Investors often use energy ETFs as proxies for the energy markets, but how well do these funds really track their underlying commodities?
- The problem with futures-based funds
- Why you can't use USO and UNG as direct proxies
- Does the 12-month solution work?
For the average investor, the commodities markets can be daunting. Many investors don't possess the capital and/or risk tolerance necessary to invest directly in the futures markets. Further, although the number of online brokerages that offer futures trading is growing, for the most part, buying or selling commodity futures requires setting up a special account, which, for many retail investors, may be more trouble than it's worth.
Enter the commodity ETF.
Commodity ETFs are relatively new, but that hasn't stopped them from attracting investors—in 2009 alone, commodity exchange-traded products brought in $30.1 billion in new investment. Although some precious metals funds actually hold the goods in question—the SPDR Gold Trust (NYSE Arca: GLD), for example, actually keeps gold bars in a vault—this strategy is impossible for most commodities, whose worth is determined by their usefulness. As such, commodity ETFs purporting to track prices in other classes of good require some type of derivative investing; in the commodities world, that means futures contracts.
However, by their very nature, commodity ETFs allow investors to make longer-term bets on the value of a given commodity than is typical in the futures market. The futures trader makes money simply if the price of oil for delivery in a given month goes up between now and that month's contract expiration. But for the fund investor, who buys shares of an oil ETF expecting the price to rise over the next several months or years, the case is not so cut and dried.
As we've covered before, it comes back to the structure of the fund. If, for example, an ETF holds only the front-month contract for a given commodity, then that contract must be sold before its expiration, and the next (soon to be front-month) contract must be purchased in its place. Since the two contracts are almost certain to have some price discrepancy, the difference on this roll must be factored into the return of the fund.
What's more, a futures-based commodity ETF is part of a secondary market on top of the already volatile futures market. As investors buy and sell shares of what is essentially ownership of other financial instruments, the fund's value may fall out of whack with the commodity it is purporting to track. After all, demand for an ETF may be very different than demand for the futures contracts that ETF tracks: Just look at the U.S. Natural Gas Fund (NYSE Arca: UNG).
With these potential pitfalls in mind, let's compare the relationships between several energy commodity ETFs and the front-month prices of the commodities they purport to track. A regression analysis should help us get an understanding for how well (or poorly) a given ETF truly tracks a commodity.
Let's look at the funds with the highest trading volume in oil, natural gas and gasoline: the U.S. Oil Fund (NYSE Arca: USO); the U.S. Natural Gas Fund (NYSE Arca: UNG) and the U.S. Gasoline Fund (NYSE Arca: UGA). Each of these funds invests in the front-month contract only, reflecting the simple, long and unleveraged approach. All data is close of day from the fund's inception to 2/12/2010, as compared with equivalent sessions for each commodity.
Crude Oil: United States Oil Fund (USO)
An R_squared of .7944 means that roughly 80% of USO's change in price can be attributed to its relationship with oil prices. If you consider the fund a proxy for buying oil directly, that is far too low a correlation to be seen as true investment in the commodity.
Looking at the graph, there's a near perfect straight line when the price of oil is above $80/barrel, indicating a tight correlation. However, at the lower end of the spectrum, two other trend lines have formed with very different trajectories. While analyzing each of these components is best saved for another time, we can generally conclude that USO is prone to erratic shifts off its intended link.