Changing behavior of crude oil futures prices

I’ve just finished a new research paper with my former student (and now University of Chicago Professor) Cynthia Wu. In our new paper, we study how increased purchases of crude oil futures contracts by financial investors may have affected the prices on those contracts.

A crude oil futures contract is an agreement between two parties to purchase oil at a future date at a price agreed upon today. For example, on Friday the November contract closed at a price of $88.18, meaning that if both parties were to hold on to the contract until expiry (which for this contract happens to be October 21), the seller would be obligated to deliver 1,000 barrels of crude oil to the buyer some time in November at a location in Cushing, Oklahoma at a price of $88.18 per barrel. Most people who buy or sell these contracts don’t actually hold them to expiry, but sell their positions to somebody else between now and then. The easiest way to think about this is if on next Wednesday, Friday’s buyer ends up selling the November oil contract back to the original seller, after which they both walk away clean. If the price of the contract goes up between Friday and Wednesday, the original buyer will have made a profit and the original seller will have made a loss.

If both parties don’t care about the risk, the price they agree to today would be the same as the price they’d expect to pay if they were just to wait until October 21 before making any agreement. However, providing a way of coping with risk that one of them would prefer to avoid is one of the principal reasons these contracts exist. For example, a producer of crude oil might be worried about its ability to maintain cash flow if the price of oil were to fall. They might then want to sell a futures contract. If the price of the oil they produce subsequently falls, the futures price will likely fall with it, and they will receive a cash gain from their futures position to offset the lower revenues from sales of oil. From their point of view, selling a crude oil futures contract is like buying an insurance policy. As with any insurance policy, if you’re risk averse you’d be willing to pay a premium for the policy that is greater than the expected value of the cash flow you receive from the policy.

But who would take the other side of the contract? The would-be buyer also faces various risks, and taking the long position in a crude oil futures contract may expose the buyer to risks that are hard for them to diversify away. If lots of oil producers are anxious to sell futures contracts for purposes of hedging, what we might see is that the price today of a November futures contract gets depressed a little bit below the rational expectation of what the price will actually be on expiry. That difference means that the buyer of the contract expects to receive some gain, on average, as compensation for the risk. The seller, on the other hand, is willing to pay that premium in order to achieve the desired hedging.

This was the basis for Keynes’ (1930) theory of backwardation, according to which hedging pressure from commodity producers might typically lead to a current futures price that is below the expected expiry price. Keynes’ theory has received empirical support from studies by
Carter, Rausser, and Schmitz (1983)
, Chang (1985), Bessembinder (1992), De Roon,
Nijman, and Veld (2000)
, and Acharya, Lochstoer, and Ramadorai (2010), among others.

The graph below plots the total open interest on NYMEX light sweet crude oil contracts since 1983. There was tremendous growth in use of these contracts after 2004. Part of this growth resulted from increased use of these contracts as a financial investment. One popular strategy is to buy a November contract today, and plan on selling that November contract in October in order to turn around and buy a December contract. Such a strategy would leave you always with a long position in oil, from which you would benefit when the futures price rises and lose when the futures price falls. The strategy might provide some portfolio diversification. For example, when oil prices rise, other assets in your portfolio such as stocks might do poorly. Price pressure from such purchases of futures contracts as an investment vehicle could have a similar effect as traditional hedging from producers but in the opposite direction. Purchases of futures contracts by financial investors basically forces arbitrageurs to take the short position on contracts, whose compensation might then come in the form of a futures price that is above the expected expiry price.



Total number of outstanding crude oil futures contracts of all maturities, daily observations March 3, 1983 to July 17, 2011. Vertical lines drawn at January 1, 1990 and January 1, 2005.
Source:Hamilton and Wu (2011).
hw4_oi.gif



In a new research paper with Cynthia Wu, we develop a simple model of how selling pressure from commercial hedgers or buying pressure from financial investors would interact with risk-averse arbitrageurs to determine the relations between futures prices of different maturities. One of our innovations is a full characterization of the dynamics and the role that risk premia seem to play in influencing those relations. We find that there have been very dramatic changes in these relationships over time, which we allow for by estimating different parameters for the model over the 1990-2004 and 2005-2011 subsamples.

