Today, I am going to address another consultation for the United Nations High Level Panel on Women’s Economic Empowerment, this one on women and entrepreneurship. I plan to focus on a specific problem: the common practice of relying on conventional business measures for assessing programs aimed at benefitting women’s entrepreneurship.
I will base my comments on ten years of experience assessing the impact of such programs, beginning with the Avon in Africa study we started in 2007 and ending with the Walmart Empowering Women Together program this year. I am putting links to the published work we have done in this area at the bottom of this post.
Many insist that investments in women’s entrepreneurship should be measured by things like sales, profitability, and hiring. These are all important things to track. But it is sometimes asserted that only these”hard” measures matter. What is also sometimes claimed is that business measures are somehow superior in their concreteness, clarity, or objectivity than other metrics. Over the past ten years, however, I feel I have learned that using only business measures leads to some very unfortunate and even dangerous problems in assessing programs.
So, I have made a list of seven good reasons why we need to assess women’s entrepreneurship support programs with other metrics besides the business outcomes. Here they are:
Business measures mask gender bias. If we are going to invest with a “gender lens,” then we need to evaluate with one, too. Studies consistently show that women-owned enterprises “underperform” relative to those owned by men. The reasons why are equally consistent: women bear a greater burden of family responsibilities, women have less access to capital, and women don’t work in the high growth industries (usually because those industries are hostile to women, as in the tech industry). Therefore, assessing programs using only business measures automatically reproduces a gender bias, especially if decisions will be made based on a perceived “ROI” relative to investing in men, as is often the case.
Women may not be able to retain control over earnings, even in a business. Thus we cannot assume that just because the business is performing, the woman who owns it is benefitting.
The performance of the business may reflect “empowerment” of the owner, but there may be disempowerment of female employees or suppliers in the same enterprise. Not all women are empowered by the same actions; indeed, their interests may be in conflict. The fact that a woman-owned business is hiring females does not mean that those employees are being paid fairly, managed compassionately, or treated with respect.
Respondents often misunderstand or misrepresent answers to business questions. We tend to assume that everyone understands what we mean when we ask “what were your sales last year,” but my experience (and that of others in the field) shows that there are many differences in local understanding of what constitutes income, what period is covered by “a year,” and many other “basic” business metrics. Further, respondents are often loathe to answer such questions: they may refuse or distort the answers. Far from being the unambiguously “hard” measures we think they are, the business data may actually be as “soft” as any other metric collected.
Business measures cannot be compared across programs. Investors, donors, and researchers assessing the relative value of various approaches to women’s empowerment (from girls’ education to mobile savings) need to be able to compare the results across efforts. If you only use business measures to represent your effort, there is no way to compare it to other types of program.
Business outcomes may vary too much to be meaningful across a small sample. In the Walmart Empowering Women Together program, we saw how the surrounding market infrastructure and ethic made a huge difference in the way women entrepreneurs were able to manage their processes. Thus, an owner in Kenya was at a vast disadvantage relative to one in the US. Of course, there are also differences in women’s rights among markets and that affects outcomes, too. Further, if there is a lot of variation in the type of business represented, the differences in organisational structure and industry potential will distort the overall outcome. (For instance, if you are selling office equipment to businesses, one new customer makes a big difference. If you are a hairdresser, one new customer is not that big a deal.) So, if you are measuring the increase in the number of clients or customers, there would be a weirdness in reporting a small dataset with a lot of different kinds of businesses.
Using only business measures raises a worrisome question about intended beneficiaries. The rhetoric around the benefits of women’s economic empowerment promises growth. That’s a good reason to support this movement. But if you only measure growth, then you run the risk of producing outcomes that benefit investors and tax collectors, but not the women who own or work in the business. These measures could potentially obscure exploitation. We need to remember that if the beneficiaries of such programs are the women, then we need to make sure that they do, indeed, benefit.
Avon in Africa, 3 year study of the Avon system in South Africa: articles and other materials
CARE Bangladesh Rural Sales Program, now “Jita”: articles and blogs about the research
Advisory Note on Measures: Women’s Economic Empowerment