What share for gold? On the interaction of gold and foreign exchange reserve returns

BIS Working Papers  |  No 906  | 
26 November 2020



Gold investments still form a significant share of central banks and governments' FX reserve portfolios. We ask whether this high share is justified from a risk-return standpoint, while investigating the variety of factors that make this a difficult question.


We make three contributions. First, we focus on how gold affects portfolios formed purely of fixed income assets, as these more closely resemble those managed by central banks and governments in practice. Second, we analyse a broad range of risk-return measures, over and above the typically applied mean-variance framework. Third, we go beyond the discussion of what is optimal for portfolios on average (as commonly seen in the literature) to focus on what might be optimal in extreme cases, ie at the tail of the risk distribution. This is of great interest to reserve managers.


From a market risk perspective, a low-duration, reserve currency fixed income portfolio may benefit only from very small gold allocations (between 0% and 5%), on average. Nonetheless, sizeable gold holdings may be justified, from a purely quantitative standpoint, for portfolios with a higher duration and for reserve managers who measure their returns in a non-reserve currency. In addition, when looking at the benefits of gold as a protection against an extreme event, we find that high allocations (of between 20% and 50%) may be adequate in some cases. Our results suggest that choosing an appropriate share for gold in reserve portfolios is a complex task. The answer depends crucially on both the purpose (policy objectives) and implementation (numéraire, risk tolerance etc) of the reserve management process.


Almost five decades after the collapse of the Bretton Woods system, gold continues to form an important share of global foreign exchange reserves. This may be because gold has traditionally offered reserve managers many benefits, such as the absence of default risk. This paper explores whether these large investment shares in gold are also justified from a risk-return standpoint, or whether any other explanations have to be brought to bear. To do this, we go beyond the simple application of portfolio optimisation techniques, comprehensively analysing all possible long-only combinations of gold and representative fixed income reserve portfolios. We conclude that the market risk associated with gold is substantial when evaluated against a broad range of criteria, such as mitigating portfolio volatility, tail-risk, the probability of loss, and measures of diversification. This will tend to limit overall allocations. Nonetheless, for portfolios with higher sensitivity to interest rates (duration) and for reserve managers who measure their returns in a non-reserve currency, we find evidence that gold may function as a hedge, making it easier to justify sizeable gold holdings from a purely quantitative perspective.

JEL Codes: E58, F31, G11, G17.