Contagion Dynamics in Multi-Asset Cryptocurrency Markets: A Hawkes Process and DCC-GARCH Framework

Abstract Existing studies of cryptocurrency contagion typically analyse either event-driven shock propagation or time-varying correlations in isolation and often focus on small asset panels. This paper integrates univariate Hawkes intensity estimation, a pairwise cross-excitation layer, and a scalar DCC-GARCH model for twenty major cryptocurrencies over January 2021 to March 2026. The Hawkes component recovers a directed spillover network, while the DCC-GARCH component quantifies how quickly market-wide correlations respond to stress and how slowly they revert. We find that contagion transmission is led by medium-cap platform tokens, with MATIC, ATOM, ADA, and SOL generating the strongest outgoing excitation, while UNI, AVAX, XTZ, and LINK absorb the densest incoming spillovers. Correlation persistence is near-integrated ( $$\hat{a} + \hat{b} = 0.9739$$ ), implying a half-life of 26.2 trading days, well above standard equity-market benchmarks. Four contagion episodes are identified, ranging from a six-day China mining-ban shock to the 33-day LUNA collapse and a renewed macro risk repricing in October 2025. The results support network-aware portfolio construction, stress testing, and systemic-risk monitoring in cryptocurrency markets.