Whenever a crisis hits, it is likely to spread simultaneously among stock markets due to their interconnection. This phenomenon, known in the literature as financial contagion, may have a long-lasting effect and manifest itself in herd behavior. A sample of worldwide MSCI (Morgan Stanley Capital International) indices covering the subprime, European, and COVID-19 crises is used to study the presence of contagion, its transmission channels, and potential herd behavior. These channels—real linkages, financial mechanisms, or investor beliefs—are proxied by macrofinance factors. Their impact on stock correlations is tested using an extension of the dynamic conditional correlation model that includes external regressors. Results show evidence of contagion during the three crises, driven mainly by investor expectations and their effects on volatility. Moreover, during the COVID-19 crisis, the dissemination of information, as measured through text-based indices, played a significant role in market contagion and led to herd behavior. • Evidence is unveiled of financial contagion/herd behavior during the three crises. • Contagion transmission channels were studied with an extension of the DCC model. • In general, investor expectations/volatility (VIX) act as main contagion drivers. • Belief channels (text-based indices) explain contagion during the COVID-19 crisis. • Herd behavior (news-driven) is detected mainly after the COVID-19 outbreak.