How Correlation Affects Implied Volatility of a Commodity Index

Summary

The document poses a portfolio-volatility question about the Bloomberg Commodity Index and its sub-sector indices. It supposes that the sub-sectors have the same average one-year at-the-money implied volatility and asks whether the index’s implied volatility should sit above or below that average when the components are uncorrelated. It also flags that correlation may behave differently in tail-risk scenarios.

The prompt highlights a key portfolio concept: aggregate volatility depends on component weights and co-movement, not simply on the arithmetic average of component volatilities. However, it supplies no answer, index weights, volatility aggregation calculation, or evidence about actual BCOM behavior. The tail-correlation caveat further limits any inference from the stated assumption of zero correlation. Readers can identify the question to investigate, but the document itself does not establish a numerical result or a general conclusion about the index’s implied volatility.

Key ideas

• Index volatility depends on component weights and correlations as well as constituent volatilities.
• The prompt asks how uncorrelated commodity sub-sectors affect aggregate implied volatility.
• It distinguishes ordinary correlation assumptions from possible changes in correlation during tail-risk scenarios.
• No calculation or answer is provided, so the relationship is left unresolved.

Tags

# commodities index volatility


# commodities index volatility












Suppose that the average one year implied at-the-money volatility of the sub sector indices making up the BCOM is at 20% and that the sub sectors are uncorrelated. Bearing in mind the effect of correlation in tail risk scenarios, would you expect the one-year at-the-money volatility of the BCOM to be higher or lower than the average of the sub sector volatilities (20%)?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.