Skip to content

Is Gold a Hedge or a Safe Haven? An Analysis of Stocks, Bonds, and Gold

Authors: Dirk G. Baur, Brian M. Lucey | Year: 2010 | Journal: Financial Review, 45(2), 217-229

Thesis

Gold is a hedge against stocks on average (negative correlation) but not against bonds. Critically, gold acts as a safe haven during extreme stock market downturns -- its correlation with equities turns significantly negative during the worst 1%, 2.5%, and 5% quantiles of equity returns. However, this safe-haven property is short-lived: it holds for about 15 trading days after a shock, then dissipates. The paper formalizes the distinction between a "hedge" (negative unconditional correlation) and a "safe haven" (negative correlation conditional on extreme losses).

Key Math

The core model is an asymmetric DCC-GARCH framework:

\[r_{\text{gold},t} = a + b_t \cdot r_{\text{stock},t} + \epsilon_t\]
\[b_t = c_0 + c_1 D(r_{\text{stock}} < q_{10}) + c_2 D(r_{\text{stock}} < q_{5}) + c_3 D(r_{\text{stock}} < q_{1})\]

Where \(D(\cdot)\) are dummy variables for stock returns falling below the 10th, 5th, and 1st percentile quantiles. Gold is a safe haven if \(c_0 + c_1 + c_2 + c_3 \leq 0\) (the conditional beta at extreme quantiles is non-positive). The DCC component captures time-varying conditional correlations:

\[Q_t = (1 - \alpha - \beta)\bar{Q} + \alpha \eta_{t-1}\eta_{t-1}' + \beta Q_{t-1}\]

Data & Method

  • Daily returns for gold (London PM fix), US/UK/German stock indices, and government bonds.
  • Sample: November 1995 to November 2005 (pre-crisis), with robustness checks on subsamples.
  • GARCH(1,1) for marginal distributions, DCC for correlations.
  • Quantile thresholds at 1%, 2.5%, 5%, 10% of the stock return distribution.
  • Rolling window analysis to test persistence of safe-haven effect (15-day, 20-day windows).

Our Replication Verdict

CONFIRMED -- The safe-haven effect at extreme quantiles is robust and has been replicated across extended samples including GFC (2008), COVID (2020), and the 2022 equity drawdown. Important limitations: (1) The safe-haven property breaks down during joint liquidity crises (March 2020 saw gold and equities sell off simultaneously for ~5 days before gold recovered). (2) Silver does NOT exhibit the same safe-haven property -- its industrial demand component creates positive beta with equities during stress. (3) The 15-day decay means this is a hedging tool, not a portfolio construction free lunch; you pay the carry cost of holding gold for a transient benefit.

Signal Mapping

  • Tail-risk hedging module (SS5.6). During detected equity stress regimes (VIX > 30, equity drawdown > 5%), the system increases gold allocation as a hedge overlay.
  • Gold/silver ratio signal: When equities are in the left tail, the gold/silver ratio spikes (gold outperforms silver). This ratio is a real-time stress indicator consumed by the regime detection engine.
  • Position sizing: The 15-day decay informs the holding period of tactical gold overlays -- positions initiated during stress are unwound within 3 weeks.

References

  • Baur, D.G. & Lucey, B.M. (2010). "Is Gold a Hedge or a Safe Haven?" Financial Review, 45(2), 217-229. DOI: 10.1111/j.1540-6288.2010.00244.x
  • Baur, D.G. & McDermott, T.K. (2010). "Is Gold a Safe Haven? International Evidence." Journal of Banking & Finance, 34(8), 1886-1898.
  • Reboredo, J.C. (2013). "Is Gold a Safe Haven or a Hedge for the US Dollar?" Finance Research Letters, 10(1), 11-22.
  • Bekiros, S. et al. (2017). "On the Time-Varying Relationship Between Gold and the Dollar." Physica A, 468, 554-562.