Networks of monetary flow at native resolution

Abstract: People and companies move money with every financial transaction they make. We aim to understand how such activity gives rise to large-scale patterns of monetary flow. In this work, we trace the movement of e-money through the accounts of a mobile money system using the provider's own transaction records. The resulting transaction sequences---balance-respecting trajectories---are data objects that represent observed monetary flows. Common sequential motifs correspond to known use-cases of mobile money: digital payments, digital transfers, and money storage. We find that each activity creates a distinct network structure within the system, and we uncover coordinated gaming of the mobile money provider's commission schedule. Moreover, we find that e-money passes through the system in anywhere from minutes to months. This pronounced heterogeneity, even within the same use-case, can inform the modeling of turnover in money supply. Our methodology relates economic activity at the transaction level to large-scale patterns of monetary flow, broadening the scope of empirical study about the network and temporal structure of the economy.