The glass ceiling, even at home: How a report on poverty in Nigeria reveals an enduring gender gap in household headship

Summary

We are told by the Nigeria National Bureau of Statistics [1] that female heads of household in Nigeria are less likely to be poor than male heads of household. It’s a provocative claim, grounded in truth, and unlikely to square with the experience of women there. Our analysis shows that among heads of household, women in Nigeria are in fact poorer than men. That is, female heads of household have higher rates of poverty, taking account of other determinants of poverty.

This factoid is an example of Simpson’s paradox, whereby a correlation that is negative in the population as a whole (say, between poverty and the gender of heads of household) can be positive within every subgroup of a population, such as small households, midsize households, and large households.

When we look at all the households in the country, those with female heads of household appear to have a lower overall rate of poverty, as reflected in the report [1]. The intuitive explanation is that women must be better off than men in the aggregate, but it does not hold up to scrutiny. If we look instead at the rates of poverty conditional on the main determinants of poverty, it turns out that female heads of household have higher rates of poverty than their male counterparts. In the aggregate, women appear to be doing better – but at a comparable household size, men are less likely to be poor.

Although it may be true that women are less likely to be heads of households in poverty, the implication of that is not that women are better off than men. It is that women are systematically absent from heading the largest – and therefore the poorest – households in Nigeria. It is not yet time to take a victory lap on behalf of Nigerian women.  At best, the Nigeria National Bureau of Statistics’ draft report on poverty among female-headed households could be a milestone on the road to economic empowerment for women in Nigeria. The dearth of female heads of household in Nigeria’s poorest demographics could nonetheless turn out to be significant for policymakers, if only by revealing the geographic limitations of poverty reduction efforts that target female heads of household.

The claim

Female-headed households suffer poverty at just two-thirds the rate of male-headed households in Nigeria, so goes the claim [1]. The result holds true at multiple levels of education and for most livelihoods, and both in cities and rural zones.

Encouraging as this good news is, none of the statistics that follow should reassure us that women are better off as a result. The statistics in the report detail differences in poverty rates that are higher for male-headed households among the least educated (66 of male-headed households vs 34% of female-headed households) and postgraduates (18 of male-headed households vs 6% of female-headed households). Similarly, more male-headed households are poor among the rural with agricultural incomes (63 of male-headed households vs 40% of female-headed households) and urban wage workers (12 of male-headed households versus 11% of female-headed households) and even among those not working at all (34 of male-headed households vs 24% of female-headed households).

Is it plausible that female-headed households are less poor?

The sleight-of-hand that produces these happy statistics has two components. First, the proportion of households that are female-headed is nowhere near 50%. Second, the gender of the head of household is not an exogenous variable in the composition of household structure. Households are not created with male or female heads and then left to get what they can in the marketplace. Female heads of household are different from the “average” household not because they are female, but because of what leads these women to become heads of household. In short – they are not married.

There is a telling omission in the summary tables comparing poverty rates between female and male heads of household; Nigeria National Bureau of Statistics included only rural vs. urban comparisons for “Table 7. Poverty Headcount Rate by Household Size.” The bureau chose not to include female household heads here, too, as they did in Tables 5 and 6 concerning educational attainment and occupation. Larger households in Nigeria are defined to be those that have more than 20 individuals. Their rate of poverty is more than 70% in both urban and rural locales, whereas just 3% of individuals (household of size 1) are poor and only one-sixth of households of size two to four.

As it turns out, women are more likely to head smaller households. These households suffer poverty at lower rates. But within each household size, those with female heads are more likely to be poor, at least as recently as 2010. (We are looking forward to the 2019 microdata becoming available to perform a similar analysis.) Nationwide, the rate of poverty is higher for male heads of household. But within comparable household sizes, it is the other way around – female-headed households are poorer.

The poverty rates in the Nigeria National Bureau of Statistics summary are conditional on gender-and-education, or conditional on gender-and-occupation. We do not yet know the share of poor heads of household who are female. We also do not yet know the distribution of educational attainment among poor, female heads of household versus male. Lastly, we do not yet know the labor participation rates of poor, female heads of household versus male. These will all become available with the microdata.

What we do know is that large households are many times more likely to be poor than small households. Therefore, we might want to determine whether it is also true that large households are equally as likely to have female heads as male. They are not.

