Kinship Systems,Gender Norms and Household B :E

Download Kinship Systems,Gender Norms and Household B :E

Preview text

Kinship Systems, Gender Norms, and Household Bargaining: Evidence from the Matrilineal Belt *


ara owes

31 October 2016 Most Recent Version Here

Abstract: I examine how broader social structures – in particular, kinship systems – affect intra-household bargaining. In matrilineal kinship systems, lineage and inheritance are traced through female members. I test the predictions of the “matrilineal puzzle,” the hypothesis that matrilineal kinship systems decrease spousal cooperation relative to patrilineal systems by creating split allegiances between spouses and by reducing a husband’s authority over his wife. I use experimental and physiological measures and a geographic regression discontinuity design along the “matrilineal belt” in Africa to test for greater discord between matrilineal couples. I show that individuals from matrilineal ethnic groups cooperate less with their spouses in a lab game and experience greater stress during game play. Despite less spousal cooperation, I find that children of matrilineal women are healthier and better educated. I explore the channels through which matrilineal kinship systems affect cooperation. First, due to split allegiances between spouses, matrilineal individuals are less altruistic towards their spouse. Second, matrilineal women have greater bargaining power and can therefore cooperate less with their husband without fear of reprisal. The results highlight how broader social structures can affect the bargaining process within the household. Additionally, at relatively low levels of women’s empowerment, there may be a trade off between increasing women’s bargaining power and household efficiency.
Keywords: Kinship systems, household bargaining, culture, gender.
JEL Classification: D13, N47, J16, Z13.

*I thank Alberto Alesina, Robert Bates, Iris Bohnet, Michael Callen, Jeffrey Carpenter, Ed Glaeser, Claudia Goldin,

Sendhil Mullainathan, Nathan Nunn, Rohini Pande, Gautam Rao, and James A. Robinson for their feedback and

guidance. Cammie Curtin, Anne DeGrave, and Adam Xu provided excellent research assistance. I thank Martin

Abel, Natalie Bau, Matt Lowe, Eduardo Montero, Raissa Fabregas, Amanda Robinson, Heather Sarsons and Jack Willis

for their feedback. I am grateful for financial support from: the National Science Foundation Graduate Research

Fellowship Program, Harvard IQSS and the Harvard Center for African Studies. This project has IRB approval from

the Harvard CUHS (IRB15-1787).

†Harvard University, Department of Economics,

Cambridge Street, Cambridge, MA,

. Email:



[email protected] Website:

1. Introduction
Large gaps in outcomes between men and women exist in many developing countries. Women often have less education, poorer health, limited autonomy, and are subjected to physical and emotional violence. One factor that may contribute to the lower status of women is social structure, particularly if social structure affects bargaining within the household. Social structure refers to norms governing how people, families, or societies are organized. A kinship system is an example of a social structure that determines how families trace group membership, descent, and inheritance. Anthropologists have long studied the variation in kinship systems and the implications of these systems for societal outcomes, but economists are just beginning to understand how kinship structure matters for women’s outcomes (Alesina et al., 2016; LaFerrara and Milazzo, 2014; Gneezy et al., 2009).
Kinship systems determine the set of people to whom an individual is considered related and their social obligations to this set of people (Radcliffe-Brown, 1950). An important element of a kinship system is the determination of descent.1 In unilineal descent systems, kin are defined using only one of the two parents (Fox, 1934). Matrilineal kinship systems, which are prevalent in Central Africa, are a type of unilineal descent system. In matrilineal kinship systems group membership and inheritance are traced through the female members. Individuals are part of their mother’s kinship group and inheritance is restricted to the children of the female members of the group. In contrast, in patrilineal systems individuals are part of their father’s kinship group and inheritance can only be passed on to children of male group members. Importantly, matrilineal systems are not symmetric with patrilineal systems, because in patrilineal systems, a wife is effectively incorporated into the lineage of her husband, while in matrilineal systems, husbands and wives maintain strong allegiances with their own (different) lineages.
This paper examines how matrilineal kinship systems affect intra-household cooperation relative to patrilineal kinship systems. Anthropologists have long puzzled over the stability of matrilineal systems, arguing that matrilineal systems create conflicting allegiances within the household (Fox, 1934). A husband in a matrilineal society has allegiances to his sisters, whose children he may support because they are his heirs, and the wife has allegiances to her brother, who provides her and her children with support. Additionally, matrilineal lineage systems un-
1Most Western societies practice cognatic descent, in which kinship ties are traced through both parents. An individual considers people related through their mother and through their father to be kin.

