Financial Contagion: Detecting Non‐Simultaneous Breaks in DCC ‐ GARCH Models
提出三步分割程序（TSSP），用SupLM检验检测DCC-GARCH模型中波动率和相关性的非同步结构断点，识别传染事件。蒙特卡洛模拟验证效果，实证发现全球金融危机、欧债危机及2022年加息与乌克兰战争相关的波动转移和传染。
ABSTRACT This paper proposes a three‐step segmentation procedure (TSSP) for detecting non‐simultaneous structural breaks in return volatility and correlations within DCC–GARCH models, using the supremum Lagrange multiplier (SupLM) test to isolate multiple parameter shifts. By detecting breaks in unconditional correlations, our method identifies potential shift‐contagion episodes. Monte Carlo simulations demonstrate the TSSP's robust performance in detecting and locating both successive and common breaks affecting different subsets of parameters. Empirical application to equity and government bond returns in advanced and emerging economies reveals volatility shifts linked to the Global Financial Crisis and shift‐contagion associated with the European Sovereign Debt Crisis, and the post‐Covid‐19 pandemic interest rate hikes alongside the war in Ukraine in 2022.