Female heads of household are likely to belong to households that are smaller, Southern, urban, divorced, and widowed. Each of these dimensions reduces the incidence of poverty regardless of the gender of the head of household. The economics literature on the determinants of poverty can tell whether, after controlling for these factors, female heads of household still suffer poverty at a higher rate than male heads of household.

What’s really going on: why female-headed households are actually poorer

The question of whether female heads of household are better off than their male counterparts depends greatly on everything else: the size of the household, the occupation and wealth of the household head, education, location, and other factors all at once.

The story of poverty in Nigeria since 1980 has not been a happy one. Professor John C. Anyanwu, the former lead research economist at the African Development Bank, found that poverty among male heads of household increased from 27 to 63% in the period 1980-96 [2]. Most of the increase occurred in just the last four years of the period. Using the same data from the National Integrated Survey of Households, CGD Fellow Zuhumnan Dapel showed that the trend continued to 2010 when the national poverty headcount hit 69% [3]. Since then there has been a considerable improvement in the overall level of poverty, with 40.1% of households indicated as poor in the 2019 survey [1].

In order to understand whether women (female-headed households and non-heads of household) are more or less poor, we might be more interested in the gender distribution of heads of household at various sizes, and the poverty rates within each group.

In 1980, female heads of household were more likely to be poor at each household size than their male counterparts. The same was true in large households as of 1985 and all households in 1992. Then something curious happened in the 1990s in the very largest households. By 1992, the poverty rate had fallen 60 percentage points in just seven years (from 100 to 40%). In 1996, no estimate was reported for poverty among female-headed households with more than 20 individuals. We suspect that large households with female heads had become vanishingly rare. Over the same period, the rates of poverty among male-headed households in every size bracket increased. Over the same period, the rates of poverty among female heads of household in small and mid-size households increased dramatically. Small households (two to four) saw an increase from 19 to 54% over the period 1985-96. Medium-size households (five to nine) saw a comparable increase from 50 to 81%. Meanwhile, among both male- and female-headed households, household size, and education were critically important determinants of household poverty.

Determinants of poverty 

When we see statistics about the rate of poverty among male and female heads of household, we assume that these describe whether men and women have an equal chance of being poor in the economy today. But the rates of poverty among male- and female-headed households tell us no such thing. Married heads of household, who have a lower chance of poverty than unmarried heads of household, overwhelmingly identify males as head of household. Divorced and widowed heads of household are less well-off than married heads of household, but still better off than unmarried heads of household [4]. An econometric regression of poverty on the other features of the household can tell us whether, at the margin, it is the marital status of the head of household, the size of the household, or the heads of household’s gender that is correlated with poverty.

Although no one has written more about this than Anyanwu, he has found contradictory results on gender at different times, using large, national datasets from the National Integrated Survey of Households. Anyanwu studies rates of poverty among households, using headcount measures at the national poverty line. He found that gender (of head of household), in and of itself, was unrelated to poverty in Nigeria during the 1990s after controlling for other factors [2]. In that paper, he found that age and education tended to reduce poverty more among female heads of household than among male heads of household. In 2013, he found that gender (of head of household) was unrelated to poverty [5] – but his 2014 paper, using a 2009-10 vintage of National Integrated Survey of Households data – found that female heads of household are more likely to be poor, all else equal [4]. In this last study, the effect of gender on poverty status is tested after controlling for household size, location, education, employment, and marital status.

In other words, after all the other demographic and economic factors are accounted for, female heads of household still faced a greater chance of poverty in 2010.

Dapel uses the same National Integrated Survey of Households datasets as Anyanwu, but with a poverty dynamics lens rather than a focus on static, cross-sectional poverty [3]. Dapel analyzes both temporary and chronic poverty. The effect of gender (of head of household) on poverty is not large, but is statistically significant in both cases, and women are more likely to be poor.

Where are all the female heads of household? 

In order to know what this means for Nigerian women, we must consider all the possibilities. Will a woman spend most of her life happily married, divorced with children, single without children, widowed and head of household, or as a member of a household headed by someone other than her spouse? Married women rarely report as head of household – a designation which, incidentally, the United States current population survey abandoned in 1980 for married couples. The USAID Demographic and Health Surveys, by contrast, still use the designation and calculate the fraction of female-headed households.