dermine men’s authority over their spouses. In matrilineal societies a woman’s children belong to her lineage, not to her husband’s lineage. Thus, relative to patrilineal societies, women maintain greater control over their children. Having the children as part of a woman’s lineage may increase the value of her outside option and increase her relative bargaining power. A large literature in anthropology suggests that matrilineal systems reduce spousal cooperation, a hypothesis which I test formally (Radcliffe-Brown, 1950; Gluckman, 1963; Richards, 1950; Douglas, 1969).
To examine the relationship between matrilineal kinship systems and cooperation within the household, I collected data from 320 matrilineal and patrilineal couples in the Democratic Republic of Congo (DRC). The DRC is intersected by the “matrilineal belt,” which describes the distribution of matrilineal ethnic groups across the center of Africa. The data are collected in a major city in the south of the DRC, where there are many matrilineal and patrilineal ethnic groups. The individuals come from villages along the border of the matrilineal belt, but they share a common institutional setting presently. Approximately 40% of the sample are from a matrilineal ethnic group.
I use laboratory experiments to measure cooperation within the household. I find that matrilineal individuals - both men and women - cooperate less with their spouses in a household public goods game. These results are driven by opportunities to hide income. When partnered with a stranger of the opposite sex, matrilineal individuals no longer differentially respond to opportunities to hide income, suggesting that the differential cooperation by matrilineal couples is behavior specific to being paired with a spouse and not more general to cooperation with a stranger of the opposite sex. To address identification concerns, I estimate a geographic regression discontinuity specification along the matrilineal belt border. The geographic RD results are consistent with the OLS results: matrilineal individuals cooperate less with their spouse and therefore earn less money in the lab experiment relative to patrilineal individuals.
Additionally, I collect physiological data during game play to provide complementary evidence of greater discord between matrilineal couples. A random subset of respondents completed ultimatum games with their spouse and with a stranger of the opposite sex in a lab environment while wearing equipment designed to monitor electrodermal activity (EDA). I find that when matrilineal individuals are paired with their spouses they experience greater stress responses than patrilineal individuals as measured by an increase in skin conductance. Matrilineal individuals do not exhibit greater stress when they are paired with strangers of the opposite sex. These results

provide physiological evidence that matrilineal individuals exhibit greater stress when engaged in bargaining interactions with their spouses.
Importantly, although I find evidence in favor of less cooperation between matrilineal spouses, I also find evidence of matrilineal women being better able to enact their preferences. Children of matrilineal women are actually healthier and better educated relative to children of patrilineal women. I find similar patterns using Demographic and Health Survey data for the DRC. Women from matrilineal areas have fewer children that have died and their children have more years of education.
I identify two potential channels for how matrilineal kinship systems decrease cooperation between spouses. These channels are related to particular structural features of matrilineal kinship systems. First, matrilineal individuals may have less altruism toward their spouse because they maintain strong allegiances to their own lineage. I find that matrilineal individuals give less money to their spouse in a dictator game relative to patrilineal individuals. Whereas patrilineal individuals are more generous with their spouses than they are with strangers, matrilineal individuals treat their spouses similarly to how they treat strangers. In a simple non-cooperative model of contributions to a public good, lower levels of altruism lead to smaller contributions to the public good.
Second, matrilineal women appear to have more bargaining power, consistent with the hypothesis that matrilineal systems reduce a husband’s authority over his wife. I use a simple non-cooperative bargaining model to demonstrate how increasing a woman’s bargaining power may actually lead her to invest less in a public good if, as her bargaining power increases, she faces less fear of reprisal for investing in her private good. In this model, her contribution is minimized as bargaining weights are equalized between husband and wife. In contrast, a husband’s contribution is always increasing in his bargaining weight. I present evidence from an ultimatum game and survey questions that suggest matrilineal women have more bargaining power than patrilineal women. I use DHS data to show that women from matrilineal areas in DRC report greater autonomy in decision making and face lower threat of domestic violence.
This paper is related to several literatures in economics, including literatures on the relationship between cultural norms and outcomes for women, on the economics of the family, and on women’s empowerment. In a recent review article, Jayachandran (2015) suggests a variety of cultural practices may affect outcomes for women, such as patrilocality (the practice of living near

