of issues, tying them together to see if the experience, motivation and management techniques used by the entrepreneur can account for differences by gender.
Studies of enterprises in Sub-Saharan Africa that examine productivity and growth have had mixed results as to whether there is a gap by gender. Some of the differences reflect sample composition, but the dimension along with gender is taken into account can matter too. Thus, McPherson (1996) and Mead and Liedholm (1998) find significant gender gaps. Their samples, however, have large representation of household based enterprises that are found to be less productive than more marketplace-based enterprises. The work here is interested in the comparison of full-time, marketplace-based activities in urban areas. The Enterprise Surveys database produced by the World Bank provides a means of examining the issue. Of particular interest for this paper, it uses the same data as other papers that have not found evidence of gap and examines how the dimensions chosen for characterizing women’s enterprises matters.
Bardasi, Blackden and Guzman (2007) show few differences in reported constraints or indicators of firm performance are statistically different between women and men in 24 North and Sub-Saharan Africa countries – once they are in business. Bardasi and Sabarwal (2009) re-examined the issues using 22 countries in Sub-Saharan Africa surveyed in 2006 and 2007. They corroborated the finding that firms with female participation in ownership are as productive and grow as fast as male-only owned firms. They do find some differences in scale, with firms with female participation constrained to be somewhat smaller than male owned firms. They find limited evidence that access to finance could be playing a role; women appear to be better able to access finance although this does not have the same positive effect on performance as it has for male owned firms.[3]
This raises two sets of questions that are relevant for this work. The first is whether the classification of male and female enterprises is along the relevant dimension. They have used ownership as the key dimension, including those with only some female participation in ownership as ‘female enterprises.’ The standard Enterprise Survey data has made it difficult to distinguish ownership from decision-making authority, the hypothesis can be tested in 5 countries with where an additional survey module was run to allow for this to be done. A second set of questions, addresses the issue they raise, of who is successful at getting into business in the first place and how this affects their subsequent performance. Gender is certainly part of this issue, but other factors such as education, motivation for being an entrepreneur and how the business was acquired could all be important in understanding who the entrepreneurs are. Again, the supplemental module to the Enterprise Surveys was designed to be able to shed light on these issues.
Both the Bardasi, Blackman and Guzman (2007) and Bardasi and Sabarwal (2009)papers restrict the firms being examined to try to minimize the difference between ownership and control. They look only at sole proprietorships and partnerships. Except for a handful of cases, sole proprietors do control their businesses. But this is not necessarily true in partnerships. In about 40 percent (varying by country) of partnerships with male and female owners, a women is not involved in the major decisions of the enterprise. The rate almost doubles for limited partnerships. Among partnerships, it is the larger and more productive enterprises that are actually run by the men, making it important whether firms are categorized by ownership or decision-making authority.[4]
Hallward-Driemeier and Shah (2009) find that gender differences in the performance of firms can be found using the same data as Bardasi and Sabarwal (2009) and even using the ownership based definition of male and female enterprises – if one differentiates countries by measures of women’s economic opportunities and by size of firm. Using female literacy, countries are divided into three groups.[5] They find that who becomes an entrepreneur does vary by country-group, with women who are able to become an entrepreneur in low-literacy countries having more education and more access to land than their male counterparts. They also find productivity differentials between men and women in lower-literacy countries, even controlling for education, access to finance and assets. In higher literacy countries, women’s participation is higher, but the effects of education and sector composition are stronger, particularly among larger firms. Men now have relatively more education and are relatively more concentrated in higher productivity sectors. Thus within sector, women perform as well as men, but not conditioning on sector, the concentration of women in traditionally low entry sectors results in a gender gap.
In looking at other characteristics of the entrepreneur that might be important – and that might account for the gender gap – the paper draws on a number of hypotheses that have been shown to be important in other regions. Djankov et al. undertook intensive surveys in Brazil, China and Russia to understand the determinants of who was likely to be an entrepreneur. They found that education mattered. They also found that having a parent who had been an entrepreneur was also a significant predictor. This hypothesis can be tested here in the 5 Sub-Sahara African countries, as well as whether the effect might be greater for sons or daughters and whether it matters if the parent was the father or the mother.
Other work in Asia has shown the significance of the composition of ownership and control in the performance of firms. Bertrand et al. (2004) and Bertrand and Schoar (2006) show that family owned business represent a significant share of the largest firms in many Asian countries. But that such control can raise concerns about succession within the firm, and with longer-run productivity deterioration. In contrast, the firms examined here are often much smaller – and exclude any publicly traded firms. The role of the family in these firms is associated with higher productivity. Succession may raise some issues, but the one highlighted here is that the benefits of the role model and business connections appear to only be passed on to the sons.