In a 2009 paper, David McKenzie and coauthors Chris Woodruff and Suresh de Mel find that giving cash grants to male entrepreneurs in Sri Lanka has a positive and significant return, while giving the same to women did not. David followed this up with work with coauthors in Ghana that compared in-kind and cash grants for women and men. Again, better returns for men (with in-kind working for some women). Taken together, these paper (with different, thoughtful explanations in each) could lead you to question the impacts of simply giving capital to female entrepreneurs.
A thought provoking new paper from Arielle Bernhardt, Erica Field, Rohini Pande, and Natalia Rigol advance our thinking on this and also complicates things. Going back to the old-school idea that households might be trying to jointly maximize income, Bernhardt and co. argue that investment will be in the household enterprise with the higher returns. And this could be a male-owned enterprise.
Bernhardt and co. look at this question using data that some of their authors have worked on (from a microfinance experiment in Kolkata), but also pulling in the data from the two experiments mentioned above that David was involved in. Their model of the household leads them to the question: what happens to male enterprises in a household when you give cash to the female?
To start with, more women in these three samples live with entrepreneurs than men do. In the Kolkata sample, over half of the women live with another microentrepreneur, and this is true for 48 percent of the women in Sri Lanka and 41 percent in Ghana. For men, on the other hand, 29 percent in Sri Lanka and 26 percent in Ghana live with another person who is classified as self-employed.
Let’s take the results country by country. In Kolkata, the sample is drawn from a randomized experiment where one group gets regular microfinance (where repayment starts after 2 weeks) and another group gets loans with a two-month grace period. For their analysis, Bernhardt and co. are focusing on two groups: those where the female client is the only entrepreneur in the house, and those where there is another entrepreneur and he is (almost always) a male. Before getting to the analysis, it’s interesting to see the structure of the female enterprises in these two groups: when the woman is the sole entrepreneur, her enterprise profits are higher and the sectoral choice mimics that of men. When a woman is not the only entrepreneur, profits are lower and the sector is different.
Bernhardt and co. take this classification to the data. First, there are no significant treatment effects (of the grace period) on average for female entrepreneurs. However, for women who are the sole entrepreneur in the household, there are significant treatment effects: weekly profits are 70 percent higher than women in the control group. For women in households with other enterprises: no significant effect. There is, however, a treatment effect on the other enterprises in the household: they go up by 50 percent relative to the control. So, in these cases the positive capital shock is being passed on.
Now, let’s move over to Sri Lanka and a cash grant for enterprises. Here again the average effect is zero. However, for women in households where they have the only enterprise profits increase by 30 percent (significant at 10 percent). And for all households Bernhardt and co. find a significant increase in total household income of 8 percent (a footnote explains that their sole/multiple enterprise regression is underpowered – they find (insignificant) results of +5 percent for sole entrepreneur households and +8 percent for multiple entrepreneur households).
In Ghana, the intervention consists of cash grants versus in-kind grants. The neat thing here is that the grants went to both men and women, which allows them to compare the effects across sexes. First off, the in-kind grants show some significant effects, while the cash do not. For women in sole enterprise households, they cannot reject the hypothesis that the in-kind grant effect is equal to men in multiple enterprise households. For women in multiple enterprise households, the effect is lower, and they can reject (at 10 percent) that it is equal to the effect for men. So women who are the only entrepreneur seem to benefit as much as men, but not so women in multiple enterprise households.
At the minimum, these intriguing results mean that we need to measure impacts for the whole household when we look at programs that seek to help programs grow – especially for programs that give something that is easily transferable such as enterprise capital. On a deeper level, these results point us to think harder about how households are making investment decisions. Indeed, this paper raises a plethora of super interesting questions as to what could be going on. Stay tuned – in two weeks I’ll be back with a look at a paper which sheds a bit of light on how the nature of household relations matter for these kinds of outcomes.
A thought provoking new paper from Arielle Bernhardt, Erica Field, Rohini Pande, and Natalia Rigol advance our thinking on this and also complicates things. Going back to the old-school idea that households might be trying to jointly maximize income, Bernhardt and co. argue that investment will be in the household enterprise with the higher returns. And this could be a male-owned enterprise.
Bernhardt and co. look at this question using data that some of their authors have worked on (from a microfinance experiment in Kolkata), but also pulling in the data from the two experiments mentioned above that David was involved in. Their model of the household leads them to the question: what happens to male enterprises in a household when you give cash to the female?
To start with, more women in these three samples live with entrepreneurs than men do. In the Kolkata sample, over half of the women live with another microentrepreneur, and this is true for 48 percent of the women in Sri Lanka and 41 percent in Ghana. For men, on the other hand, 29 percent in Sri Lanka and 26 percent in Ghana live with another person who is classified as self-employed.
Let’s take the results country by country. In Kolkata, the sample is drawn from a randomized experiment where one group gets regular microfinance (where repayment starts after 2 weeks) and another group gets loans with a two-month grace period. For their analysis, Bernhardt and co. are focusing on two groups: those where the female client is the only entrepreneur in the house, and those where there is another entrepreneur and he is (almost always) a male. Before getting to the analysis, it’s interesting to see the structure of the female enterprises in these two groups: when the woman is the sole entrepreneur, her enterprise profits are higher and the sectoral choice mimics that of men. When a woman is not the only entrepreneur, profits are lower and the sector is different.
Bernhardt and co. take this classification to the data. First, there are no significant treatment effects (of the grace period) on average for female entrepreneurs. However, for women who are the sole entrepreneur in the household, there are significant treatment effects: weekly profits are 70 percent higher than women in the control group. For women in households with other enterprises: no significant effect. There is, however, a treatment effect on the other enterprises in the household: they go up by 50 percent relative to the control. So, in these cases the positive capital shock is being passed on.
Now, let’s move over to Sri Lanka and a cash grant for enterprises. Here again the average effect is zero. However, for women in households where they have the only enterprise profits increase by 30 percent (significant at 10 percent). And for all households Bernhardt and co. find a significant increase in total household income of 8 percent (a footnote explains that their sole/multiple enterprise regression is underpowered – they find (insignificant) results of +5 percent for sole entrepreneur households and +8 percent for multiple entrepreneur households).
In Ghana, the intervention consists of cash grants versus in-kind grants. The neat thing here is that the grants went to both men and women, which allows them to compare the effects across sexes. First off, the in-kind grants show some significant effects, while the cash do not. For women in sole enterprise households, they cannot reject the hypothesis that the in-kind grant effect is equal to men in multiple enterprise households. For women in multiple enterprise households, the effect is lower, and they can reject (at 10 percent) that it is equal to the effect for men. So women who are the only entrepreneur seem to benefit as much as men, but not so women in multiple enterprise households.
At the minimum, these intriguing results mean that we need to measure impacts for the whole household when we look at programs that seek to help programs grow – especially for programs that give something that is easily transferable such as enterprise capital. On a deeper level, these results point us to think harder about how households are making investment decisions. Indeed, this paper raises a plethora of super interesting questions as to what could be going on. Stay tuned – in two weeks I’ll be back with a look at a paper which sheds a bit of light on how the nature of household relations matter for these kinds of outcomes.
Join the Conversation