the groom’s parents after marriage), payment of bride price or dowry, and patrilineality. There is a growing literature that examines the effects of some of these specific cultural practices (Alesina et al., 2016; Ashraf et al., 2016; Bau, 2016; Gottlieb and Robinson, 2016; LaFerrara and Milazzo, 2014). I focus on a region where many different ethnic groups share a similar geographic setting and history. By comparing individuals along the matrilineal belt, I am better able to isolate the effect of matrilineal kinship on spousal cooperation. The paper is also related to the literature on observed inefficiencies in the household. For example, Udry (1996) finds that household agricultural production does not meet the Pareto-efficient assumption of collective models of the household. Recent lab experiments have also rejected productive efficiency in a variety of settings including Ethiopia, Uganda and India (Kebede et al., 2013; Iversen et al., 2011; Castilla, 2013). Finally, this paper relates to the literature on the relationship between economic development and outcomes for women (Duflo, 2012; Doepke and Tertilt, 2014), by providing evidence that increasing women’s bargaining power may actually decrease spousal cooperation, but may have positive benefits for investment in children.
The remainder of the paper is organized as follows. Section 2 defines matrilineal kinship and describes its origins and practice. Section 3 describes the data collection process and the experimental design. Sections 4 presents the OLS and geographic RD results. Section 5 presents the physiological results. Section 6 examines the implications of matrilineal kinship systems for child outcomes using survey and DHS data. Section 7 outlines a non-cooperative model of contributions to the public good under threat of violence. Section 8 explores channels and Section 9 concludes.
2. Background
2.1. What are matrilineal kinship systems?
In matrilineal kinship systems, individuals trace lineage and descent through women. Biologically, of course, an individual is related to family on both the mother and father’s side; however, in matrilineal systems individuals are considered kin only if they share a common female ancestor. Figure 1 illustrates the structure of matrilineal kinship systems. In the diagram, men are represented by triangles and women are represented by circles. Membership in the same matrilineal group is denoted with red. Children are in the same matrilineal group as their

mothers. Likewise, a mother is in the same matrilineal group as her male and female siblings. In many matrilineal societies, the mother’s brother has an important role relative to his sister’s children. His inheritance and lineage will be traced through his sister’s children, and he has obligations to financially support her children. Importantly, husband and wife do not share the same lineage - for all married couples one spouse is blue and the other spouse is red.
Figure 2 presents the structure of patrilineal kinship. Now, children are in the same group as their father, as denoted in blue. In a patrilineal society, rather than maintaining strong ties with her own lineage, a woman is effectively incorporated into the lineage of her husband upon marriage. This is because once she is married, she is not relevant for determining descent and inheritance for her lineage. This is illustrated in the patrilineal kinship diagram by the married women denoted in grey, while the unmarried daughter shares the same color as her father.
Fig6urBe;m:`2DiRa,grJamio`fBMHBMa2trilHinEeaBMl Kb?inBTship



