No labels, no problem: Identifying investment fund cohorts through clustering

This paper applies a tailored clustering approach to identify cohorts of investment funds for financial stability assessment. To define clusters, we use regulatory data on asset class exposures reported under the Alternative Investment Fund Managers Directive (AIFMD). We applied our approach to more than 10,000 alternative investment funds (AIFs) holding €3.7 trillion in assets, revealing 12 economically interpretable fund cohorts, including traditional bond and equity funds, liability-driven investment (LDI) funds, and private asset funds. Our cluster-based approach substantially outperforms traditional AIFMD categories in explaining fund return variance and points to material vulnerabilities, notably concentrated leverage in GBP-denominated LDI funds and widespread liquidity mismatches. The dispersion of these cohorts across EU jurisdictions underscores the need for oversight and cross-border coordination in ensuring the macroprudential oversight of the investment fund sector. This framework provides regulators, supervisors and macroprudential authorities with a practical framework for identifying funds whose collective behaviour could amplify systemic risks during periods of market stress. JEL Classification: G01, G11, G18, G23, C38