Ethnicity and Networks in African Trade

This paper investigates the role of ethnicity, religion, gender, and networks in domestic agricultural trade in Africa. Using a theoretical model of self-disciplining markets, we begin by demonstrating that statistical discrimination and networks can generate similar patterns of ethnic concentration. The model also predicts that traders who start with an ethnic or network advantage grow faster. We then test these ideas using original survey data collected in Benin, Malawi, and Madagascar. We find no evidence that members of a particular sex or ethnic group are more easily trusted by suppliers and trust clients more easily. In contrast, network effects have a strong and systematic effect on trust and information sharing. We also find some evidence that religion matters, but the effect is not due to discrimination or better networks. Ethnicity also has little effect on start-up networks and working capital but women accumulate working capital slower than men, including in Benin where women represent 80% of surveyed traders. Agricultural trade appears fairly open to all, irrespective of gender, ethnicity, or religion.