Can safe-haven assets still protect investors? Evidence from dynamic connectedness and R^2-decomposed portfolio strategies
Abstract
Rising global crises and heightened market uncertainty have elevated the role of safe-haven assets in portfolio stabilization. This paper focuses on four major safe-haven assets, namely, Bitcoin, Gold, JPY.USD, and CHF.USD and investigates the direction and the extent to which information transmission among the four assets changes over time. Using the DCC-GARCH $$\text {R}^{2}$$ decomposed connectedness framework, We uncover marked heterogeneity and regime dependence in shock transmission: Bitcoin persistently functions as the primary net receiver, Gold, JPY.USD and CHF.USD switch between receiver and transmitter roles contingent on market conditions. The information embedded in the connectedness analysis is then used to construct two minimum-connectedness portfolios (bivariate and multivariate) designed to minimize bilateral and system-wide shock transmission. Based on our results, the two connectedness-based portfolios show superior in-sample risk-adjusted performance compared to conventional minimum-variance, minimum-correlation, and static hedging benchmarks in terms of risk-adjusted return and hedging effectiveness. Strikingly, whereas variance-minimizing strategies marginalize Bitcoin, connectedness-based portfolios assign it substantial weight, confirming its significant contribution to portfolio diversification within the examined network. This result also underscores the limitations of static covariance-based approaches for resilient portfolio construction in volatile markets. Overall, our findings carry direct implications for investors pursuing robust risk-adjusted performance across market regimes and for policymakers tasked with monitoring evolving systemic risk transmission among safe-haven assets.