Skip to main content

Causal Inference in Sarafu Network

Online Event
CEU N13 Atrium
Tuesday, May 26, 2020, 2:00 pm – 3:00 pm

Please note that this is an online event for the CEU Community only.

If you wish to attend, please email for an internal meeting invite.

ABSTRACT / In this work I evaluate the effect of a promotional campaign on the individual performance in the Sarafu network, a group of Community Currency Systems circulating alongside the national currency system in Kenya. The transaction data analyzed in this paper covers the period March-September 2019. The promotional campaign took place in the villages of Kilifi county in June 2019. The villages in Kwale county are used as control group. A differences-in-differences estimation method is used with controls for individual network-related features. The causal effect turns to be significantly positive, but the anticipatory effect significantly negative. Therefore, the parallel trends assumption can be considered as invalid.

However, in this work I tested two new measures for economic network analysis: cyclic participation and star centrality. Their comparison with other economic network measures and their statistical significance is reported.