11 ms·

Amber Waves of Pain (Do Not Buy Commodity ETFs)

62 points by leelin 16y ago

_delirium 16y ago
The biggest gap in expectations seems to be that people think they're buying a share in some sort of underlying commodity, as opposed to a rolling series of 1-month futures. It'd be as if S&P 500 index ETFs, instead of holding a weighted index of the S&P 500 stocks, were just buying and rolling over SPY options. Any particular reason they're 1-month contracts, though? It seems the treading-water aspect would be greatly reduced if these funds could trade, say, 1-year futures. Do longer-dated futures just not exist, or at least not exist in markets that are liquid enough?
- hristov 16y ago
  The problem with one year futures is that they do not track the spot price of oil as well. The whole purpose of an oil ETF is for it to give you the ability to buy or sell a bunch of oil at any time without having to deal with the actual oil. Thus, the value of an ETF should track the current (spot) price of oil. But if the fund buys one year contracts, their costs will not reflect the current price but what people believe the price will be at the end of the year. So the ETF will be mis-priced in relation to the current price. I think that is why they try to buy contracts that are relatively short, because they track the current price better.
  - Confusion 16y ago
    The whole purpose of an oil ETF is for it to give you the ability to buy or sell a bunch of oil at any time without having to deal with the actual oil. If that's the purpose, then they were flawed when conceived, because, as the article shows, the ETF´s cannot abstract from the costs involved with dealing with the actual oil.
- biotech 16y ago
  They do both. For example, compare the ETFs "United States Oil Fund" (ticker USO) with "United States 12 Month Oil Fund" (ticker USL). Every month, USO buys the next month's future as the previous month's future expires. Also every month, USL buys the next year's future as the future bought a year previous expires. You can look at their charts here: http://bit.ly/ar6Kxm http://bit.ly/ar6Kxm
leelin 16y ago
Quick summary: 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-year-us-treasury-note.html 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-crude.html 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?
- Retric 16y ago
hristov 16y ago
This is the reality of Wall Street, any time you tie your self to a certain predictable future behaviour, people will figure that out and figure out a way to exploit you. Those exchange traded funds screwed themselves because they committed themselves to buying contracts at a specific time. So other traders knew about it and made sure the prices of the contracts at that time were extra high. And it does not seem like there is a good answer. On one hand you can have the ETFs stop their requirements to buy or sell at a particular time. Thus you would entrust the managers of those funds to find good times to buy and/or sell. But then again can you trust the managers? There is a huge incentive for the managers to take bribes from another trader to buy or sell at times that are bad for the fund but good for the other trader. Or, maybe the managers simply are not very good and choose bad times. In a way you have an agency problem, and ETFs were invented in order to remove any agency issues. That is, the whole philosophy behind ETFs is for them to be automatic so you do not have to trust the judgement of some manager to buy them. I think the only answer is for investors is to buy the contracts directly and not to go through ETFs. Of course that is a bit dangerous, because if you do not clear your position in a timely manner, you always risk somebody actually delivering a thousand barrels of oil on your front lawn, but hey that is a risk you have to take if you want to be commodities investor.
- leelin 16y ago
  This is a very interesting problem. Basically the algorithm for when to roll needs to be: 1.) Random and hard to predict before-the-fact 2.) Totally verifiable after-the-fact by the public (or by an uninterested authority the public trusts) That means a randomized algorithm whose seed is something the ETF manager knows before the general public, but which everyone else can verify once they publicize the info. Quick brainstorming: 1.) They seed based on a hash of their total AUM or PNL on a certain date, down to the penny, which is almost impossible for outsider's to predict but which they know internally and can publicize in the next quarterly statement. 2.) They tell the SEC a list of state-run lotteries in secret, then the seed is the winning lottery number on a particular night. The public can never know which lottery is the one that matters until after-the-fact. 3.) Seed based off a hash of the concatenation of every employee's birthday or social security number, which is never public, but if there was ever a serious investigation or audit, they could verify to authorities that they followed the rules.
joubert 16y ago
This also illustrates the important point of only ever investing in stuff if you understand them and how they're packaged.
perkoff 16y ago
It does not have anything to do with contango. In a paper called "Facts and Fantasies about Commodity Futures" (Yale, 2004), Gorton and Rouwenhorst came to the following conclusion after studying 45 years of data on a wide array of commodities: - "commodities that have been more backwardated (by the second definition) have not earned larger historical returns" - "During our sample period, this commodity futures risk premium has been equal in size to the historical risk premium of stocks (the equity premium), and has exceeded the risk premium of bonds." George Rahal wrote a paper recently that also showed that contango/backwardation did not have any affect on commoditiy returns: http://www.hardassetsinvestor.com/features-and-interviews/1/2072-george-rahal-contango-has-no-effect-on-commodity-returns.html http://www.hardassetsinvestor.com/features-and-interviews/1/... "The other thing is, commodities are an artifact of futures, so they appear to have more volatility in the front month than in the back. Volatility is not the friend of a long-term index-type investor. So an index that avoids the front month is ideal." There are some commodity ETFs that are not suitable for long term investors though. UNG, which tracks US Natural Gas, generally have bad returns. This is mostly due to the frequent turnover, since it holds the front month futures.