According to the World Bank, female-headed households account for just 15% of Nigerian households. Since 1990, the National Population Commission has estimated that the fraction of female-headed households is in the range of 15 to 19%, according to the Demographic and Health Surveys Program. Moreover, the proportion of households with female heads is lower in the rural sector (15 female heads among rural households vs. 22% female heads among urban households).

Systematic differences exist in household size and composition between male- and female-headed households. Nigeria’s share of households with female heads is 19%, but those households have just 13% of the population [6]. The disparity in household size is common to urban and rural populations, with female-headed households accounting for 15% of rural households and just 10% of the rural population. For urban locales, female-headed households have 23% of households and 17% of population.

In 1994, the Demographic and Health Surveys (DHS) program found that one-quarter of female-headed households had no dependents, and another one-quarter had one adult and dependent children. Yet among male-headed households, less than one-tenth had no children, and less than 2% had a single adult with dependent children. That implies that the households with no dependents and a single adult were about twice as likely to have a male head, whereas those with dependents and a single adult were about three times as likely to have a female head. More than half of female heads of households were over age 50.

Low poverty rates among female heads of household should be interpreted as an absence of female heads of household among the poor, not a low incidence of poverty among female-headed households. What happens to poor women who should naturally become heads of household? Why are there effectively no female heads of household in the poorest of Nigeria’s demographics? Is it true that females cannot become heads of large households in the poorest regions of Nigeria, and why?

One hypothesis is that female heads of household intentionally seek to have a male relative or dependent declared the head of household, whereas widowed male heads of household retain their status as head of household. This would be worthy of further study.

A second hypothesis is that dependents in large households with female heads calculate that they will be better off in a different household. The avenues by which they can leave are several: create a new household, marry out, move away for work or education, or seek to join another household in the extended family such as a widowed mother joining her adult son’s household. Large households with female heads may shrink or even break up. Here again, this is a hypothesis for further study – no evidence is presented in support of these dynamics.

Further readings

Other empirical papers may be of interest to economists. Contrary evidence that female heads of household are less likely to be poor is found by Oginni, Ahonsi, and Ukwuije, using the Nigeria Demographic and Health Survey 2008 dataset [7]. Supporting evidence for higher poverty among female heads of household is the subject of a study of urban households in Nigeria by Akerle, Momoh, Adewuyi and co-authors [8]. Evidence from a study of 285 rural households in Ondo State showed higher headcount rates of poverty among female heads of household than their male counterparts, by Iglbalaljobe and Fatuase [9]. Multidimensional poverty rates were higher among women in the 2010 National Integrated Survey of Households survey, focusing on rural households, in a paper by Adeoti [10]. For a discussion of the concepts and measures of poverty, see Olatomide Olowa’s discussion of Nigeria in perspective [11]. An earlier UN WIDER discussion paper serves as a precursor to more recent empirical works [12]. Systematic reviews of poverty are rare outside of a specific policy intervention, but Jung, de Bairros, Pauli, and Neutzling offer a global, systematic review of the determinants of household food insecurity that is tangentially related and contains one empirical paper on Nigeria [13]. On the topic of the feminization of poverty, Sylvia Chant gives a review of decades of discourse, which seems to be at the heart of our collective interest in this problem [14].

A word about poverty

The headcount rate of poverty is defined as a fraction of households that can maintain expenditures equal to or greater than the “cost of basic needs,” on a per capita basis, taking account of the age and sex of the members of the household. In the Nigeria National Bureau of Statistics analysis of the 2019 national household budget survey [1] that is the occasion for this essay, the national poverty line is equal to NGN137,430 per person per year.

Poverty can also be measured by other indicators. The poverty gap measures how far beneath the poverty line is the average poor household’s expenditure. According to the Nigeria National Bureau of Statistics, the current poverty gap in Nigeria is 12.9%. The multidimensional poverty index measures health, education, and living standards, rather than income or consumption expenditure. The national poverty line in Nigeria is less stringent than metrics produced by international bodies. The national poverty headcount in 2009 was just 46%, whereas the World Bank estimated that in 2011, 53% of the country fell below $1.90/day and 78% fell below $3.10/day per capita income [14]. More recently in 2017, the Oxford Poverty and Human Development Initiative found that 51% of the population suffered multidimensional poverty [14]. The initiative shows further that poverty is heavily concentrated in the north, where some regions have more than 50% “severe” poverty, falling below the poverty line on more than half of the weighted indicators in the multidimensional poverty index.

