Analysis · Finance
How Gold Changed the Benchmark for Explaining Stock Returns
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
Changing the benchmark reduced the number of unexplained returns in a study of 49 US industry portfolios. In their 2023 paper, Muhammad Abdullah and colleagues found that replacing Treasury bill returns with gold returns improved several stock pricing models. The finding concerns how historical returns are measured and explained.
The researchers used 420 months of data, from January 1981 through December 2015. They combined dollar-denominated London bullion prices with US equity portfolios grouped by industry, company size and other characteristics. They compared conventional models with versions that subtracted gold returns from both portfolio returns and market returns instead of subtracting the one-month Treasury bill rate.
Their hypothesis was that gold could serve as a zero-beta benchmark: an asset whose returns have little systematic relationship with market returns. Gold’s estimated market beta was 0.027 and was not statistically distinguishable from zero over the sample. However, its monthly returns were much more volatile than Treasury bill returns. A near-zero beta does not establish that an asset is risk-free.
For the five-factor model applied to industry portfolios, the number of statistically significant pricing errors fell from 18 to 11. These errors, called alphas, represent average returns the model leaves unexplained. The researchers also examined differences in average returns across portfolios, using a two-stage regression procedure with corrections for uncertainty in estimated exposures.
The gains were uneven. Many gold-based models still failed tests asking whether pricing errors were jointly zero, and some comparisons showed little improvement. The evidence covers US equities during one historical period; it does not demonstrate future forecasting accuracy or establish that gold causes stock returns. Proposed uses during pandemics remained conjectures.
The practical lesson is about benchmark sensitivity. An apparent unexplained return can change when the reference return changes. Gold offers an alternative specification to investigate, while the remaining errors and limited setting constrain what that comparison establishes.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Assessing the Use of Gold as a Zero-Beta Asset in Empirical Asset Pricing: Application to the US Equity Market M. H. Abdullah, Hussein A. Abdou, Christopher Godfrey …
