Major Canadian news media, last week, triumphantly noted the latest research by Bank of Canada’s analysts in which they deflect blame from speculators being responsible in rising energy prices.
Titled The Role of Financial Speculation in Driving the Price of Crude Oil, Bank of Canada’s in-house researchers, Ron Alquist and Olivier Gervais, present a three-prong argument whose real aim may be to provide, veiled into statistics, defensive powder to Canada’s policy makers that have consistently defended market speculation be it in oil or in food.
Alquist and Gervais warn Canada’s policy makers that what they did is of tremendous importance for Canada:
Understanding the relationship between financial speculation and oil prices is an important issue confronting Canadian policy makers. Since oil makes up about 35 per cent of Canadian commodity production, understanding the effect of speculative financial flows on oil prices has immediate implications for the Canadian economy… relationship between energy prices and the Canadian dollar implies that fluctuations in oil prices affect the overall competitiveness of Canadian exports.
Since fundamentals (presumably) and not the speculation are behind the oil price rise, Alquist and Gervais also warn Canada’s policy makers not to fall for speculation blame because it may ruin efficiency of sectoral capital and labor allocation.
“A sectoral reallocation of capital and labour unrelated to fundamental macroeconomic conditions would be inefficient and may call for offsetting macroeconomic policy actions,” write Alquist and Gervais.
Every paragraph thereafter goes on to defend speculation.
Argument 1: Futures contracts and oil consumption cannot be compared because futures are stock while consumption is a flow.
Alquist and Gervais say that “it is common to compare the amount of open interest in the futures market to U.S. daily consumption and use the fact that open interest is many times larger than U.S. daily consumption to infer the presence of speculative pressures (Cho 2008; Masters 2008).” and in the footnote conclude that “Once the maturities of different contracts are reconciled, the volume of futures contracts for delivery in a given month was a fraction, rather than a multiple, of global physical production during 2007/08, the period when futures contracts were the most heavily traded (Ripple 2008, Table 2).”
Argument 2: Statistical testing methods show that changes in futures contracts and price are predictable both ways.
Researchers use data from the data come from the CFTC’s weekly Commitments of Traders reports to conduct a “bivariate and conditional Granger causality tests” in which they reject the hypothesis that changes in futures contracts predict price changes as well as reject a claim that changes in prices predicts positions. “Thus, the test for the full sample is inconclusive – the evidence suggests that the direction of predictability goes both ways,” write Alquist and Gervais.
Argument 3: If speculators drove the oil price, then that would be reflected in increased hoarding of oil stocks but evidence does not show any oil inventory accumulation.
“If non-commercial firms had created expectations of persistently high prices, it would have been more natural to observe the accumulation of stocks and an upward sloping futures curve. Furthermore, in the United States, where the data on privately held stocks of oil are the most reliable, inventories of oil during the first half of 2008 were below those prevailing in 2005/07,” write the researchers.
Authors note that the oil inventories, blue line, fell dramatically during the 2008 price spike while production remained steady.
As a result of Arguments 1, 2 and 3, an explanation for the surge in oil price has to be found elsewhere – like low real interest rates and increase in demand – and regrettably, Alquist and Gervais do not convincingly find that any of these “alternative” causes did anything to the price of oil.
Of course, all of this fancy statistical analysis, and ultimately all the conclusions, hinge on the choice of the time interval used and the authors admit that.
“It is important to stress, however, that this rejection of the null hypothesis is highly sensitive to the lag-length specification and sensitive to the start date of the sample. For instance, using the Schwarz criterion and/or starting the sample in 1996 would result in non-rejection of this null hypothesis.” [my emphasis]
If choice of the time interval matters, then why mesh the oil bubble that occurred in 2008 with data going back to 1993?.. unless, perhaps, that a stylized dataset is the only way to obtain facts necessary to support the pre-existing conclusion.
In fact, the value of the t- statistic authors use to validate or reject a claim is obviously higher for the shorter interval they present – from 2003 to 2008.
But even this interval is stylized so that 2008 ends in June omitting the period when the oil price really spiked – after June – thus fueling additional suspicion that facts are harvested here to support a pre-existing conclusion.
The text of Alquist and Gervais oil price study is available here.
Additional readings on oil speculation:
- The Accidental Hunt Brothers
- Black Gold & Fool’s Gold: Speculation in the Oil Futures Market
- Stock Price Manipulation by Hedge Funds
- Causes and Consequences of the Oil Shock of 2007–08
- How Wall Street Speculation is Driving Up Gasoline Prices Today
- Peak Energy and the Limits to Global Economic Growth