G2;2M/, 4 JH2b- 4 62KH2b- f 4 aK2 Ji`BHBMv
6B;m`2 k, Si`BHBM2H EBMb?BT




4 JH2b-

4 62KH2b- f 4 aK2 Si`BHBMv 5

G2;2M/, 4 JH2b- 4 62KH2b- f 4 aK2 Ji`BHBMv
Fig6urBe;m2:`2Dika,grSami`oBfHBPMa2trilHinEeaBMl bK?inBTship



G2;2M/, 4 JH2b- 4 62KH2b- f 4 aK2 Si`BHBMv
The kinship grou6psB;dme`fi2nejd, bSymi`aBtHrBiMli2n[email protected]@satevmKsKar2eio`Bf+ten important in sub-
Saharan Africa. They form a basic political unitƌin which members recognize each other as
kin and often have certain obligations toward each other (Fox, 1934). For example, members of the same matrilineal group may share land and may contribute to bride price payments for
lineage members. They may also pƌrovide financial support in thƌe form of school fees or burial
payments. Thus, membership in a matrilineal or patrilineal society determines your obligations and privileges relative to your kin group.
Matrilineal groups are relatively rare. Of the 1267 societies in the Ethnographic Atlas, only 12 percent are matrilineal (while 46 percent are patrilineal).2 Within sub-Saharan Africa, 15 percent
of the 527 Gso2c;ie2tiMes/i,n the4EtJhnoHg2rba-phic 4Atl6a2sKareH2mb-atrilfineal 4anad7K0 2peSrcein`tBHBaMrve patrilineal.
The vast majority of these matrilineal societies are distributed across the center of Africa in
the so called “matrilineal belt” (Richards, 1950R, p.207). The matrilineal belt intersects present
day Angola, Republic of Congo, DRC, Gabon, Malawi, Mozambique, Namibia, Tanzania and Zambia. Figure 3 illustrates the matrilineal belt across Africa, with matrilineal groups denoted in blue, patrilineal groups denoted in green, and bilateral and other groups in beige. For more information on the historical development and spread of matrilineal kinship systems in sub-Saharan Africa, see Appendix A.
Historically, matrilineal kinship systems are correlated with other cultural traits. Table 1 shows
2The Ethnographic Atlas is a data set compiled by George Murdock that documents the practices and customs of various societies across the world.

some of the traits that are correlated with matrilineality within Africa in the Ethnographic Atlas. The table presents traits that other work in economics has shown to be important for development, including: bride price, residence after marriage, jurisdictional hierarchy, plough use, and presence of animal husbandry (Ashraf et al., 2016; Bau, 2016; Michalopoulos and Papaioannou, 2013a; Michalopoulos et al., 2016; Alesina et al., 2013, 2016; Alsan, 2015). Not surprisingly, matrilineality is highly correlated with matrilocal residence patterns, which is when a couple lives in the same village as the bride’s mother’s kin group.3 Historically, matrilineal groups are less likely to pay bride price, to use the plough, or to rely on animal husbandry. There is no difference in levels of jurisdictional hierarchy between matrilineal groups and other groups.
Figure 3: Ethnic Group Boundaries and Matrilineal Belt
To motivate the study of matrilineal and patrilineal ethnic groups near the border of the matrilineal belt, Panel B of Table 1 presents the same historical correlates of matrilineality restricting the Ethnographic Atlas observations to those ethnic groups that can be matched to groups in
3There are many potential living arrangements after marriage. In matrilocal (or uxorilocal) groups, couples live in the same village as the bride’s mother’s group. A type of uxorilocal residence pattern is avunculocal residence, in which the couple lives in the village of the bride’s maternal uncle. In patrilocal (or virilocal) groups, couples live in the same village as the groom’s father’s group. In natolocal groups couples stay in their natal homes on marriage, and in neolocal groups they establish a new residence upon marriage (Fox, 1934)).

