Do oil price variations reliably affect returns for U.S. industry-level stock portfolios? In the June 2011 draft of their paper entitled “U.S. Industry-Level Returns and Oil Prices”, Qinbin Fan and Mohammad Jahan-Parvar apply several tests to investigate how oil price changes impact stock returns for 49 U.S. industries. They test economic significance by: (1) using a 60-month rolling historical window to model the predictive relationship between spot oil price changes and industry returns; (2) applying this relationship each month to the last observed oil price change to predict future industry returns; and, (3) investing in either industry portfolios or 4-week Treasury bills according to which has the higher expected return. They assume an industry portfolio-Treasury bill switching friction of 0.10%. Using monthly and weekly prices for West Texas Intermediate crude oil spot (January 1979 through January 2009) and nearest contract Cushing, Oklahoma light sweet crude oil futures (February 1986 through January), along with contemporaneous U.S. industry returns, they find that:
- Spot oil price changes have significant predictive power for some U.S. industry-level stock returns, but oil futures price changes generally do not.
- The relationship between spot oil price changes and future stock returns concentrates (based on statistical significance) in about a fifth of U.S. industries: autos, boxes, business services, construction, meals, personal services, retail, rubber and telecom.
- There is a delay of six to ten trading days between spot oil price changes and industry-level returns, and the effect does not persist more than a month.
- The spot oil timing strategy described above, equally weighted across all 49 industries, generates an average annual net return of 12.9%, compared to 10.7% for a comparable buy and hold strategy. The corresponding average annual net Sharpe ratios are 0.67 for the spot oil timing strategy and 0.48 for a buy and hold strategy.
- For the equally weighted nine industries with strong evidence of dependence on spot oil price changes, the oil timing strategy generates an average annual net Sharpe ratio of 0.72.
In summary, evidence indicates that changes in spot oil prices usefully predict some U.S. industry-level stock returns at an horizon of less than one month.
Cautions regarding findings include:
- Given the number of positions involved within industry portfolios, the assumed industry portfolio entry/exit friction of 0.10% may be quite optimistic over the 1984-2008 trading strategy test (see “Trading Frictions Over the Long Run”).
- The study appears to ignore the incremental and likely material trading frictions involved in monthly rebalancing across industries implicit in equally weighted outcomes.
- Analysis of subperiods to gauge trend in the predictive power of spot oil price changes for industry-level returns would be interesting.
