American Economic Review: Papers & Proceedings 2017, 107(5): 565–571 https://doi.org/10.1257/aer.p20171088 COOPERATION AND CONFLICT: PERSPECTIVES FROM ECONOMICS AND BEYOND Keeping It in the Family: Lineage Organization and the Scope of Trust in Sub-Saharan Africa† By Jacob Moscona, Nathan Nunn, and James A. Robinson* In the last two decades economists have thus far emphasized historical political institu- increasingly come to recognize that economies tions as a source of variation. Tabellini 2010 , are embedded in societies. For an economic sys- for example, uses historical constraints (on the) tem to work, a large number of complementary executive as a source of variation in social cap- social norms need to exist and these cannot be ital today, while Guiso, Sapienza, and Zingales taken for granted. While such ideas are latent in 2016 instead use whether or not an Italian city the debate between substantivists and formalists was( a )commune in the Middle Ages.1 about how to understand historical economic This study examines the relationship between systems e.g., Polanyi 1944 and in research by traditional political institutions and one aspect sociologists( Granovetter 1985) , their modern of social capital, trust. Within Africa, which articulation is( in the guise of “social) capital.” is the context of this study, studies of the long In his seminal book Making Democracy Work, run importance of traditional institutional struc- Robert Putnam 1993 argues that the better tures have emphasized the role of political cen- governance and (prosperity) of Northern Italy tralization e.g., Gennaioli and Rainer 2007; compared to Southern Italy was caused by the Michalopoulos( and Papaioannou 2013, 2014 . North having greater “social capital,” which he However, precolonial African societies also had) defines as the “connections among individuals— many other important characteristics besides social networks and the norms of reciprocity and their level of political centralization, many of trustworthiness that arise from them” Putnam which are determined by the social structures of 2000, p. 19 . ( the society. Examples of these social structures There is )now significant evidence that social include the nature and strength of kin relations; capital is associated with many important eco- the presence of age-based occupations or politi- nomic outcomes, such as economic develop- cal offices; rules of marriage and residence; and ment Tabellini 2010; Algan and Cahuc 2013 settlement patterns. and trade( Guiso, Sapienza, and Zingales 2009 ). In this article, we investigate the relationship Yet, why (does social capital vary across soci)- between a canonical form of social structure in eties? Following Putnam, most scholars have Africa—segmentary lineage organization—and trust. Early anthropological work recognized that many African societies, both centralized and * Moscona: Department of Economics, Massachusetts not, were systematically organized on the basis Institute of Technology, E52-380, 50 Memorial Drive, of kinship. Evans-Pritchard 1940 , in his study Cambridge, MA 02142 e-mail: [email protected] ; ( ) Nunn: Department of Economics,( Harvard University, 1805) of the Nuer of the South Sudan, classified these Cambridge Street, Cambridge, MA 02138 e-mail: nnunn@ groups as a “segmentary lineage society.” Such fas.harvard.edu ; Robinson: Harris School (of Public Policy, University of Chicago,) 1155 East 60th Street, Chicago, IL 60637 e-mail: [email protected] . 1 An exception to this approach is Nunn and Wantchekon ( ) † Go to https://doi.org/10.1257/aer.p20171088 to visit the 2011 who show that social capital in Africa is systemati- article page for additional materials and author disclosure cally( lower) in places that were more impacted by the slave statement s . trade. ( ) 565 566 AEA PAPERS AND PROCEEDINGS MAY 2017 a society is characterized by unilineal descent, and the levels of trust today. Our analysis dis- which means that people trace their ancestry tinguishes between trust of family members and back, usually to some mythical founder, either more generalized trust of nonfamily members. through the male patrilineal , or female line We find that segmentary lineage organization is matrilineal . When( such a descent) group takes associated with a larger gap between the trust on( a corporate) form, meaning that it is used to of one’s relatives compared to nonrelatives. organize economic and political life, it is said to Disaggregating the difference, we find that this form a lineage. Subsets segments of lineages is primarily due to segmentary lineage societ- function as coherent corporate( groups) and are ies having lower trust of nonrelatives, and not assigned political, administrative, and judicial higher trust of relatives. Thus, to use the termi- functions, usually under the guidance of elders. nology of Putnam 2000 , segmentary lineage Fortes 1953, p. 26 describes the system as fol- organization appears( to )decrease the level of lows: “the( individual) has no legal or political bridging social capital while having no effect on status except as a member of a