Risk　correlation　is　an　important　feature　of　financial　institutions　and　a　source　of　systemic　risk.　Despite　the　importance　of　risk　correlation,　few　studies　have　examined　its　mechanism.　Based　on　538,269　textual　risk　disclosures　of　1,953　financial　institutions　during　2006-2020,　this　study　uses　text　mining　methods　to　identify　the　risks　and　construct　risk　disclosure　similarity　measurements　to　verify　that　financial　institutions　with　similar　risk　disclosures　have　a　stronger　risk　correlation.　The　channel　analysis　reveals　that　when　different　financial　institutions　disclose　infective　risks　simultaneously,　the　market　perceives　their　susceptibility　to　common　factors,　leading　to　a　more　pronounced　left-tail　dependence.　Additionally,　a　cross-sectional　analysis　is　conducted　to　explore　the　heterogeneity　among　different　subsectors,　suggesting　that　risk　disclosure　similarity　within　the　same　subsector　is　higher　due　to　the　peer　effect.　However,　this　heightened　similarity　tends　to　mitigate　the　positive　impact　of　risk　disclosure　similarity　on　risk　correlation.　A　two-stage　regression　analysis　considering　endogeneity　supports　the　main　findings.　©　2023