heatdeath 16y ago
I've held GSG for a year or so and wondered why it wasn't gaining while gold, oil, and everything else was going up. Is there any way to invest in oil using a standard brokerage account? Doesn't seem like it.
- piguy314 16y ago
  Precious metal ETFs seem to track spot much better due to, as has been said, the cheap cost of storing gold/silver. Plus the stated goal of a precious metal ETF is to buy and hold precious metals as a hedge against inflation, their aim is to make each share worth exactly 1 ounce of whatever metal they are buying, so there is no contract turnover on futures. I know Peter Schiff, who some people say is a scam artist, prefers to buy physical gold, gold ETFs, and mining companies. Probably the only way to buy and hold and invest in oil is to buy oil company stocks.
heyrhett 16y ago
It is possible for the reverse of contango to happen (called backwardian) where the etf would actually go up in the short term while the commodity stays flat. Ultimately though, the reason why contango has to win in the long run is that there is a real cost associated with holding oil. Contango should be a lot cheaper with gold, since it's cheaper to store gold. The amount you might lose to contango in an ETF is probably a lot cheaper than if you tried to horde a bunch of oil in a tank on your property anyway, so it's not really a terrible investment. Moral of the story: don't invest in instruments you don't understand.
tocomment 16y ago
So it sounds like these ETFs aren't a good investment. Do you guys have any recommendations for good investments that also offer reasonable protection from inflation? It's a shame about these oil ETFs being no good. It seems like the perfect investment as I can't imagine the price of oil ever going down in the long term.
- binaryfinery 16y ago
  It does not seem like a perfect investment. The only way it can seem like a perfect investment is if you do not investigate what it is. How many other of your investments are "perfect" do you think? And why would you be worried about inflation?
  - lsc 16y ago
    because it is in the interest of perhaps the majority of Americans to have inflation, and because it seems that the fed is trying to generate inflation. On the other hand, there are some pretty big deflationary forces (like, say, bad loans) at work here, so being /certain/ of inflation is probably a bad idea, but I can see why a person my worry that we might have inflation.
    - binaryfinery 16y ago
      I don't think the Fed is trying to generate inflation, simply because if they wanted to they could. If you look at their own reports, however, we clearly don't have inflation and aren't going to for a long time. About 20 trillion dollars have been wiped out of the money supply. The trillion dollars of printing so far is like hoping taking a piss will turn a tsunami. And most of that trillion dollars has been taken by the banks and put right back into the federal reserve system. Check out banks' reserves. Any money the Fed has given out hasn't affected spending in any way. There is a way they could get inflation going: if the government prints money (i.e. not borrows printed money) and gives it away to people, e.g. by lowering taxes a lot. But they can't. And we all know they cant.
      - 16y ago
zach 16y ago
There was a very clever hack that econ professor Robert Shiller (of housing-market fame) came up with to avoid having to deal with secondary markets like futures trading. It goes like this: there are a pair of ETFs, with equal amounts of outstanding shares. Each share in an ETF represents one portion of a claim on $120 (for example). But the exact amount that each of the two is entitled to seesaws based on some external variable, such as the future market for crude oil. So if a barrel of crude is priced at $40, one fund (the "up" ETF) will be valued at $40 per share and its paired fund (the "down" ETF) will be valued at $80 a share. The shares come into existence in pairs and can only be redeemed at the end of the fund's lifetime. As you can see, this offers a nice solution to the horrendous problems seen in the article. However, when this idea was actually turned into a real product (MacroShares), it ran into real-world problems again and again. Their first oil funds had to be closed early because it was actually based on that $120 figure -- and when the price of oil shot up close to that level, it triggered an automatic termination. And when they got their $200-a-barrel-max follow-on running, it had lost a lot of the momentum of the earlier funds. But their major problem was that their complexity was all in full view of the traders, and people found it hard to really understand. Because even though its current asset value was based on the price of oil on a commodity exchange, the "up" version often had a very large premium over that price (and the "down" ETF a discount). That was the effect of contango -- people were figuring in the rising future value of the oil. So in fact, it was actually providing a more accurate price than the commodity market for the storage-free value of a barrel of oil! Now, that's fascinating and all, but investors wanted a product based on the commodity market value, so it made things annoyingly obscure. Add to that the really complex situation when factoring in the possibility of early termination -- in which case the future price doesn't matter at all because the funds are paid off according to the commodity price at the point of termination -- and things just got to be too much. MacroShares ended up closing their oil funds in the middle of last year, replacing them with a pair of US city real estate index (yes, the Case-Shiller index) ETFs which seemed interesting but were wildly unpopular. It looks pretty bad for the concept right now, but they are probably just waiting in the wings for the next opportunity since they can provide a way to invest in things that otherwise would never be feasible.
sabj 16y ago
This is a fascinating article, and a good summary about contango -- a quite fascinating topic that has been exploited for great profit and success to those with large amounts of capital and savvy to deploy in the last decade, especially around the various times of great volatility. What I _don't_ appreciate is an article on the web artificially sliced into 7 pieces for more pageviews :( Thanks BusinessWeek... "print article"