lineage; … all bonding social capital. legal and political relations in the society take We now turn to a description of the data and place in the context of the lineage system.” our empirical results. In segmentary lineage societies, family is cen- tral. All forms of interaction—social, economic, I. Data and political—are structured around family ties and lineages. While this has the benefit of Our analysis uses information on the histor- strengthening associations and most likely ical presence of segmentary lineage organiza- trust among blood relatives,( it also reduces tion as coded by Moscona, Nunn, and Robinson associations) with those with whom one is not 2016 . Commonly used ethnographic sources, related. Thus, it is likely that this shrinking of, including( ) the Ethnographic Atlas, do not include what Putnam 2000 calls “associational life” information about segmentary lineage organi- i.e., interactions( outside) of one’s kin would zation. Therefore, to identify the presence or tend( to reduce the extent of trust of nonrelatives) absence of segmentary lineage organization, in society. Putnam 2000 argues that greater Moscona, Nunn, and Robinson 2016 extract social capital is facilitated( ) by more associa- information from the Ethnographic( Survey) of tional life, and the more people are members Africa, a multi-volume work compiled during of and involved with different forms of groups the mid-twentieth century that contains ethno- and societies outside of the family.2 Thus, we graphic information about a large sample of eth- expect segmentary lineage organization to limit nic groups in Africa. In total, 145 ethnic groups the extent of generalized trust in a society due to were determined to definitively either have 74 the greater reliance on family which limits asso- in total or not have 71 in total a segmentary( ciational life.3 lineage )organization. (Ethnic groups) in the sam- We test this hypothesis by examining the rela- ple reside across sub-Saharan Africa, including tionship between an ethnic group’s traditional several in Uganda, Tanzania, Kenya, Ethiopia, reliance on segmentary lineage organization Mozambique, the Democratic Republic of the Congo, Zambia, Nigeria, Ghana, Liberia, Sudan, and Sierra Leone see Moscona, Nunn, 2 Interestingly, Banfield’s 1958 book, which has been ( ( ) and Robinson 2016 . the foundation research in the area of social capital, attri- ) butes the low levels of social capital in the south of Italy to Segmentary lineage organization can be found its social structure. Specifically, it is argued that it can be among groups with both centralized and less partially explained by the dominance of the nuclear family centralized political administrations. Examples in the south. of segmentary lineage societies that are less cen- 3 There are other well-established hypotheses about tralized i.e., “stateless” include the Nuer, Tiv, the social consequences of such social structures. Sahlins ( ) 1961 , for example, hypothesized that segmentary lineage and Rega, while examples of segmentary lineage societies( ) would be “expansionary” territorially and would societies that have centralized states include the tend to be associated with conflict. In Moscona, Nunn, and Somali, Duala, and Ndembu Moscona, Nunn, Robinson 2016 we develop the first systematic database of ( ( ) and Robinson 2016 . However, there is a cor- the presence of segmentary lineage societies in Africa and ) show that the presence of such societies is associated with relation between the two characteristics. On greater levels of conflict today. average, ethnic groups that are less politically VOL. 107 NO. 5 KEEPING IT IN THE FAMILY 567 centralized are slightly more likely to have a been developed, linking the ethnic groups from segmentary lineage organization. Among the the Afrobarometer to the ethnic groups from 145 ethnic groups in our sample, there is a neg- Murdock’s Map of Africa. These are from ative relationship between the presence of seg- Nunn and Wantchekon 2011 for round 3 and mentary lineage organization and the levels of Deconinck and Verpoorten( )2013 for round political hierarchy beyond the local community 4. For round 5, we manually( matched) ethnic from the Ethnographic Atlas corr 0.21 ; groups that were not in the round 3 or 4 surveys. p 0.01 . Thus, in our analysis,( we are= −sure to In total, there are 68 ethnic groups according to control= for) this historical characteristic of ethnic Murdock’s classification in our round( 3 sample, groups.4 73 ethnic groups in our round) 4 sample, and 92 To measure the nature of social capital, we ethnic groups in our round 5 sample.5 focus on responses about trust in others from the 2005 round 3 , 2008 round 4 , and 2015 II. Results round 5 Afrobarometer.( ) The( database) compiles nationally( ) representative surveys conducted To investigate our questions of interest, we in local languages from 18, 20, and 34 coun- use the same basic specification as from Nunn tries, respectively. In all surveys, respondents and Wantchekon 2011 , but with two import- are asked how much they trust their “relatives.” ant changes.