Benchmark-sensitive persistence in exchange-rate connectedness after the Russia–Ukraine conflict

Research on geopolitical shocks has mainly documented immediate market responses, leaving less evidence on whether exchange-rate connectedness remains persistently elevated once benchmark choice, volatility clustering, and sanctions-related pricing distortions are considered. Using daily observations from 15 January 2021 to 27 December 2024, this study analyses 14 consistently quoted USD exchange rates with a time-varying parameter vector autoregression and frequency-domain connectedness measures. Returns connectedness rises locally after 24 February 2022, but the persistence inference is strongly benchmark-sensitive. Post-conflict returns TCI exceeds the recent-pre-conflict mean-plus-one-standard-deviation and 90th-percentile thresholds on 68.33% and 57.01% of days, respectively, yet exceeds neither threshold when the full pre-conflict sample is used. In a dynamic model with Newey–West inference that controls for average FX volatility and lagged TCI, the post-conflict coefficient is positive but only marginally significant ( β = 0.0716, p = 0.0636), while volatility-based TCI shows no comparable post-conflict elevation. The DXY relationship is also regime-dependent: its full-sample slope is insignificant for returns connectedness but strongly negative for volatility connectedness, whereas the volatility slope becomes positive during the event window and marginally positive in the post-conflict period. No-RUB, dislocation-window, weekly-frequency, alternative-basket, and parameter checks indicate that the aggregate patterns are not attributable solely to RUB or one model specification. These findings support a bounded interpretation of benchmark-sensitive network persistence and sanctions-conditioned reconfiguration, rather than unconditional geopolitical hysteresis or a causally identified Federal Reserve channel.