This is the twelfth in our series of posts by students on the job market this year.
In many developing countries we observe large gaps in outcomes between men and women. Women often have fewer years of education, worse health, limited autonomy, and are subjected to physical and emotional violence ( Jayachandran, 2015). My research is motivated by the observation that many important outcomes for women are determined through household bargaining, yet our models of intra-household bargaining rarely consider the broader social context ( Lundberg and Pollak, 1993). In my job market paper, I examine how kinship systems affect spousal cooperation. Specifically, I test whether matrilineal kinship systems affect spousal cooperation relative to patrilineal systems using original survey data, lab experiments, physiological data, and a geographic regression discontinuity design along the “matrilineal belt” in the Democratic Republic of Congo (DRC).
What are matrilineal kinship systems?
In matrilineal kinship systems, lineage and inheritance are traced through a group’s female members and children are part of their mother’s kinship group. In contrast, in patrilineal systems, group membership is determined through men and children are part of their father’s kinship group. Within sub-Saharan Africa, matrilineal ethnic groups are about 15% of the 527 societies represented in the Ethnographic Atlas, a data set compiled by anthropologists to document the cultural practices of various societies across the world. Many of these matrilineal groups are located in the matrilineal belt, which describes the distribution of matrilineal societies across the center of Africa. (See Figure 1 for matrilineal groups in the DRC and the delineation of the matrilineal belt.) Despite being relatively common in sub-Saharan Africa, economists are just beginning to understand how matrilineal kinship systems affect outcomes for women ( Alesina et al., 2016; Bau, 2016; LaFerrara and Milazzo, 2014).
Anthropologists have noted that matrilineal systems may decrease spousal cooperation for two reasons. First, matrilineal systems create split allegiances between spouses. In a patrilineal system a wife is effectively incorporated into her husband’s kinship group, while in a matrilineal system both husband and wife maintain strong ties with their own kinship groups. Second, matrilineal systems may increase a woman’s outside option. Therefore, women can cooperate less with their husbands without fear of reprisal. If her husband mistreats her, she can more easily return to her kinship group relative to patrilineal women (Richards, 1950; Gluckman, 1963).
Do matrilineal individuals cooperate less with their spouse?
Evidence from Lab Experiments
To examine whether matrilineal individuals cooperate less with their spouse relative to patrilineal individuals, I collect data from 320 matrilineal and patrilineal couples in the DRC. Individuals originate from villages along the matrilineal belt border, but all presently reside in Kananga, the provincial capital of Kasai Central province. Approximately 40% of the sample is from a matrilineal ethnic group, and there are 28 different ethnic groups represented in the sample.
Figure 1: The Matrilineal Belt in DRC and Villages of Origin of Sample
I use laboratory experiments to measure cooperation in the household. Individuals participate in a public goods game with their spouse, where they are asked to allocate an endowment between a private envelope and a household envelope. To better simulate the real life coordination problem couples face, individuals have a 50% probability of receiving an unobservable positive income shock to their initial endowment prior to making their allocation decision. Contributions to the household envelope are combined with the contributions of the spouse, multiplied by either 1.5 or 2, and then divided evenly between the spouses. An individual’s total payoff is thus the amount contributed to the private envelope and half of the amount earned in the household envelope. To maximize household income in the game, both spouses would need to allocate their full endowment to the household envelope.
To identify the effect of being from a matrilineal ethnic group on spousal cooperation, I use a geographic regression discontinuity design with individuals’ origin villages relative to the matrilineal belt border (see Figure 1 for the villages of origin of individuals in the sample relative to the matrilineal belt). I find that individuals from matrilineal groups – both men and women – contribute 13% less to the household envelope in the public goods game relative to patrilineal individuals, leading to monetary losses at the household level. This is particularly the case when matrilineal individuals receive the unobserved income shock and can more easily hide money from their spouse. Individuals also complete the same experiment with a stranger of the opposite sex. Importantly, matrilineal individuals no longer respond differentially to opportunities to hide income when the other player is a stranger of the opposite sex, suggesting that differential matrilineal behavior is specific to being paired with a spouse.
