wilmott.com

Hello everyone,I'm currently working on a research paper about cointegration on the commodity market. More precisely, I am trying to find out if the growing presence of Commodities Index Funds has had an impact on the cointegration of the underlying commodities. Any thought about this matter would be greatly appreciated.In order to guide the discussion, here are a few questions I am trying to answer myself:1/ I assume Commodities Index Funds have really started to have an impact on the market since 2005 - do you think it makes sense to try and find a change in cointegration relationships between commodities on such a short time frame (2005 - today)?2/ What would the rationale be for a new cointegration relationship between underlying commodities due to Commodities Index Funds? My thought is that they are long-term investors holding a broad range of underlying commodities, which eventually leads to a stronger correlation and/or cointegration (to be discussed) on the long run between these underlyings.I am open to discussion and hopefully some of you will be able to help me,Best regards,

It's been shown that ETFs have increased correlation among commodities by these two working papers, make sure you're not duplicating them...1. Ke Tang and Wei Xiong, ?Index Investment and Financialization of Commodities,? National Bureau of Economic Research Working Paper Series No. 16385 (September 2010), http://www.nber.org/papers/w16385.2. Kenneth Singleton, ?Investor Flows and the 2008 Boom/Bust in Oil Prices? (2011), http://papers.ssrn.com/sol3/papers.cfm? ... id=1793449.

Hi,Thank you for your answer. However I've already read these two articles, and although they are very interesting and well informed, they are mainly about correlation and not cointegration implied by commodities index funds.As you know of course, the two concepts are close but quite different at the same time.I was wondering if anyone had ideas about the impact of commodities index funds on cointegration (the impact on correlation is pretty clear indeed).Thank you in advance for your help.Regards

Hi,Here is a question to start the discussion: since commodities index funds are long only, they drive the prices of commodities futures up. Since the ponderation of these funds (e.g. GSCI) are more or less stable, these upward pressure is stable on the long term, hence creating a common long-term trend among the underlying commodities of the fund. Do you think that could "possibly" lead to a cointegration between these underlying commodities?Thanks

For energy commodities at least, in my opinion, cointegration relationships are primarily driven by fundamental relationships e.g. the relationship between power and gas is driven by the efficiency of CCGTs. When cointegation relationships change, I would expect that this is due to a change in the fundamentals. I am not sure that there would be much influence of commodity index funds on cointegration (although I agree completely with your distinction between corrleation and cointegration). Perhaps index funds can speed up or slow down the propagation of fundamental effects to the market (by sustaining or bursting price bubbles, for example), but other than that, I expect their role to be limited. I would be interested to hear if your research contradicts my views.

Hi Bazzat,Thanks a lot for your answer. Could you please elaborate on your example regarding energy and gas cointegration through efficiency of CCGTS? It seems interesting, however I lack the basic knowledge to understand it well.Thanks in advance.Any other input from anyone would be greatly appreciated.

CCGTs are gas-powered electricity-generators and depending on the particular generator it will do this conversion with some degree of efficiency. Older generators are typically less effective than newer ones. Efficiency is obviously a vaguely defined concept and mostly interesting for the engineers, but it suffice to say that when you put in one unit of gas you don't get the same energy equivalent out in electricity at the other end. Numbers range from high 40s to 60 (percent wise).Now, since in most countries you also need to include costs of CO2-emissions, you will have the relationship power_t = alpha + beta*gas_t + gamma*co2_t + err_t. Alpha will capture O&M etc, beta will reflect the overall efficiency and gamma the emission-costs. Power, gas and emissions all have a unit-root if you look at them separately, but this linear combination should not have a unit-root. I've not checked this myself, but it definitely seems reasonable.HTH, S