The graph below plots our estimate of the risk premium associated with a contract that would expire 8 weeks from the indicated date, where the risk premium represents how much higher the price of the contract would have been at that date if there had been no hedging pressure from commercial producers or financial investors. For the period before 2005, the premium is uniformly positive. We calculate that the price would typically have been about 2-1/2% higher over this period in the absence of any risk aversion. This is consistent with Keynes’ theory, which would interpret that risk premium as the compensation to arbitrageurs for providing insurance to commercial producers.



Estimated risk premium on a crude oil contract that would expire 8 weeks from the indicated date, January 1990 to June 2011, quoted as a fraction of the current price, and based on models estimated separately on the two subsamples. Vertical line drawn at January 2005.
Source:Hamilton and Wu (2011).
hw4_tp.gif



The above graph also reveals how differently the risk premium seems to have behaved in more recent data. The average compensation to the long position is largely gone and even reversed in recent years, consistent with the view that arbitrageurs are now profiting not by buying from commercial hedgers but instead from selling to financial investors. However, the volatility of the risk premium is also significantly greater than it used to be.

Our paper is focused entirely on the relation between the prices of futures contracts of different maturities, and we do not attempt to make any statements about the more interesting question of what effects increased financial participation in crude oil futures markets may have had on the spot price of oil, a topic on which I offered some thoughts a few weeks ago. But our new findings are consistent with the claim that use of futures contracts as a vehicle for financial investment has influenced the dynamic behavior of futures prices.

15 thoughts on “Changing behavior of crude oil futures prices

  1. Bruce

    Financialization of the commodities sector; that is, commodities having become an “asset class” increasingly for leveraged speculators instead of primarily as business inputs for the production process.
    Producers (and non-financial intermediaries) now are forced to become commodities speculators to manage prices increasingly into the future, which contributes to increasing volatility and still more leverage.

  2. Vangel

    Isn’t the Cushing location a big problem? How can we try to draw any conclusions about the general market if there are such unique issues about deliveries to an area that has a number of significant constraints?

  3. westslope

    Great stuff JDH. You must be very proud of Professor Wu. Questions follow.

    Why divide the time series at 2005? It does not strike me as a natural break.

    With regards to volatility, would a chart of a model estimating 1990 to 2011 data clearly illustrate the increasing volatility?

    If the risk premia disappears as investors bid up future contracts, hedging producers should be benefiting, at the very least, if they are risk neutral. Consumers are paying higher prices or shouldering more risk or both. I should stop asking questions and read the paper…

  4. Steven Kopits

    Q4 2004 is when the oil supply stalls; hence using 2005 as a break-point makes sense from my perspective.

  5. Chris Cook

    This is interesting, particularly the changes since 2005.
    That was the year that ‘inflation hedging’ by exchange traded funds started in earnest, as Shell – transparently – caught on to what BP and Goldman had been doing opaquely for the previous ten years.

    Investors in these ETFs are the precise opposite of speculators intent on transaction profit. Their motive is to avoid loss, and they off-load dollar risk, and take on oil price risk. This is, of course precisely the opposite of producers who are offloading oil price risk and taking on dollar risk.

    What started to happen in earnest from 2005 onwards is that producers like BP and Shell were able to monetise oil in tank, or even stored for free in the ground, by leasing it to ETFs in exchange for an interest free loan. This was and is achieved by a sale and repurchase using an OTC forward physical contract.

    The physical market dog wags the futures tail and not vice versa: the exchange positions were hedges by the investment banks of OTC transactions as they facilitated this financial oil leasing.

    The sheer weight of financial flows into leasing positions drove up the price through to 2008, aided by large portions of investment bank hype. The bubble burst in 2008 after demand started to be choked off, and also following a (some say) manipulated spike representing a profitable trading coup.

    The speculators pulled out, and many flipped to short positions, and in late 2008 a lot of fund money pulled out as well and the market overshot on the down side to $30.

