Polygamy and poverty are both widespread in sub-Saharan Africa.1 Several arguments have been made suggesting this correlation is causal. Scholars have suggested, for example, that polygamy crowds out productive investment (Tertilt 2005), reduces investment in girls’ education (Edlund and Lagerlöf 2006) or diminishes the labour supply of unmarried men (Edlund and Lagerlöf 2012).
A ‘polygamy belt’ stretches across Africa, from Senegal through to Tanzania, in which it is not uncommon for a third of married women to share their husbands (Jacoby 1995). In the present day, this prevalence is almost unique to Africa. All Demographic and Health Survey (DHS) data sets report that at least 92% of married women are monogamous – except for those from Haiti and sub-Saharan Africa.
Less well known is the rate at which polygamy has declined in Africa in recent decades. In Benin, Burkina Faso, Guinea, and Senegal, more than 60% of married women recorded in DHS data in 1970 were polygamists at the time they were surveyed. For women married in 2000 in these countries, the polygamy rate is less than 40%. Several other countries with DHS surveys have experienced similar erosions of polygamy.
In recent research (Fenske 2013) I use DHS data on 494,157 women from 34 countries to document these trends over space and time. I then test nine hypotheses about these patterns. These hypotheses are motivated by a simple model, and by previous theories and findings from economics, anthropology, and African history. I test whether polygamy responds to economic incentives, to economic shocks, and to the process of economic development.African polygamy over space
I map African polygamy in Figure 1. Each point is a married woman for whom the DHS data gives geographic coordinates. Red dots are polygamists, while blue dots are monogamists. Polygamy is most concentrated in West Africa, though it is by no means limited there. Polygamy in the data is largely bigamy – 72% of respondents report that they are the only wife, 19% report that their husband has two wives, 7% report that he has three wives, and fewer than 2% report that he has 4 wives or more.
Figure 1. Polygamy in Africa
The three hypotheses I test about the spatial distribution of polygamy relate to geographic, historical, and cultural variables that are slow to change. First, Jacoby (1995), building on Boserup (1970), has linked differences in the demand for wives across regions of the Ivory Coast to the productivity of women in agriculture. I find, by contrast, that polygamy is least common in those parts of Africa where women have historically been most important in agriculture. Second, economists since Becker (1974) have argued that polygamy can only exist where there is inequality between men. I am not able to find any correlation between wealth inequality recorded in the DHS and the probability that a woman is polygamous. I find, however, that historical inequality predicts polygamy today. Similarly, geographic predictors of inequality that have been used in other studies also predict the existence of polygamy in the present. Third, I confirm the result of Dalton and Leung (2013) that greater slave-trade exposure predicts polygamy today. I add the caveat, however, that the result stems largely from the broad contrast of West Africa to the rest of the continent.African polygamy over time
I show the decline of polygamy over time for a selection of countries in Figure 2. A raw correlation between year of birth and polygamy will confound time trends with age effects, since a young lone wife may later become a polygamist’s senior wife. Thus, I estimate time trends by using countries that have multiple DHS surveys, regressing polygamy on quartic polynomials in age and year of birth.
Figure 2. Predicted polygamy by year of birth for 30-year-old women
I calculate the predicted probability that a woman aged 30 is polygamous as a function of her year of birth. I present sample trends in Figure 2. In the paper, I show similar trends for all countries in which the data permit this. Though the speed of the decline has differed across countries, its presence has been widespread.
The six hypotheses I test that have the potential to explain these trends examine time-varying factors that may influence polygamy. First, I exploit two natural experiments that have increased female education in Nigeria and Zimbabwe, and find no causal effect of women’s schooling on polygamy. By contrast, I use data on colonial teachers and missions from Huillery (2009) and Nunn (2011) to show that areas that received more colonial-era educational investment show lower levels of polygamy in the present.
Second, I use exogenous variation in country-level rainfall to predict country-level incomes during the period of a woman’s adolescence that typically precede her marriage. These in turn predict lower rates of polygamy. Third, I show that a similar pattern exists at the local level. Adverse local rainfall shocks experienced in a woman’s adolescent years make her more likely to marry a polygamist. Fourth, conflict exposure at the local level acts like a detrimental rainfall shock, increasing the prevalence of polygamy among adolescent girls exposed to it. Both of these effects, however, are small.
Fifth, I use a regression discontinuity design to test for breaks in the prevalence of polygamy across national borders. With a few exceptions, I find that polygamy largely passes smoothly over borders, demonstrating that national bans and other policies have been mostly ineffective. Last, I use a country-level differences-in-differences approach and a natural experiment from Uganda to test whether declining child mortality predicts reduced polygamy rates, and find a large effect.Conclusions
These results pose challenges to existing theories of polygamy. The distribution of polygamy in Africa does not fit an explanation rooted in the gender division of labour. I find no evidence that educating women in the present reduces polygamy. Further, I find that history matters. Pre-colonial inequality, the slave trade, and colonial education all predict polygamy rates in the present. I find limited evidence that African marriage markets have responded to economic growth and fluctuations. The largest elasticities that I find are in response to changes in child health. This is consistent with theories that see polygamy as a strategy for men to increase fertility, making wives and surviving births per wife substitutes.Bibliography
Becker, G (1974), “A Theory of Marriage: Part II”, in Theodore W Schultz (ed.), Marriage, Family, Human Capital, and Fertility: 11–26.
Boserup (1970), Woman’s role in economic development, New York: Martin’s Press.
Dalton, J and T Leung (2013), “Why is polygyny more prevalent in Western Africa? An African slave trade perspective”, forthcoming in Economic Development and Cultural Change.
Edlund, L and N-P Lagerlöf (2006), “Individual versus parental consent in marriage: Implications for intra-household resource allocation and growth”, The American Economic Review 96(2): 304–307.
Edlund, L and N-P Lagerlöf (2012), “Polygyny and Its Discontents: Paternal Age and Human Capital Accumulation”, working paper.
Fenske, J (2013), “African Polygamy: Past and Present”, CSAE Working Paper WPS/2012-20.
Huillery, E (2009), “History matters: The long-term impact of colonial public investments in French West Africa”, American Economic Journal: Applied Economics 1(2): 176–215.
Jacoby, H (1995), “The economics of polygyny in Sub-Saharan Africa: Female productivity and the demand for wives in Côte d’Ivoire”, Journal of Political Economy 103(5): 938–971.
Nunn, N (2011), “Gender and Missionary Influence in Colonial Africa”, forthcoming in African Poverty of the Longue Durée.
Tertilt, M (2005), “Polygyny, fertility, and savings”, Journal of Political Economy 113(6): 1341–1374.
1 ‘Polygamy’ here refers exclusively to polygyny (multiple wives). Polyandry (multiple husbands) is uncommon everywhere.