Bad data problem worsens as oil prices skyrocket
Bad data problem worsens as oil prices skyrocketAs oil prices have climbed to 10-year highs, the global markets need for timely, accurate production data is greater than ever. Yet, the quality and quantity of output statistics are worse than ever the "missing barrels" phenomenon is alive and wellGeorge S. Littell, Groppe, Long & Littell, Houston nxiety over predictions of Y2K problems in computer systems worldwide led millions of consumers to greet year 2000 with full gasoline tanks in their automobiles. Reported deliveries and inventories were affected in both 1999 and 2000. Meanwhile, none of the other problems with published statistics have been resolved. The "missing barrels" are still missing, and there are more of them. Largely by default, the monthly Oil Market Report (OMR) compiled by the International Energy Agency (IEA) has become the source of published oil balances not only for the financial community but also for the industry, itself. IEA is also a statistical agency. Based on the data available, recent oil balances particularly stock movements have differed considerably with the OMR version, as shown in Table 1.
After running slightly high in 1997 and 1998, IEAs estimate of global oil supply was about 600,000 bpd too low last year. Apparently sensing that it was underreporting supply, IEA seems to have overcompensated this year, with the OMR estimate now running too high again. The discrepancies in stock movements are even more pronounced. While estimates at Groppe, Long & Littell (GL&L) showed only 300,000 and 500,000 bopd were added to global stocks in 1997 and 1998, respectively, OMR reported increases that were 1.4 million and 1.5 million higher. Then, almost inexplicably, OMR data reversed course and underreported net stock declines in 1999 by 200,000 bopd. IEAs inconsistency is alive and well this year, as the agency is 900,000 bopd off of GL&Ls estimate. OMR data are showing stocks as building, rather than continuing last years decline. The Y2K Effect A large number of apprehensive consumers took pains to begin year 2000 with full fuel tanks in their cars, trucks, houses, etc. Normally consumers tanks are a little more than half-full. Oil statistics are collected from producers, refiners and marketers. The Y2K effect in 1999 manifested itself as a large reduction in industry stocks at the end of the year. In addition, abnormally high reported deliveries were reported for the fourth quarter. In 2000, the Y2K effect has been the reverse low reported deliveries and not much inventory drawdown in the first quarter, as shown in Table 2.
If it turns out that consumption of products increases by 800,000 bpd in 2000, versus 1999, then the Y2K effect will make reported deliveries about the same for both years. An example can be seen in Table 3.
The Missing Barrels The large balancing items in OMR stocks 1.4 million bopd in 1997 and 1.7 million bopd in 1998 for a total of 1.13 billion bbl are often called the "missing barrels." They are, in concept, increases in stocks in parts of the world not covered by the statistical system. Changes in stocks in the member-nations of the Organization for Economic Cooperation and Development (OECD) are usually small numbers over the course of a year. The Y2K effect was, of course, an extremely unusual event. In 1999, OMR reported no missing barrels. However, in first-quarter 2000, another 500,000 bopd have turned up missing. That phenomenon can be explained by the figures in Table 4. Analysis shows that OMR should have reduced its figures for production and deliveries, and increased estimates of processing gains.
Basic Concepts Before continuing a critique of IEAs methods in OMR, a quick review of certain market definitions and fundamentals is in order. For instance, the production of oil has three categories:
Refineries convert crude oil, condensate and NGLs into useful products, such as gasoline, naphtha, jet fuel, diesel, residual fuel oil, lubricants, wax, asphalt, coke, etc. In addition to the international trade in crude oil, condensate and NGLs, there is extensive commerce in refined products. The simple basis for calculating oil balances is to begin with production, i.e., the "production basis." In the production basis, deliveries for a country or area are obtained by this formula:
The more difficult method is to begin with deliveries of refined products and NGLs, or the "deliveries basis." Via this method, the formula for determining balance is:
Aside from the small amounts of hydrogen that may be added, refining processes do not change the weights of the hydrocarbons refined. Many of the processes (cracking, reforming, etc.) result in products that are less dense than the input. The volumes of products from a refinery not their weights typically exceed the input volumes of crude oil, condensate, etc. by 1% to 6%, depending upon the combination of processes employed. That difference between output and input volumes is termed processing gain. Processing Gains For the entire world, the recent history of processing gains is displayed in Table 5. Statistics for the U.S. collected by the Energy Information Administration (EIA) are by volume (bbl), as is true for Canadian statistics (cu m). For other OECD members, statistics are by weight (metric t), although it is easy enough to convert them to volume and then calculate processing gains. For the rest of the world, exact calculations can be made through 1996. The estimates for 1997 and later are reasonably close.
