The Impact of Promotions in Sarafu Network

June 2, 2020

Teodoro Criscione, a 2nd year PhD candidate at the Department of Network and Data Science, presented his ongoing research on the network of the Community Currency System (CCS) in Kenya, the so called Sarafu Network. The emergence of CSS is driven by the need for an alternative credit system outside the official monetary system in a small and close-knit environment. In order to increase the well-being of a community, this aforementioned scheme focuses on financing community projects. 

Initiated in 2013, the Sarafu Network works in parallel with the national currency system. In 2019, 12 community currencies were connected to Mombasa-Pesa and Nairobi-Pesa, involving about 45 different locations in 4 different main counties: Mombasa, Kwale, Kilifi, and Nairobi. In order to upscale the implementation, a promotional measure was taken in June 2019 in Kilifi through such mechanism: pre-existing users are entitled to 10% discount on the next transaction for every three new members they could attract. Afterwards, a fundamental question is worth investigating: did the promotional campaign in Kilifi have an effect on the individual performance of users? 

In this research, individual features and network-related characteristics are used to control for confounders (Cyclic Participation and Star Position) and test the causal effect of the campaign. Cyclic Participation is defined as individual embeddedness into the economic network in terms of trade cycles, meanwhile, Star Position is framed as relative Monopolistic and Monopsonistic positions of the individual. Under this setting, Kilifi falls into treatment group, while Kwale is considered as control group. Operativity Score (total weighted outcoming degree over the total weighted incoming degree) is measured and employed as an outcome in the regression model below.

This research finds that operativity score is not significant with respect to the chosen null model. On the left hand side, Cyclic Participation is significant at 1% in Kilifi, but not significant in Kwale. Moreover, Star Position is significant at 1% in the case of Kilifi White Stars during pre-treatment and also significant at 5% in the case of Black Stars during post-treatment. It could be concluded that Cyclic Participation plays an important role in promotion instead of Cyclic Centrality. In addition, the dying White Star and the growing Black Star on reciprocal dyadic relationships are considerably pivotal. 

Blog post by Rafiazka Hilman