What can female headship tell us about women’s well-being? Probably not much


This page in:

In the last week alone, we reviewed two studies that assessed gender inequality by comparing outcomes between female- and male-headed households. This was surprising to us, as it seems an outdated approach. We thought it would be worthwhile to make some observations about why comparing female- and male-headed households tells us very little (if anything) about gender gaps, at least in Africa (if not everywhere). There is more than enough here to argue against comparing households with male and female heads to capture differences in male and female well-being or gender equality.

Maybe the most obvious reason not to study gender (in)equality by comparing households by the gender of their heads stems from the fact that many women live in male-headed households, and this includes many women who are deprived in specific ways. In the latest DHS for 77 countries, on average one quarter of households are headed by a woman, but there’s a lot of variation across countries. The rate of female headship ranges from less than 2% in Afghanistan to 45% in Haiti. Barring very heavily skewed male/female population ratios, women can’t mostly be residing in female-headed households. Indeed, a lot are living in male-headed households. In Cote d’Ivoire, where about 20% of households are headed by a woman, about 70% of women (15 and older) reside in male-headed households. Similar rates are found in Ethiopia, Rwanda, and Tanzania. On the other hand, far fewer men live in female-headed households in these countries (around 10%). Also, men are rarely the spouse of the household head (arguably alleviating concerns of varying social norms across countries in regards to gender and assigning headship). So unlike for women, male headship captures socio-economic conditions for the vast majority of men.

Why is it problematic to ascribe differences between male- and female-headed households as an indicator for gender (in)equality when two thirds of women reside in the former? Because it assumes that resources are shared equally among members in households. This might not be a problem for indicators like dwelling characteristics (which arguably do not vary across household members), but it is a problem for household measures of consumption and income which get divided up among members (and poverty measures derived from them). For example, a recent paper finds that a majority of Africa’s nutritionally deprived women and children are not found in poor households as conventionally defined. For over 25 years, the unitary model of households in economics has been questioned (see this paper, just one of many examples). We still struggle with measuring well-being within households. Chapter 5 in the World Bank Poverty and Shared Prosperity (PSPR) report from 2018 discusses the existing literature and new methods to measure intrahousehold differences in resource allocation, both from a monetary and non-monetary perspective.  

A second problem lies in the heterogeneity in marital status among female-headed households. In sub-Saharan Africa, for example, over 80% of male heads are married. Among those who are not, most have not yet married. On the other hand, the women who head households are much more heterogenous with respect to marital status than men. In Africa, widows make up the largest segment of female heads (about 45%). Divorced women are about 20% and the never married are about 10%. And close to a quarter of the women who are female heads in the region are actually married with a non-resident husband. The husband has either migrated or is non-resident in a polygamous marriage. Implicit in comparing female- and male-headed households to judge gender gaps is the assumption that the former have no or weak links to male-owned or controlled incomes and resources. This is unlikely the case for women who are household heads and have husbands living elsewhere, with their potential income from remittances and access to other resources via their husbands. Consistent with this, in Africa female heads who are married have lower poverty rates than female heads who are not married. Rather than focus on sex of the household head, it would be better to focus on the specific attributes of the head that make some female-headed households poorer (such as widowhood) or wealthier (migrant husbands).

A third problem to be aware of in the way male- and female-headed households are being compared relates to how we measure the economic welfare of people. The use of per capita consumption/income in making poverty comparisons between female- and male-headed households ignores economies of scale in consumption. Female-headed households have, on average, a different demographic composition to their male peers. Specifically, they are smaller and have higher dependency ratios. This raises concerns about whether findings based on per capita income or consumption are robust to different assumptions about scale economies and adult equivalence. Globally, female-headed households are on average less poor using a per capita measure (see the above PRSP) and this is true in Africa also where there is a growing share of female-headed households happening alongside falling poverty rates (see this paper). But controlling for household size (to allow for economies of scale), female headship is associated with significantly lower consumption for all but one of the 9 countries examined from Africa (look for the paper out soon!). And allowing for even modest economies of scale in consumption can result in a reversal in the poverty comparison, as this work finds for East and Central Africa. Current measurement practices may not even be reliable for telling us whether female-headed households are in fact poorer.

We need to go beyond simple male- and female- headship comparisons, and toward a richer typology of households, taking account of marital status, demographics (such as here), income sources, and (as much as possible) individual well-being.



Kathleen Beegle

Research Manager and Lead Economist, Human Development, Development Economics

charles gilman
June 11, 2019

These are great points about what comparing FHH and MHH isn't good for and I hope that your point of going beyond household surveys is appreciated by all. But are there questions where comparing FHH and MHH might be useful (even if not optimal)? Perhaps customary land titling means widows and divorced women are deprived land rights and that community decision making, for example through water user associations, are ascribed through land titles. And so in this case the head of household does confer a sort of public good to the rest of the household. Maybe there are other examples? Access to markets? Thanks for the blog- Cheers

June 24, 2019

I have been thinking about this exact question! And I can't think of relevant examples. One would likely still want to look at household composition to understand deprived land rights etc. (is it the presence of an adult male that brings access etc - regardless of whether that person gets a label of head).
A point we did not raise but that is also relevant is that across surveys (both within and across countries), there is not consistency is how headship is defined. So in one context, it might go to the elder male, in another the prime age male. Another reason to avoid the headship classification.

Danny Gotto
June 11, 2019

This means that we should encourage women to remain married and those widowed should be protected to remain in charge of the family assets.

June 24, 2019

Hmmm, I would not infer this from our blog. Our blog was not intended to speak to the socioeconomic/other benefits/costs of marriage or widowhood for women.

Fred Dzanku
June 11, 2019

Supposed that we compared woman and men rather than male versus female headed households, would the confoundedness of heterogeneity go away? No. Just wanted to make the point that while we should do much better than comparing male and female headed households, the challenge of heterogeneous units would not necessarily disappear.

June 24, 2019

Thanks for this comment. I don't think the confoundedness of heterogeneity is what the blog touches on. But maybe I am not entirely sure what you mean.

July 13, 2020

A very interesting article. I have two more imagined analytical problems in comparing FHH and MHH.
First is the impact of the transition of marital status as a shock i.e. MHHs to FHHs. For example, the economic welfare of a widow or a divorcee during the first year might be different in about 5 years later. Do you consider the duration marital status especially in the case of FHHs?
Secondly, in some cases especially among low-income households in Africa, divorced women, or those who never got married tend to stay with their parents mostly in male-headed household settings. How would you handle these mixed female-male-headed households?