Pre-Conference GAO Zhiyuan | Causal Nowcasting For Country-Pair Payment Flows When Disruptive Events Occur

Causal Nowcasting For Country-Pair Payment Flows When Disruptive Events Occur

To what extent does payment-related data offer useful information about what is occurring in bilateral relationships between nations? Can changes that occur in different kinds of payment messages be tied to shocks in country-pair relationships? Can financial telecommunications data act as useful indicators of such shocks? This research examines the potential for anomaly detection methods, coupled with explanatory data analytics methods, to identify intertemporal and geospatial shocks that occur in bilateral country relationships with respect to their payment activities. We explore a large historical dataset for over 200 countries, with 20,000 country-pairs for non-financial market payments for the period 2004 to 2015, provided by the SWIFT Institute. We identify both temporary and permanent effects of shocks on the payment messaging volume data, resulting in payment anomalies that exhibit different reactions to shocks. We further explore why different country-pairs have different propensities for payment anomalies to occur. We find that countries with more stable and resilient forms of governance are less likely to exhibit payment anomalies when there are various kinds of shocks, while countries with more trade openness are more likely to exhibit anomalies. Our findings are consistent with economic resilience and vulnerability theory.

This a pre-conference talk for Fourteenth Symposium on Statistical Challenges in Electronic Commerce Research (SCECR 2018).

About the Speaker