QuoteOriginally posted by: bazzatFor energy commodities at least, in my opinion, cointegration relationships are primarily driven by fundamental relationships e.g. the relationship between power and gas is driven by the efficiency of CCGTs. When cointegation relationships change, I would expect that this is due to a change in the fundamentals. I am not sure that there would be much influence of commodity index funds on cointegration (although I agree completely with your distinction between corrleation and cointegration). Perhaps index funds can speed up or slow down the propagation of fundamental effects to the market (by sustaining or bursting price bubbles, for example), but other than that, I expect their role to be limited. I would be interested to hear if your research contradicts my views.There are no power contracts in Commodity index futures or ETFs, so there is no "real" co-integration to be expected between power and gas (atleast due to index funds). Co-integration if any, can be expected between the components of the SP GCSI for example. Due to the existence of these index funds, as people buy more of these ETFs/futures... they are indirectly buying more energies, precious metals, agriculture, etc. But the amount of buying/selling of these index funds, has to EXCEED the net buys/sells of the individual components (that are fundamentally based) for it to appear as though there is cointegration.So even if you think there is co-integration among commodities, it would be relatively weak (i.e. you can expect the relationship blowout, imagine if someone did Gold vs. Silver cointegration, they would have been destroyed in May).Anyway, no one will be willing to fund any nonsensical co-integration strategies anymore, it's all a bunch of bogus technical analysis bullshit. Do it for your personal trading account, to prove to yourself why everytime a 2-3 standard deviation move occurs in your residual and doesn't revert back to the mean... that's when you go bankrupt (i.e. you will have to change your co-integration coefficient, i.e. there is a fundamental relationship change). You will nicely give the market back all the money you made.What Stale mentions is an example of fundamental co-integration, I don't know if his relationship is correct, but if its a structural relationship between the three variables, then you can expected residuals to be noise without a unit root. If it isn't a structural relationship, then it is just curve fitting.

Take care with the gas-power relationship because CCGTs can be switched on and off, so it also depends what other generation assets are tied to the grid. For example, if there is a lot of full Hydro storage in the system then this will be run instead of the more expensive CCGT, messing up any co-integration measurements. To get anything meaningful you are going to have to filter your data into various regimes. To get any sense out of an cointegration relationship between commodities and ETFs, then the size of the ETF funds must be sizeable in comparison to the underlying market.

It's necessary for different forms of power generation to be abke to be switched on or off for the whole cointegration idea to work.One of the interesting parts to aspects to this discussion is that one of traditonally power & gas price modelling has assumed demand is essentially exogenous (especially domestic heating diven demand) and hence the modelling is just a question of adding up the cheapest sources of supply till you get to demand and the price is simple the marginal cost of production. From this you simply get to the cointegration style price relations of power to gaswith some regime switching as the marginally generated megawatt switches fuel.However, add investor involvement and the whole thing ratchets up in complexity (think the fundamentals of crude oil or precious metals versus those of power).QuoteOriginally posted by: EdgeyTake care with the gas-power relationship because CCGTs can be switched on and off, so it also depends what other generation assets are tied to the grid. For example, if there is a lot of full Hydro storage in the system then this will be run instead of the more expensive CCGT, messing up any co-integration measurements. To get anything meaningful you are going to have to filter your data into various regimes. To get any sense out of an cointegration relationship between commodities and ETFs, then the size of the ETF funds must be sizeable in comparison to the underlying market.

QuoteOriginally posted by: twOne of the interesting parts to aspects to this discussion is that one of traditonally power & gas price modelling has assumed demand is essentially exogenous (especially domestic heating diven demand) and hence the modelling is just a question of adding up the cheapest sources of supply till you get to demand and the price is simple the marginal cost of production. From this you simply get to the cointegration style price relations of power to gas with some regime switching as the marginally generated megawatt switches fuel.However, add investor involvement and the whole thing ratchets up in complexity (think the fundamentals of crude oil or precious metals versus those of power).Demand is becoming less and less exogenous and more and more a negatively correlated function of price. I know some large companies that modulate their factories' power consumption on 5-minute intervals based on real-time electricity pricing. And at the consumer level, some utilities are experimenting with both time-variable rate structures (in which the utility tells consumers if today's price of electricity will be high due to high anticipated demand) and direct control of consumer demand (the utility installs a switch in people's homes that lets the utility modulate home air conditioner's on/off cycles to level the local load).

I think long-only funds do not just drive prices up- like any other normal investor they have trading limits; stop losses etc. so they would all be herding to rush for the door to unwind long positions when the market tanks.