    At this point zero % interest rates and QE saw ‘inflation hedging’ set in on a massive scale, as investors bought anything but dollars.

    In my view the majors were simply incapable of accommodating these flows, since they didn’t have enough to lease.
    What then occurred, I believe, is that the Saudis accommodated investors by leasing oil to them within an agreed ‘peg’ set at mutually acceptable levels for the US and the Saudis.

    This unstable equilibrium ended with the Libyan supply shock and the post Fukushima demand shock, as speculators entered the market.
    I believe that the market price is being artificially supported within the opaque and increasingly illiquid Brent/BFOE complex, and that the recent backwardation is evidence that without a QE-fuelled flow into the market the overhang of forward sales depresses the forward price relative to the front months.

    In a nutshell, as with the tin market pre 1985, and the Hamanaka copper market, the oil market is suffering ‘macro’ manipulation on a cosmic scale.

    Cui bono from high oil prices? It’s the producers, stupid. It’s certainly not speculators, who don’t care what level the price is provided it’s moving in the right direction.

    As for the funds, I have been saying for a long time that ETFs are an accident waiting to happen. Inflation hedgers have ironically caused the very inflation they set out to avoid: and the investment banks probably set out the massive market risk on these instruments – which were being sold to risk averse investors, remeber – on page 10; section (b); clause (iii) of the boilerplate agreement.

    All this is about to end in tears.

  6. ppcm

    The financial results of financial corporations are not rewarding since on pure speculation and on a life span of 4 years, they hardly break even .http://www.gao.gov/new.items/d11529.pdf, and drive more doubts than hopes in Keynes theory. These results are consistent with previous empirical analysis made on banks forex proprietary trading (excluding customers based profits) on a long term they are not profitable,including “when the markets change, I change my mind, ” Keynes.This decade was a paradise for insiders trading,what if it changes?
    M Chinn “futures are not great predictors since they explain about 6% of the variation in changes in oil prices at the one year horizon”.
    More interesting is the correlation between the increasing futures in volume and prices,the figure 4 (Econbrowser Petroleum Prices and the International Dimension) shows that in absence of volume in the derivatives instruments the price of the oil remained stable 25 $ to 35 $ /b for five years in the row 2000 till 2005 (Yes then China India but not before).
    Do speculators obtain the right rewards as oil vectors, carrying prices convexity? the compression starting year 2005 is casting doubts.
    How geopolitical history 2000 2011 would have been written without the futures?

  7. jonathan

    Reading this lucid post makes me think: authors should be required to summarize every paper published in this manner. If you can’t say it clearly, then you should not publish it. Not limiting this to econ, but man oh man econ really needs this.

  8. JDH

    Chris Cook: Thanks for the very interesting information. What are the names of some of the ETFs to which you are referring? And are you using the term “leasing” literally, or rather as a functional description of what the arrangement amounts to?

    s jay: The Litzenberger and Rabinowitz paper, as well as much of the related literature, is addressing “backwardation” as normally defined as a relation between the spot price and the futures price. This determines the incentives for putting the physical product into inventory and/or leaving it in the ground. By contrast, our paper is looking at the relation between the futures price and the expected price at expiry. This is a separate arbitrage relation, in which one can participate solely using futures contracts without ever needing to take physical delivery of any oil.

  9. Eric

    Isn’t a potential problem with the interpretation of the risk premium analysis the structural break in the oil price time series in about 2005? This is right around when the time series stops being mean reverting and begins an upward trend. Since this upward trend is predominantly due to long-term changes in supply and demand and not long only financial speculation, couldn’t we also interpret the increase in risk premium volatility to changes in the market as opposed to the increase in long only speculation?
    To provide convincing evidence that speculators are driving the changes in risk premium it seems like it would make sense to look changes in premiums of commodities that haven’t experienced upward trends since 2005. (maybe nickel?)

  10. Santiago Prehoda

    I just wanted to comment and say that I really enjoyed reading your blog post here. It was very informative and I also digg the way you write! Keep it up and I’ll be back to read more in the future.

Comments are closed.