62 points by leelin 16y ago

The biggest gap in expectations seems to be that people think they're buying a share in some sort of underlying commodity, as opposed to a rolling series of 1-month futures. It'd be as if S&P 500 index ETFs, instead of holding a weighted index of the S&P 500 stocks, were just buying and rolling over SPY options. Any particular reason they're 1-month contracts, though? It seems the treading-water aspect would be greatly reduced if these funds could trade, say, 1-year futures. Do longer-dated futures just not exist, or at least not exist in markets that are liquid enough?

hristov 16y ago
The problem with one year futures is that they do not track the spot price of oil as well. The whole purpose of an oil ETF is for it to give you the ability to buy or sell a bunch of oil at any time without having to deal with the actual oil. Thus, the value of an ETF should track the current (spot) price of oil. But if the fund buys one year contracts, their costs will not reflect the current price but what people believe the price will be at the end of the year. So the ETF will be mis-priced in relation to the current price. I think that is why they try to buy contracts that are relatively short, because they track the current price better.
- Confusion 16y ago
  The whole purpose of an oil ETF is for it to give you the ability to buy or sell a bunch of oil at any time without having to deal with the actual oil. If that's the purpose, then they were flawed when conceived, because, as the article shows, the ETF´s cannot abstract from the costs involved with dealing with the actual oil.
biotech 16y ago
They do both. For example, compare the ETFs "United States Oil Fund" (ticker USO) with "United States 12 Month Oil Fund" (ticker USL). Every month, USO buys the next month's future as the previous month's future expires. Also every month, USL buys the next year's future as the future bought a year previous expires. You can look at their charts here: http://bit.ly/ar6Kxm http://bit.ly/ar6Kxm

leelin 16y ago

Quick summary: 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-year-us-treasury-note.html 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-crude.html 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?

