Confessions of a statistician
In its August 2004 issue of the Oil Market Report (OMR), the IEA posed the question whether the current oil market situation justify $45 oil and stated the result as irrational exuberance. What did the IEA do in order to justify that fundamentals are there? It adjusted Non-OECD demand upwards in OMR. It also revised NGL figures for many countries.
The editorial of Arab Oil and Gas Journal on 16 December 2005 stated that “The development of oil prices since 2003 has followed a rising curve that can in no way be justified solely by the so-called market "fundamentals", such as the volumes of demand, supply and stocks.” Right! Even though prices have remained at high levels even during periods when supply has comfortably exceeded demand and when stock levels have been rising, even to their highest level in the past 6 years, almost everybody still tries to point out fundamentals. When all the fundamentals looked good enough, people blamed insecurity, geopolitical tensions, speculators, peak oil analysts, foreign exchange market, weather, you name it.
What is wrong with fundamentals? Well, we know that oil market fundamentals are interpreted by looking at the world oil supply demand balances. And there we better ask what is correct in those balances. What do we really know about oil market fundamentals then? Well, certainly not much!
Oil market reports of the IEA and EIA are by far the most recognized and carefully analyzed sources of the oil market fundamentals, and are believed to be a leading indicator about the oil price movements. That is why the world oil balance produced by them is closely watched by many in the oil markets, including Wall Street and particularly OPEC.
Missing Barrels Phenomenon: The Oil Market Report of the IEA was blamed for the price panics of 1999. The report went under attack when it showed 1.9 Mb/d stock built and miscellaneous to balance for the year 1998 in its Annual Statistical Supplement - the perceived existence of ‘missing barrels.’
Several speculations have been made about ‘missing barrels.’ Mabro stated that there was an upward bias in IEA’s production estimates. Horsnell claimed that ‘missing barrels’ were almost certainly comprised of firstly an underestimation of World oil demand, and secondly an overestimation of OPEC supply. In addition, Matthew Simmons emphasised that ‘missing barrels’ simply cannot be true because it is not possible to hide that much oil even though there exists many off-shore storage facilities, meaning something is wrong with the data. Recently, those “missing barrels” came back on the world oil balances and yet nobody makes any comment.
In fact, there has never been “missing barrels,” there has been missing statistics!
Here I will explore a few items about data quality on market fundamentals by pointing out the pitfalls and by focusing on overseen, neglected, misunderstood, mispresented, not properly handled or missing actors whose role and magnitude are not insignificant at all.
We all know that, in general, oil data are not very transparent and reported figures for some countries are mostly guesstimates (OPEC Members) and come from third party sources like independent tanker traffic consultants. Just check what the sources of data for OPEC members are in OPEC reports. In cases where everything seems to be transparent (e.g., OECD Members) it is still possible to end up with different figures for the same thing.
Let us have a look at some important items in oil balances:
Production of oil generally refers to the marketable production within national boundaries including off-shore. There exist some countries, however, that underreport production due to the exclusion of the production by concession companies (e.g., in Egypt, at least years up to 2002). Timing, revisions, and discrepancy in definitions and coverage are only some of the reasons. A. Littlle claims that definitions matter a lot especially for OPEC countries where quotas apply to crude oil only. Laherrere also discussed the quality of oil production data in several articles. But …
In most world oil supply demand balance tables you will see liquid fuels produced from coal, gas and from renewables (e.g, ethanol, biodiesel) added into total demand. Some call them oil from non-conventional sources. Fine. But how do you add them into oil supply? Supply must balance demand, no? One remedy is to add the exact amount into supply, and call it supply of all liquids (including NGLs, petrochemical feedstocks and refinery gains), knowing that not all supply come from crude oil. Then why do we call the end result as world oil supply demand balance? Existing world oil supply demand balances do not yet tackle this issue properly. Some would argue that the share of liquids from non-conventional sources is not that important. Yes, but how long?
Furthermore, it is not correct to make a peak oil analysis based on “all liquids,” because the non-conventional fuels I mentioned above do not come from an oil reservoir.
International trade statistics are by-products of customs procedures. They are prepared from customs declarations filled in by exporters and importers and verified by customs authorities. Under perfect conditions one would expect the following statement to be true: Export of country A to B equals to the import of country B from A. Unfortunately, in practice this rarely holds. Unless there is an item called ‘unallocated’, whatever publication you consult, you should always see a discrepancy between world oil exports and imports. Some hints: for example, for NGL, heavy fuel oil and LPG.
Besides fiscal and political reasons, in many countries imports may not even include the imports of private companies (e.g. in India for oil and oil products), in some countries trade is underreported simply because of black market, or exports are underreported due to possible cheating (most OPEC Member countries). Moreover, customs office and ministry/statistics office may not agree with each other on trade statistics (e.g. Russia). On the other hand, as is the case in Egypt, exports may not be reported from foreign concessions. On top of this, there is the problem of barter trade which does not appear in most statistics.