Evidence from Physiological Measures
As an alternative measure of spousal discord, I also collect physiological data from a subset of couples during game play. Individuals wore devices that record electrodermal activity (EDA). EDA describes the involuntary changes in how well the skin conducts electricity. When the sympathetic nervous system, which controls the body’s “fight or flight” response, is aroused, sweat activity increases, and skin is better able to conduct electricity. Often, increased arousal of the sympathetic nervous system is associated with an increase in stress ( Boucsein, 1992). I find that in a simple bargaining game with a spouse, both matrilineal men and women exhibit larger increases in skin conductance relative to patrilineal individuals. However, when paired with a stranger of the opposite sex, matrilineal individuals no longer exhibit larger stress responses.
Are there benefits to matrilineal systems?
Thus far I have presented evidence that matrilineal individuals cooperate less with their spouses and exhibit greater stress in a bargaining task with a spouse. However, even if matrilineal systems undermine spousal cooperation, there may be benefits to a social structure that empowers women and allows them to enact their preferences. One natural outcome to examine is investment in children. In my sample, children of matrilineal women are 9 percentage points less likely to have been sick in the last month, even controlling for various characteristics of the mother, including education and wealth. Additionally, children of matrilineal women have 10% more years of education. I am able to examine similar outcomes using Demographic and Health Survey (DHS) data from 2007 and 2014 for the DRC. In the DHS data, matrilineal women have fewer children that have died and their children are better educated relative to children of patrilineal women.
I provide suggestive evidence that the threat of violence may be an important factor explaining why matrilineal women are relatively less cooperative. Matrilineal women are .13 standard deviations less accepting of domestic violence and are .11 standard deviations less likely to have experienced domestic violence in the DHS. In a context where women are expected to share their income with their husbands, matrilineal women may be relatively more empowered to retain that income for themselves.
Bringing it all together
I find that kinship systems matter for understanding bargaining outcomes in the household. I present lab evidence that matrilineal men and women are less cooperative with their spouses and physiological evidence that they experience greater stress in a bargaining task with their spouses. Households with matrilineal individuals are less able to reap the benefits of cooperation. However, children of matrilineal women are healthier and better educated, and matrilineal women are less likely to experience domestic violence. This project speaks to the broader literature on the relationship between women’s empowerment and economic development ( Duflo, 2012). The results suggest that at low levels of women’s bargaining power, empowering women may actually lead to greater inefficiencies within the household, but that children and women benefit from kinship systems that empower women.
Sara Lowes is a PhD student at Harvard University. More information on her research can be found on her website .
In many developing countries we observe large gaps in outcomes between men and women. Women often have fewer years of education, worse health, limited autonomy, and are subjected to physical and emotional violence ( Jayachandran, 2015). My research is motivated by the observation that many important outcomes for women are determined through household bargaining, yet our models of intra-household bargaining rarely consider the broader social context ( Lundberg and Pollak, 1993). In my job market paper, I examine how kinship systems affect spousal cooperation. Specifically, I test whether matrilineal kinship systems affect spousal cooperation relative to patrilineal systems using original survey data, lab experiments, physiological data, and a geographic regression discontinuity design along the “matrilineal belt” in the Democratic Republic of Congo (DRC).
What are matrilineal kinship systems?
In matrilineal kinship systems, lineage and inheritance are traced through a group’s female members and children are part of their mother’s kinship group. In contrast, in patrilineal systems, group membership is determined through men and children are part of their father’s kinship group. Within sub-Saharan Africa, matrilineal ethnic groups are about 15% of the 527 societies represented in the Ethnographic Atlas, a data set compiled by anthropologists to document the cultural practices of various societies across the world. Many of these matrilineal groups are located in the matrilineal belt, which describes the distribution of matrilineal societies across the center of Africa. (See Figure 1 for matrilineal groups in the DRC and the delineation of the matrilineal belt.) Despite being relatively common in sub-Saharan Africa, economists are just beginning to understand how matrilineal kinship systems affect outcomes for women ( Alesina et al., 2016; Bau, 2016; LaFerrara and Milazzo, 2014).
Anthropologists have noted that matrilineal systems may decrease spousal cooperation for two reasons. First, matrilineal systems create split allegiances between spouses. In a patrilineal system a wife is effectively incorporated into her husband’s kinship group, while in a matrilineal system both husband and wife maintain strong ties with their own kinship groups. Second, matrilineal systems may increase a woman’s outside option. Therefore, women can cooperate less with their husbands without fear of reprisal. If her husband mistreats her, she can more easily return to her kinship group relative to patrilineal women (Richards, 1950; Gluckman, 1963).
Do matrilineal individuals cooperate less with their spouse?