Retric 16y ago

hristov 16y ago

This is the reality of Wall Street, any time you tie your self to a certain predictable future behaviour, people will figure that out and figure out a way to exploit you. Those exchange traded funds screwed themselves because they committed themselves to buying contracts at a specific time. So other traders knew about it and made sure the prices of the contracts at that time were extra high. And it does not seem like there is a good answer. On one hand you can have the ETFs stop their requirements to buy or sell at a particular time. Thus you would entrust the managers of those funds to find good times to buy and/or sell. But then again can you trust the managers? There is a huge incentive for the managers to take bribes from another trader to buy or sell at times that are bad for the fund but good for the other trader. Or, maybe the managers simply are not very good and choose bad times. In a way you have an agency problem, and ETFs were invented in order to remove any agency issues. That is, the whole philosophy behind ETFs is for them to be automatic so you do not have to trust the judgement of some manager to buy them. I think the only answer is for investors is to buy the contracts directly and not to go through ETFs. Of course that is a bit dangerous, because if you do not clear your position in a timely manner, you always risk somebody actually delivering a thousand barrels of oil on your front lawn, but hey that is a risk you have to take if you want to be commodities investor.

leelin 16y ago
This is a very interesting problem. Basically the algorithm for when to roll needs to be: 1.) Random and hard to predict before-the-fact 2.) Totally verifiable after-the-fact by the public (or by an uninterested authority the public trusts) That means a randomized algorithm whose seed is something the ETF manager knows before the general public, but which everyone else can verify once they publicize the info. Quick brainstorming: 1.) They seed based on a hash of their total AUM or PNL on a certain date, down to the penny, which is almost impossible for outsider's to predict but which they know internally and can publicize in the next quarterly statement. 2.) They tell the SEC a list of state-run lotteries in secret, then the seed is the winning lottery number on a particular night. The public can never know which lottery is the one that matters until after-the-fact. 3.) Seed based off a hash of the concatenation of every employee's birthday or social security number, which is never public, but if there was ever a serious investigation or audit, they could verify to authorities that they followed the rules.

joubert 16y ago

This also illustrates the important point of only ever investing in stuff if you understand them and how they're packaged.

perkoff 16y ago

It does not have anything to do with contango. In a paper called "Facts and Fantasies about Commodity Futures" (Yale, 2004), Gorton and Rouwenhorst came to the following conclusion after studying 45 years of data on a wide array of commodities: - "commodities that have been more backwardated (by the second definition) have not earned larger historical returns" - "During our sample period, this commodity futures risk premium has been equal in size to the historical risk premium of stocks (the equity premium), and has exceeded the risk premium of bonds." George Rahal wrote a paper recently that also showed that contango/backwardation did not have any affect on commoditiy returns: http://www.hardassetsinvestor.com/features-and-interviews/1/2072-george-rahal-contango-has-no-effect-on-commodity-returns.html http://www.hardassetsinvestor.com/features-and-interviews/1/... "The other thing is, commodities are an artifact of futures, so they appear to have more volatility in the front month than in the back. Volatility is not the friend of a long-term index-type investor. So an index that avoids the front month is ideal." There are some commodity ETFs that are not suitable for long term investors though. UNG, which tracks US Natural Gas, generally have bad returns. This is mostly due to the frequent turnover, since it holds the front month futures.

heatdeath 16y ago

I've held GSG for a year or so and wondered why it wasn't gaining while gold, oil, and everything else was going up. Is there any way to invest in oil using a standard brokerage account? Doesn't seem like it.

piguy314 16y ago
Precious metal ETFs seem to track spot much better due to, as has been said, the cheap cost of storing gold/silver. Plus the stated goal of a precious metal ETF is to buy and hold precious metals as a hedge against inflation, their aim is to make each share worth exactly 1 ounce of whatever metal they are buying, so there is no contract turnover on futures. I know Peter Schiff, who some people say is a scam artist, prefers to buy physical gold, gold ETFs, and mining companies. Probably the only way to buy and hold and invest in oil is to buy oil company stocks.

heyrhett 16y ago

It is possible for the reverse of contango to happen (called backwardian) where the etf would actually go up in the short term while the commodity stays flat. Ultimately though, the reason why contango has to win in the long run is that there is a real cost associated with holding oil. Contango should be a lot cheaper with gold, since it's cheaper to store gold. The amount you might lose to contango in an ETF is probably a lot cheaper than if you tried to horde a bunch of oil in a tank on your property anyway, so it's not really a terrible investment. Moral of the story: don't invest in instruments you don't understand.