The focus on headship in this paper should not diminish our concern for women’s wellbeing in male-headed households, where women suffer higher rates of poverty and food insecurity than the rest of the household [16]. A small but growing empirical literature on female-headed households in Nigeria suggests that in those households, women suffer less disparity of health, food security, and poverty [17-18]. We would be thrilled to read that women and men in Nigeria had equally low odds of facing poverty as heads of household. That day has not yet arrived.

Reference list

1. National Bureau of Statistics (NBS). 2019 Poverty and Inequality in Nigeria: Executive Summary. 2020;(May):1–27.
2. Anyanwu JC. Poverty in Nigeria : A Gendered Analysis. J Stat Africain. 2010;(11):38–61.
3. Z. Dapel, 2018. Nigeria: The trapped, the freed, and the never trapped. Center for Global Development. CGD Working Paper 485. June 2018.
4. Anyanwu JC. Marital Status, Household Size and Poverty in Nigeria: Evidence from the 2009/2010 Survey Data. African Dev Rev [Internet]. 2014 Mar 1 [cited 2020 Jun 17];26(1):118–37. Available from: http://doi.wiley.com/10.1111/1467-8268.12069.
5. Anyanwu JC. The correlates of poverty in Nigeria and policy implications. African J Econ Sustain Dev [Internet]. 2013 [cited 2020 Jun 17];2(1):23–52. Available from: https://www.researchgate.net/publication/264823947.
6. Milazzo A, van de Walle D. Women Left Behind? Poverty and Headship in Africa. Demography. 2017 Jun 1;54(3):1119-45.
7. Ayodeji Oginni, Babatunde Ahonsi, Francis Ukwuije (2013). Are female-headed households typically poorer than male-headed households in Nigeria? The Journal of Socio-Economics. Volume 45 , Pages 132-137, 2013.
8. Akerele, D., Momoh, S., Adewuyi, S.A., Phillip, B.B. and Ashaolu, O.F. (2012), “Socioeconomic determinants of poverty among urban households in South‐West Nigeria”, International Journal of Social Economics, Vol. 39 No. 3, pp. 168-181. https://doi.org/10.1108/03068291211199341.
9. O. Igbalaljobe, AI Fatuase et al., 2013.
10. ADEOTI, Adetola I. (2014) “Trend and determinants of multidimensional poverty in rural Nigeria” Journal of Development and     Agricultural Economics Vol.6(5), pp. 220-231, May 2014.
11. Olowa, Olatomide. (2012). Concept, Measurement and Causes of Poverty: Nigeria in Perspective. American Journal of         Economics.  2. 25-36. 10.5923/j.economics.20120201.04.
12. Okojie CE. Gender and education as determinants of household poverty in Nigeria. Wider discussion paper; 2002.
13. Jung NM, de Bairros FS, Pattussi MP, Pauli S, Neutzling MB. Gender differences in the prevalence of household food insecurity: a   systematic review and meta-analysis. Public health nutrition. 2017 Apr;20(5):902-16.
14. Sylvia Chant (2008) The ‘Feminisation of Poverty’ and the ‘Feminisation’ of Anti-Poverty Programmes: Room for Revision?, The   Journal of Development Studies, 44:2, 165-197, DOI: 10.1080/00220380701789810.
15. OPHI Oxford Poverty & Human Development Initiative Global MPI Country Briefing 2019: Nigeria (Sub-Saharan Africa) [Internet].   [cited 2020 Jun 17]. Available from: www.ophi.org.uk.
16. Haddad L, Hoddinott J, Alderman H, International Food Policy Research Institute. Intrahousehold resource allocation in developing   countries: models, methods, and policy. Baltimore Md.; London: Johns Hopkins University Press; 1997.
17. Aromolaran AB. Household income, women’s income share and food calorie intake in South Western Nigeria. Food Policy. 2004   Oct  1;29(5):507-30.
18. Oni OA, Fashogbon AE. Food Poverty and Livelihoods Issues in Rural Nigeria. African J Agric Resour Econ [Internet]. 2013 [cited   2020 Jun 29];8(2):108–35. Available from: https://ageconsearch.umn.edu/record/160631.