This is a guest post by David Frenk, Executive Director of Better Markets:
Last week, OECD published a report co-authored by two Illinois professors, Scot Irwin and Dwight Sanders. The report, entitled Speculation and Financial Fund Activity, purports to find statistical evidence that speculation played no role in generating the damaging volatility in food and energy prices witnessed during 2008-9. In fact, it claims that speculation by long-only index investors with no understanding of underlying supply and demand conditions actually helped reduce volatility, by providing liquidity.
The study and its findings can be disregarded for three reasons:
- The statistical methods applied are completely inappropriate for the data used.
- The study is contradicted by the findings of other studies that apply more appropriate statistical methods to the same data
- The overall analysis is superficial and easily refuted by looking at some basic facts.
Please find attached a review of the study, which goes into more detail on these points:
(Many thanks to Babak, Prof. Wei Jong, and Prof. Kenneth Medlock, all of whom granted permission to reproduce figures.)
Specifically, the authors of the OECD study use a Granger causality test, which is a perfectly good statistical tool in its own right (we have used it ourselves in other contexts). However, Granger tests can only be used for certain kinds of data. They are known to give unreliable results when applied to highly volatile data sets like commodities prices. In fact, there are several academic papers that argue convincingly for the inapplicability of Granger tests to stock and commodities price data.
Second, even disregarding the inapplicability of the test itself, the test parameters used in the OECD test are inappropriate: they mostly study fluctuations over the space of a week, whereas those who argue speculation has affected commodities prices are generally talking about a much broader time period.
Finally, a Granger test can only be interpreted with reference to some underlying theory. As an alternative to the view that speculation drove commodities prices in 2008, the OECD study claims that prices, including those of wheat, cotton and oil, were purely determined by supply and demand from producers and consumers. As the review discusses, EIA data shows conclusively for the case of oil that supply was rising while demand was falling during the fastest price rise in history, which indicates strongly that supply and demand from consumers and producers was not the driving factor.
Enjoyed this? Don't miss the next one, grab the feed or