Looking at the data, global processing gains reported in OMR are ridiculously low they are even less than the total GL&L reports for just OECD countries. If one assumes that EIAs figures for deliveries and production are on target in OMR, then the balancing item for stocks should be negative (i.e., phantom drawdowns). To achieve a balance between deliveries and supply, the other options are:
In practice, OMR has employed the third option. Figures for deliveries in non-OECD countries which are mislabeled "demand" are calculated on the production basis, while the OECD numbers are close to being figured on the deliveries basis. This mixture is inaccurate, although the result would result in a balance the error in deliveries being offset by the error in processing gains provided that the production figures are accurate. Definitions Affect Output Estimates The "missing barrels" (i.e., positive balancing items in stocks) result from OMRs overstatement of production. For 1999, OMRs production estimates are compared with those by EIA, Oil and Gas Journal (O&GJ) and British Petroleum (BP) in Table 6. One should note that OMR routinely reports more production both OPEC and non-OPEC than anyone else would believe exists.
A better approach is to begin with trade statistics imports and exports and work back toward estimates of production. Large amounts of money change hands in the oil trade, and most of it is routine. For 1999, our efforts at GL&L to track trade flows resulted in the production figures listed in Table 7.
Whether or not either bitumen or tar should be classified as crude oil can be debated. If they are classified as such in Canada the usual practice then they should be treated the same in Venezuelan data, as well. In the U.S., EIA carefully labels what is usually termed crude oil as production of crude oil and "lease condensate." In EIAs system, lease condensate refers to the volume of liquids recovered at lease separators about 470,000 bpd in 1999. Condensate separated at the inlets of gas processing plants 200,000 bpd or so is classified as part of "pentanes plus," which is predominantly natural gasoline. OPEC adopted a formal definition of condensate in 1988, because the groups quotas apply only to crude oil. That action ended an attempt by Venezuela to reclassify a substantial volume of light crude oil as condensate. In addition to bitumen, Venezuela is notable for production of heavy oil. Whether all or some of that output is something other than crude oil has never been debated. In assessing OPEC production, it is natural for the trade press to follow non-OPEC statistical convention and include condensate in figures for crude oil. Algeria is a notable exception, because condensate is rarely blended with crude oil there. Furthermore, Algerian production of both condensate and NGLs is almost never discussed with the media. Statistics Do Matter OPEC controls the oil market through its members production quotas. Because oil consumption does not change much over a wide range of prices, relatively small mistakes in setting quotas can have a big impact on prices in either direction. At the November 1997 meeting in Jakarta, Indonesia, the belief among OPEC ministers that they were already producing 27.2 million bpd of crude was a basic rationale for raising quotas to a total of 27.5 million bpd. Disaster quickly ensued, even before Iraq escaped the UN oil embargo in June 1998. The production "cuts" of 1998 resulted from a wildly exaggerated, February production number published ad nauseam by the media. Actual OPEC output was up in 1998, not down versus 1997. In March 1999, OPEC ministers finally reduced actual production and were surprised by the results. Including bitumen from Venezuela, OPEC oil production in 1999 was almost exactly equal to the average quotas that comprised the 25.93-million-bpd target. Reports of production of 26.5 to 26.8 million bopd in the media gave rise to allegations of quota-cheating. At the lower end, the 600,000-bopd difference is about equal to the volume of condensate produced by OPEC members, excluding Algeria. If OPEC ministers believe OMR and the media, then it will be easy to set quotas too high. It is very possible for these ministers to cause another period of low prices, even though most of OPEC members are operating at, or close to, oil production capacity.
|