tocomment 16y ago

So it sounds like these ETFs aren't a good investment. Do you guys have any recommendations for good investments that also offer reasonable protection from inflation? It's a shame about these oil ETFs being no good. It seems like the perfect investment as I can't imagine the price of oil ever going down in the long term.

binaryfinery 16y ago
It does not seem like a perfect investment. The only way it can seem like a perfect investment is if you do not investigate what it is. How many other of your investments are "perfect" do you think? And why would you be worried about inflation?
- lsc 16y ago
  because it is in the interest of perhaps the majority of Americans to have inflation, and because it seems that the fed is trying to generate inflation. On the other hand, there are some pretty big deflationary forces (like, say, bad loans) at work here, so being /certain/ of inflation is probably a bad idea, but I can see why a person my worry that we might have inflation.
  - binaryfinery 16y ago
    I don't think the Fed is trying to generate inflation, simply because if they wanted to they could. If you look at their own reports, however, we clearly don't have inflation and aren't going to for a long time. About 20 trillion dollars have been wiped out of the money supply. The trillion dollars of printing so far is like hoping taking a piss will turn a tsunami. And most of that trillion dollars has been taken by the banks and put right back into the federal reserve system. Check out banks' reserves. Any money the Fed has given out hasn't affected spending in any way. There is a way they could get inflation going: if the government prints money (i.e. not borrows printed money) and gives it away to people, e.g. by lowering taxes a lot. But they can't. And we all know they cant.
    - 16y ago

zach 16y ago

There was a very clever hack that econ professor Robert Shiller (of housing-market fame) came up with to avoid having to deal with secondary markets like futures trading. It goes like this: there are a pair of ETFs, with equal amounts of outstanding shares. Each share in an ETF represents one portion of a claim on $120 (for example). But the exact amount that each of the two is entitled to seesaws based on some external variable, such as the future market for crude oil. So if a barrel of crude is priced at $40, one fund (the "up" ETF) will be valued at $40 per share and its paired fund (the "down" ETF) will be valued at $80 a share. The shares come into existence in pairs and can only be redeemed at the end of the fund's lifetime. As you can see, this offers a nice solution to the horrendous problems seen in the article. However, when this idea was actually turned into a real product (MacroShares), it ran into real-world problems again and again. Their first oil funds had to be closed early because it was actually based on that $120 figure -- and when the price of oil shot up close to that level, it triggered an automatic termination. And when they got their $200-a-barrel-max follow-on running, it had lost a lot of the momentum of the earlier funds. But their major problem was that their complexity was all in full view of the traders, and people found it hard to really understand. Because even though its current asset value was based on the price of oil on a commodity exchange, the "up" version often had a very large premium over that price (and the "down" ETF a discount). That was the effect of contango -- people were figuring in the rising future value of the oil. So in fact, it was actually providing a more accurate price than the commodity market for the storage-free value of a barrel of oil! Now, that's fascinating and all, but investors wanted a product based on the commodity market value, so it made things annoyingly obscure. Add to that the really complex situation when factoring in the possibility of early termination -- in which case the future price doesn't matter at all because the funds are paid off according to the commodity price at the point of termination -- and things just got to be too much. MacroShares ended up closing their oil funds in the middle of last year, replacing them with a pair of US city real estate index (yes, the Case-Shiller index) ETFs which seemed interesting but were wildly unpopular. It looks pretty bad for the concept right now, but they are probably just waiting in the wings for the next opportunity since they can provide a way to invest in things that otherwise would never be feasible.

sabj 16y ago

This is a fascinating article, and a good summary about contango -- a quite fascinating topic that has been exploited for great profit and success to those with large amounts of capital and savvy to deploy in the last decade, especially around the various times of great volatility. What I _don't_ appreciate is an article on the web artificially sliced into 7 pieces for more pageviews :( Thanks BusinessWeek... "print article"