Dual safe havens during turbulence? Wavelet-based TVP-VAR evidence from oil–gold dynamics under uncertainty shocks

Under the background of geopolitical risk and economic policy uncertainty, understanding the dynamic interaction between key commodity markets is essential. This study examines the time–frequency relationship between China's crude oil and gold under multiple sources of uncertainty. Using a comprehensive dataset spanning March 2018 to December 2025, we employ a wavelet-based time-varying parameter vector autoregression framework to derive several key insights. First, the wavelet correlation between crude oil, gold, and uncertainty is mainly dominated by medium and long-term scales. Second, the risk contagion among crude oil, gold, and uncertainty increases with the increase of the time scale, and extreme events will aggravate the risk contagion among variables. Third, EPU acts as a net contributor to risk in the short term, while GPR shows strong long-term risk spillover. Finally, there is a two-way risk contagion relationship between oil and EPU, and gold is mainly the recipient of uncertain shocks. These findings are highly relevant for designing effective hedging and risk-management strategies in the face of heightened uncertainty, particularly through positions in crude oil and gold. • Wavelet TVP-VAR reveals oil–gold–uncertainty links across time–frequency. • Oil, gold, and uncertainty correlations are dominated by medium/long-term scales. • Contagion intensifies at longer horizons; extreme events amplify spillovers. • EPU is a short-run net risk transmitter; GPR drives long-run spillovers. • Oil–EPU contagion is bidirectional; gold mainly absorbs uncertainty shocks.