Evidence from Lab Experiments
To examine whether matrilineal individuals cooperate less with their spouse relative to patrilineal individuals, I collect data from 320 matrilineal and patrilineal couples in the DRC. Individuals originate from villages along the matrilineal belt border, but all presently reside in Kananga, the provincial capital of Kasai Central province. Approximately 40% of the sample is from a matrilineal ethnic group, and there are 28 different ethnic groups represented in the sample.
Figure 1: The Matrilineal Belt in DRC and Villages of Origin of Sample
I use laboratory experiments to measure cooperation in the household. Individuals participate in a public goods game with their spouse, where they are asked to allocate an endowment between a private envelope and a household envelope. To better simulate the real life coordination problem couples face, individuals have a 50% probability of receiving an unobservable positive income shock to their initial endowment prior to making their allocation decision. Contributions to the household envelope are combined with the contributions of the spouse, multiplied by either 1.5 or 2, and then divided evenly between the spouses. An individual’s total payoff is thus the amount contributed to the private envelope and half of the amount earned in the household envelope. To maximize household income in the game, both spouses would need to allocate their full endowment to the household envelope.
To identify the effect of being from a matrilineal ethnic group on spousal cooperation, I use a geographic regression discontinuity design with individuals’ origin villages relative to the matrilineal belt border (see Figure 1 for the villages of origin of individuals in the sample relative to the matrilineal belt). I find that individuals from matrilineal groups – both men and women – contribute 13% less to the household envelope in the public goods game relative to patrilineal individuals, leading to monetary losses at the household level. This is particularly the case when matrilineal individuals receive the unobserved income shock and can more easily hide money from their spouse. Individuals also complete the same experiment with a stranger of the opposite sex. Importantly, matrilineal individuals no longer respond differentially to opportunities to hide income when the other player is a stranger of the opposite sex, suggesting that differential matrilineal behavior is specific to being paired with a spouse.
Evidence from Physiological Measures
As an alternative measure of spousal discord, I also collect physiological data from a subset of couples during game play. Individuals wore devices that record electrodermal activity (EDA). EDA describes the involuntary changes in how well the skin conducts electricity. When the sympathetic nervous system, which controls the body’s “fight or flight” response, is aroused, sweat activity increases, and skin is better able to conduct electricity. Often, increased arousal of the sympathetic nervous system is associated with an increase in stress ( Boucsein, 1992). I find that in a simple bargaining game with a spouse, both matrilineal men and women exhibit larger increases in skin conductance relative to patrilineal individuals. However, when paired with a stranger of the opposite sex, matrilineal individuals no longer exhibit larger stress responses.
Are there benefits to matrilineal systems?
Thus far I have presented evidence that matrilineal individuals cooperate less with their spouses and exhibit greater stress in a bargaining task with a spouse. However, even if matrilineal systems undermine spousal cooperation, there may be benefits to a social structure that empowers women and allows them to enact their preferences. One natural outcome to examine is investment in children. In my sample, children of matrilineal women are 9 percentage points less likely to have been sick in the last month, even controlling for various characteristics of the mother, including education and wealth. Additionally, children of matrilineal women have 10% more years of education. I am able to examine similar outcomes using Demographic and Health Survey (DHS) data from 2007 and 2014 for the DRC. In the DHS data, matrilineal women have fewer children that have died and their children are better educated relative to children of patrilineal women.
I provide suggestive evidence that the threat of violence may be an important factor explaining why matrilineal women are relatively less cooperative. Matrilineal women are .13 standard deviations less accepting of domestic violence and are .11 standard deviations less likely to have experienced domestic violence in the DHS. In a context where women are expected to share their income with their husbands, matrilineal women may be relatively more empowered to retain that income for themselves.
Bringing it all together
I find that kinship systems matter for understanding bargaining outcomes in the household. I present lab evidence that matrilineal men and women are less cooperative with their spouses and physiological evidence that they experience greater stress in a bargaining task with their spouses. Households with matrilineal individuals are less able to reap the benefits of cooperation. However, children of matrilineal women are healthier and better educated, and matrilineal women are less likely to experience domestic violence. This project speaks to the broader literature on the relationship between women’s empowerment and economic development ( Duflo, 2012). The results suggest that at low levels of women’s bargaining power, empowering women may actually lead to greater inefficiencies within the household, but that children and women benefit from kinship systems that empower women.
Sara Lowes is a PhD student at Harvard University. More information on her research can be found on her website .
Join the Conversation