Table 1: Historical Correlates of Matrilineality

Matrilocal Residence

Bride Price

Panel A: All of Africa

Jurisdictional Hierarchy

Plough Use

Animal Husbandry

Observations Mean Dep. Var.

0.643*** (0.054)
527 0.104
Matrilocal Residence

-0.228*** (0.056)

-0.172 (0.110)

-0.0578*** (0.022)

527 0.831

472 2.201

527 0.074

Panel B: Sample Ethnic Groups

Bride Price

Jurisdictional Hierarchy

Plough Use

-0.831*** (0.192)
500 2.516
Animal Husbandry


















Mean Dep. Var.






Notes: Robust standard errors are in parentheses. The data are from the Ethnographic

Atlas and are restricted to groups in Africa in Panel A and to ethnic groups in my sample

in Panel B. Matrilineal is an indicator variable equal to 1 if the society has inheritance and

descent traced through women. The other types of descent systems include patrilineal,

bilateral, duolateral, ambilineal, quasi-lineages and mixed. Matrilocal Residence is an indi-

cator variable equal to 1 if the society has avunculocal, uxorilocal or matrilocal residence

patterns. Bride price is an indicator variable equal to 1 if the society has bride price. This

does not include token bride price or bride service. Jurisdictional hierarchy is coded from 0

to 4, with 0 being no levels of political hierarchy to 4 being a large state. Plough use is an

indicator variable equal to 1 if the society had the plough prior to colonialism or adopted

it subsequently. Animal husbandry is coded from 0 to 9, corresponding with percentage

dependence on animal husbandry. *** p<0.01, ** p<0.05, * p<0.1


the study sample.4 In the restricted sample many of the differences observed in Panel A are no longer statistically significant, though the sample size is also quite a bit smaller. Reassuringly, the magnitudes on the coefficients are also small. Although, matrilineal ethnic groups are still more likely to be matrilocal historically, this is less relevant for individuals in the study sample, since all of the respondents now live in a common urban environment away from their villages of origin. Additionally, survey data confirm that most individuals in the sample practiced neolocal residence after marriage, moving to a location different than that of either spouses’ families. Finally, while matrilineal groups were less likely to pay bride price historically, it is now the custom for all ethnic groups in the study sample to pay bride price. There are no differences in amount of bride price paid between matrilineal and patrilineal individuals in my sample.5
2.2. The Matrilineal Puzzle
Much of the early anthropological scholarship on matrilineality focused on the so called “matrilineal puzzle”. The matrilineal puzzle is the hypothesis that matrilineal kinship systems decrease spousal cooperation, and therefore it is puzzling to observe them as a kinship system. Anthropologists note that matrilineal systems (1) split an individual’s allegiance between their spouse and their lineage and (2) undermine male authority. First, in patrilineal systems, women effectively relinquish membership in their own lineage to be de facto members of their husband’s lineage. However, in matrilineal systems both partners retain strong ties with their own lineages. This leads to split allegiances within a matrilineal household. Second, given requirements of exogamy, or marrying outside of the kinship group, a woman produces children with a man outside of her group, but these children are to belong to her lineage, rather than her husband’s lineage. Thus, a husband in a matrilineal society does not have the same authority and control over his wife or children as a husband in a patrilineal society, in which the children are members of the husband’s group. As Richard’s writes in her work on matrilineality among the Central Bantu, "the matrilineal system makes for certain elements of conflict for which some kind of
4Not all ethnic groups in my sample can be matched to an observation in the Ethnographic Atlas. This is for two reasons. First, the Ethnographic Atlas sometimes aggregates smaller groups into a larger ethnic group. Additionally, some groups in my sample are just not represented in the Ethnographic Atlas.
5The survey data confirm that the payment of brideprice is a common practice for all ethnic groups. In the survey data, men report whether they paid brideprice and how much they paid. Almost all couples (99%) report having paid bride price.

Preparing to load PDF file. please wait...

0 of 0
Kinship Systems,Gender Norms and Household B :E