Bivariate sudden stop analysis of equity and bond fund flows to emerging markets using isolation forest | Susanne Griebsch, Andreas Röthig
Sudden stops and surges in emerging-market equity and bond fund flows hit their highest intensity in a decade in 2022, a Bundesbank working paper shows.
Source: Deutsche Bundesbank · July 24, 2026 at 10:59 PM · AI-assisted report
KUALA LUMPUR, 25 JULY 2026 —
Listen to this article
DomainFork Audio · read aloud
Sudden stops and surges in emerging-market equity and bond fund flows hit their highest intensity in a decade in 2022, a Bundesbank working paper shows.
Market Impact
Deutsche Bundesbank economists Susanne Griebsch and Andreas Röthig used an isolation forest algorithm on weekly EPFR Global data from 2004 to 2024 to measure anomalies. Traditional methods flag sudden stops and surges using fixed thresholds and analyze equity and bond flows separately, but the authors say their bivariate approach captures the joint dynamics of these markets.
The model recorded the most bivariate anomalies since 2013 in 2022, with the median weekly anomaly score reaching its peak in the sample. “Most anomalies occur when both equity and bond funds move in the same direction,” the authors write. Mixed patterns—such as equity inflows with bond outflows—were less common and may reflect shifts in global risk appetite.
The paper compares the isolation forest to a static threshold model that labels flows as surge, normal, or sudden stop. The algorithm detected prolonged anomalous periods in 2021–2022 that the threshold approach missed, the authors say. “Traditional methods often misclassify normal data points as anomalies while missing prolonged stress,” Röthig said.
The study argues that simultaneous reversals in equity and bond markets can amplify macroeconomic risks in emerging economies. It links the sharpest bivariate stop in the sample to the 2022 taper tantrum and subsequent Federal Reserve rate hikes.
The authors propose expanding the framework to include additional asset classes and country-specific data. They also suggest integrating the anomaly scores into central-bank dashboards for real-time monitoring of capital-flow volatility. The paper will appear in Bundesbank Discussion Paper No 15/2026.