Forever Blowing Bubbles
Was it speculation rather than economic fundamentals that caused the boom and bust in the oil price in 2008 and 2009? According to Daniel O?Sullivan, the statistics make a convincing case for that argument.
Throughout the great oil price blow-out of 2008 ? in which crude first breached $100 per barrel in January, soared to almost $150 in July, but then abruptly collapsed to $35 by Christmas ? there was constant criticism of data on the NYMEX oil futures trade setting the benchmark global oil price made available by the Commodity Futures Trading Commission, the US regulator.
Argument was raging over whether speculation was driving the oil price, but the CFTC?s figures did not sufficiently differentiate between genuinely commercial oil market participants and financial sector institutions such as hedge funds and so-called investment bank ?swap dealers?. This opacity was a key theme in my book Petromania: Black Gold, Paper Barrels & Oil Price Bubbles, in which I argue, from a wealth of evidence, that triple-digit oil prices were indeed the result of a classic asset price bubble, driven primarily by financial-sector speculators.
Yet I am not actually going to detail here the arguments in Petromania. Instead, I will show something I simply could not when writing in early 2009, as the data required was not available.
In October 2009, the CFTC both began releasing data showing specific trader classes and also provided three years-worth of previous data likewise disaggregated.
So we now have some broad-brush information on how different classes of market participant traded both through the obvious speculative bubble, and since then. Using even simple statistical tools on this data, there is striking proof not only that financial sector-led speculation drove the 2008 oil price bubble, but also drives prices now.
Our two charts show the oil price in BLACK across two separate 39-week periods, plus three other lines showing what three ?nested? sets of variables ? extracted both from CFTC disaggregated trader data and other key statistical sources ? say what the oil price should have been if we were only relying on these indicators.
The relationships between these indicators and the oil price are modelled through fairly standard statistical number-crunching, specifically multiple linear regression analysis. One chart, Through the peak, covers the period from start-2008 to end-September 2008, thereby capturing the window through which oil first broke through $100, charged up to its $140-plus peak and then collapsed just as quickly back to sub-triple-digit levels. The other chart, The recent rally, sketches a period running from early June last year (2009), to the most recent available data, late February this year (2010).
Many commentators (myself included) reckon this latter period ? which now sees oil about $80 per barrel despite an OPEC spare production capacity buffer of monumental proportions, high inventory levels and persisting global economic weakness ? also shows evidence of speculation overtaking fundamentals. Yet, being an identical number of weeks to our earlier period, it also provides a useful ?control? comparison. Why not just one whole timespan from start-2008 to now? From the end of September 2008 into early 2009, the aftermath of the Lehman Brothers bankruptcy wreaked such havoc globally, that all players, commercial or speculative, were liquidating all assets and acting on extremely bearish suppositions. Structurally, this was a completely different market environment from that seen up to October 2008, and from mid-2009 onwards.
Trying to model a continuous, consistent set of behaviours across these ruptures is pointless.
Instead, the question is: do the variables that seem to most fully ?explain? the oil price movement through the height of its boom and into its subsequent bust (just up to where the global economy itself imploded), also explain the oil price through the same length of time well over a year later, despite everything that happened in-between?
The answer is yes, with the variables crucial to the most accurate prediction of the oil price across both periods being those relating to the NYMEX market positioning of both investment bank swap dealers and hedge funds ? together, two of the most obvious conduits for financial sector speculation on the oil price.
In both charts, the green line shows the oil price as predicted by supply and demand ?fundamentals?. These are represented by changes in inventory and end-demand figures supplied by the US government Energy Information Administration (EIA), whose weekly bulletins are closely watched on NYMEX, and also by the change in net positioning (expressed in contracts to buy versus contracts to sell) held by commercial NYMEX market players such as producers, refiners and physical traders. Their positioning should capture any physical market fundamentals beyond EIA figures.
Across both periods, the green line shows these fundamentals as a poor predictor of the actual oil price ? although, in the second chart it at least manages to generally shadow an upward drift in prices, which is unsurprising given that the macro-environment since mid-2009 has been one of recovery from incredibly weak demand.
The blue line meanwhile adds to supply and demand fundamentals, two other variables widely seen as linked to the oil price ? dollar strength (reflected in the euro-dollar forex rate) and the gold price. Conventionally, as the dollar weakens, gold tends to strengthen and so too does oil.
Some might say allowing these elements a significant role in oil price formation is already normalising speculation. I myself am not so doctrinaire ? while dollar strength and the gold price certainly do not affect physical oil supply, and while financial speculators will indeed often tie a bearish view on the dollar with a bullish punt on gold and/or oil, there are nevertheless good arguments as to why these monetary factors can have a justifiable effect on oil pricing. Yet, while echoing oil?s peaks and troughs better than the green line, the blue line is also not a good predictor.
It is only by adding in a third set of variables (in the red line) that our model generates a predicted oil price close to the actual oil price. The fit is extremely good through the peak, in statistical terms showing an ?R-squared? of 0.83 compared to 1.0 being a perfect match (also known as the coefficient of determination, R-squared is a statistical expression of how closely two sets of figures relate to each other, and provides a good guide to the effectiveness of predictive models). For the recent rally, the fit is not so close, yet still compelling with an R-squared of 0.62.
The fresh variables link the changes in net NYMEX positioning across the combined swap dealer and hedge fund grouping with the size of their commitments relative to the overall market. That these variables add such explanatory power after already controlling for fundamentals, the genuinely commercial player positioning and dollar and gold prices alike, shows it is these trading patterns in themselves that are the crucial inputs.
Simply put, supply and demand fundamentals explain little, gold and dollar moves little more, but the way financial speculators adjust their market positioning is crucial in explaining oil price movement.
DANIEL O?SULLIVAN is a journalist covering the energy and mining sectors for the past decade. He is currently employed by Energy Intelligence. His book, Petromania (£20) is published by Harriman House. For details see www.harriman-house.com
If you’d like to get in touch with the author for interview or comment, or you’d like a review copy of this book, please contact us at email@example.com or call +44 (0)1730 269809.Rights
For information on available rights, please contact firstname.lastname@example.orgBulk purchases
Discounts for bulk purchases available. Please contact email@example.com for a quote.