We're all taught that receiving $1 tomorrow is better than receiving $1 a year from tomorrow. Even if you can't use the $1 tomorrow, someone else likely can, and capital markets make it easy to get your money to them. That's why the interest rate yield curve is normally upward sloping, and the discount factor curve is almost always monotonic.
http://www.cmegroup.com/trading/interest-rates/us-treasury/2... (for the purists, I know there is a lot else going on in the treasury bond future curve, but it's hard to find a good DF curve)
In commodities, however, receiving a barrel of crude oil tomorrow is not necessarily more valuable than receiving a barrel a few months from now. If there is more oil available than people who can readily use it, then receiving crude oil tomorrow means storing it and potentially transporting it in the future. It's quite common for oil to be delivered next month to be cheaper than oil to be delivered the following month, and so on.
http://www.cmegroup.com/trading/energy/crude-oil/light-sweet...
Commodity ETFs at first attempted to buy a stake in the underlying commodity. It worked for easy to store items like gold or cash-equivalents. When they tried to do the same for oil or natural gas, it was too costly or too big a headache.
The work-around was to trade futures contracts for each of the commodities, and avoid ever taking physical delivery of millions of barrels of oil by selling contracts shortly before their delivery date and buying new contracts with further away delivery dates.
Unfortunately, the futures contracts already implicitly include the cost of storage and the investor aversion to taking physical delivery, making the ETF manager pay a cost every time they move from almost-expiring contracts to longer-dated contracts.
The rest of the article explains anecdotes about how a few large players take advantage of the situation. Some people can predict when large ETFs need to roll their contracts and profit by front-running. Others invest in efficient storage and transporting frameworks in order to take physical delivery and sell at a profit in the future.
My questions:
Why not minimize the front-running problem by making the dates you roll harder to predict? Instead of rolling the front month every month, why not include some basket of CL1 thru CL12 and find opportune times to roll? Once these ETFs got popular, why not invest in some of the storage infrastructure and charge a slightly higher fee but give much better tracking to investors?