By John Kemp
LONDON, May 9 (Reuters) - Research and analysis into commodity prices and derivatives almost always relies on the commitments of traders (COT) report published by the U.S. Commodity Futures Trading Commission (CFTC).
Except for some correlation studies published by the European Central Bank and International Monetary Fund, every major report on the formation of commodity prices, supply/demand, the role of index funds, and the possible “financialisation” of the asset class, published in the last five years, has relied on data taken from the COT report.
Most big investment banks in commodities, as well as smaller brokers and dealers, publish an analysis of the weekly COT data for their clients, and COT numbers appear somewhere in the tables and charts included in regular reports.
But how much does the commitments of traders report really reveal about activity in commodity derivatives markets?
The report is less insightful than many regular users realise. But in a classic example of availability bias the data is (mis)used as a proxy for speculation and hedging, and over-interpreted to track the correlation between prices and positions, simply because it is the only information researchers have.
The COT report provides almost the only information on the type of traders and size of their positions in the otherwise secretive futures and options markets for energy products like crude and natural gas, metals such as copper, and agricultural items ranging from grains to livestock and coffee.
The report’s origins can be traced back to 1924, when the U.S. Department of Agriculture began publishing an annual report on hedging and speculation in regulated futures markets. It has since appeared with increasing frequency, becoming monthly in 1962, bi-monthly in 1990, fortnightly in 1992 and weekly from 2000.
As trading on regulated exchanges has spread from grains and other agricultural products to oil and refined products, natural gas and power, precious and base metals, currencies, equity indices, bonds, and volatility, more and more commodities and other financial assets have come within the scope of the reporting system.
Until recently, no other regulator or exchange published data on traders and positions in comparable detail. By default, therefore, the COT report has become the cornerstone of research into how supply, demand and inventories, as well as speculation and hedging, affect cash and futures prices, for everything from retail gasoline to steak on restaurant menus and the cost of breakfast cereal.
The report’s main limitations and problems stem from the way the CFTC classifies types of traders and aggregates information on their positions to preserve the anonymity of market participants.
Under the CFTC’s large-trader reporting system (LTRS), all clearing members of exchanges, futures commission merchants and foreign brokers must file reports with the CFTC showing futures and options positions they hold on their own account and on behalf of each client above specific reporting levels set by the Commission.
Reports are filed daily, but the COT report is based on returns for positions held at the close on Tuesdays. Reported positions usually represent 70-90 percent of all live contracts (open interest) in any given market.
All traders with a position above the threshold must file a Form 40. It provides details about their enterprise, and identifies whether they are using futures and options to hedge business activities (which Form 40 defines as including “production, merchandising or processing of a cash commodity, asset/liability risk management, security portfolio risk management, etc”).
Based on Form 40, and the judgment of CFTC staff, all traders with reportable positions are assigned to one of more than 20 detailed categories set out in Schedule 1 of Form 40 and in Appendix 1 of a CFTC report on “Fundamentals, Trader Activity and Derivative Pricing” published in December 2008.
The information in the daily reports enables the Commission to track the positions of individual traders or classes in real-time and considerable detail, identifying unusual trading patterns or the accumulation of positions (though as Amaranth illustrated, having the data is not the same as acting on it in an effective and timely manner).
By law, the CFTC is required to preserve the confidentiality of individual traders, so most of the detail in the LTRS is suppressed in the published COT reports.
Until 2009, the CFTC aggregated large traders into just two categories: commercial traders (assumed to be hedging) and non-commercial traders (who were not). The remaining (small) positions were allocated to a “non-reportable” category.
Unfortunately, as Form 40 makes clear, “hedging” covers a wide range of activities. The “commercial” category therefore contained firms that were hedging in the popularly accepted sense (producers, consumers, merchants hedging operational inventories) as well as businesses managing financial risks (such as index operators hedging commodity-price related transactions they had done with pension funds).
In 2008, following the commodity price spike, the CFTC decided to improve the report by offering more detailed categories. Swap dealers, which use futures and options to “hedge” their financial rather than physical exposure to changing commodity prices, for example as a result of offering commodity index products, were removed from the commercial category.
The CFTC began publishing a disaggregated COT report which classified traders into four major groups: (1) producer/merchant/processor/user; (2) swap dealers; (3) managed money; and (4) other reportables. The remaining small positions are again classified as “non-reportable”.
In theory, this should offer greater clarity. The producer/merchant/processor/user category corresponds more closely to the traditional idea of hedging physical transactions. Financial hedgers and dealers have been separated out into the swap dealer category. The old “non-commercial” speculators are reported in the remaining two categories of money managers and other reportables.
The data is still too aggregated to be useful. The major problem is that the CFTC classifies traders rather than positions.
Take a major oil company, which engages in operational hedging, offers “risk management services” to its clients, and runs a speculative trading book. All its positions will be allocated to just one of the four categories, depending on whether the CFTC decides it is primarily a producer, a swap dealer or some other type of trader. The result is that most categories contain a mixture of hedging and speculative positions.
The CFTC does not reveal how individual firms have been classified. For example, we don’t know if BP, Shell, Exxon and Chesapeake are classified as producers, swap dealers or in some other category. In 2008, the CFTC had to retrospectively reclassify one large trader, changing all the reports going back more than a year, after it discovered they could no longer be described as hedging because they had disposed of their refinery.
As a result, research that attempts to separate out the effect of hedging versus speculation on prices, using either the old two-way commercial/non-commercial classification, or the new four-way one, is based on flawed data.
The second problem is the CFTC does not break down the age distribution of positions. In 2008, CFTC economists provided a rare glimpse of the maturity profile, based on the full confidential LTRS database. It showed swap dealers and money managers accounted for most contracts with one-three years left to maturity and almost all open interest beyond three years ().
But the data has been reported only for three points in the past and is not regularly updated. It is impossible to replicate the exercise because the full data set remains confidential and the CFTC will not release it to other researchers, even anonymised.
In its current form, the COT data provides an incomplete and often misleading picture of the distribution of positions in the futures and options markets. Unfortunately, most of the research which has been published based on it is worthless because it is based on over-broad classifications (the “garbage in, garbage out” principle).
How could the COT data be improved? The first step would be to publish a full list of how individual traders have been classified, with a more detailed explanation of classification policy. The second step would be to start breaking positions down by age distribution (perhaps in broad buckets like the 2008 report).
The most radical step would be to publish data on individual positions with an appropriate delay to ensure they are no longer commercially sensitive. Publication after three or five years of individual positions, or at least classifications, would not expose any commercial secrets. But it would allow researchers to get a much better understanding of how commodity prices have been formed in the past.
Some of these changes could be implemented by the Commission itself. Others might require legislation. But if the COT data is ever to yield the sort of useful conclusions that academics, commodity analysts and policymakers want, much more disclosure will be necessary. In the meantime, users should treat the COT data with care.