It is mostly believed that the item stock changes is one of the most important factor driving the market and hence the price of oil. Stock changes are extremely uncertain, often unreliable and mostly subject to big revisions. My experience tells me that at least half of the countries in the world do not report the data for stock changes in their oil statistics. In many cases they are combined with statistical differences. In most of the countries military stocks are considered as secret information and they are not reported.
In particular there is also a misunderstanding as to what is being measured and covered. For instance, American Petroleum Institute and EIA regularly publish weekly data on the US stock position (with sometimes conflicting with each other). Note that the weekly bulletin of API is based on voluntary reporting with coverage of about 90-95% on most major series. Best data come from EIA petroleum supply monthly but it comes out 2 months after the fact. (see here for more on that). The EIA uses the same definitions as API. Their reports are released the day after API’s. The problem is sometimes is not with stocks but with logistics like frozen waterways and ports being closed due to bad weather.
The IEA keeps track of the stock positions in OECD region (including the stocks held on behalf of the governments under the emergency stockpiling requirements). All these sources exclude stocks held by players such as producer governments, stocks at sea, offshore stocks (especially in Caribbean, Netherlands and Singapore.) and also secondary stocks held by wholesalers and tertiary stocks held by consumers. Hence only a small part of total stocks are measured and even that data is considered to be dubious.
Bunkers is a tricky item and hence very much confused in oil balances. There are two types of bunkers – domestic and international. If fuel is delivered to a foreign entity (like embassies) they are considered as domestic bunkers (e.g., in Russia). Fuels delivered to sea going ships and to plains destined abroad are considered as international bunkers. In national publications, international bunkers are treated either correctly, or considered as exports (most Latin American countries), or domestic consumption (e.g., Iran, Thailand). Even if they are correctly identified as international bunkers, this time, confusion arises in terms of flag type. Many countries consider the fuels delivered to ships or to air planes as bunkers if the delivered entity has a foreign flag (in case of marine) or belongs to another country (in case of planes). For example, Brazil, defines Marine Bunkers as bunkers carrying strictly “foreign" flags, and therefore the bunkers with "national" flags are reported under domestic. According to the international standards, however, the entity is not important and international marine bunkers, for instance, are defined as fuels delivered to sea-going ships of all flags including warships. God help statisticians!
Demand is mostly calculated as apparent consumption in case end use data by sector are not available. If, indeed, detailed end use data by sector were available then it would be possible to get a much better picture of demand. For this reason, using apparent demand overestimates the demand. This is also one of the biggest problems in green house gases inventory calculations. How much do we know about the real demand then? Well, not much!
Oil Piracy and Smuggling are getting more and more problematic to track. Illegal oil bunkering and oil piracy from fields and pipelines in the Niger Delta is estimated to be 300 kb/d according to the Governor of Delta State. Shell argues that up to 100 kb/d is being stolen from its operations. In Port Harcourt gangs are reported to steal crude from pipelines.
In Russia, about 200 kb/d of oil (mostly crude) is stolen from pipelines. In recent years, the Basra oil terminal in Iraq and the southern port city of Umm Qasr have emerged as two of the world’s new piracy hotspots. Crude and products tankers are among the most affected vessel types. Indonesia and the Malacca Straits remain the most prone areas. The magnitude of oil products smuggled into Turkey is estimated at 70 kb/d. The production of the smuggled and stolen amounts is most probably reported in production statistics but its consumption definitely does not show up in demand and hence missing.
A common feature of the oil balances is processing gains. Refined petroleum products are less dense than the input. The volume of products from a refinery — not their weight — typically exceeds the volume of the oil input by 1% to 6%, depending upon the combination of processes employed, and this is termed processing gain. The IEA’s Oil Market Report, for instance, is blamed for having very low processing gain. I completely agree. World refinery input has been going up but processing gains are set somehow always around 1.9 Mb/d. This subject is widely covered by Littell (1999, 2000) and Auerback (2002). By the way, you can not have processing gains if input and output are measured in tons. The IEA should be more careful about that.
Missing link - Conversion Factors: Oil statistics in the world over are expressed by using four most common conventions - litre, cubic metre, barrel, and ton. Since world oil balances are usually expressed in barrels, the selection of appropriate conversion factor is essential. Interestingly, not enough attention is paid how different measurement units for oil are aggregated and which conversion factors are in use.
Although, the relationship between volume and mass measured by density or specific gravity, the number of barrels of crude oil per ton for the same country varies widely, depending on the source. The discrepancies in conversion factors in use, although seem small, are not negligible and in fact already play an important role in world oil balance. This can be demonstrated by using the Russian crude oil production. Russia releases its oil production in tons. Take an original Russian production figure in tons for any year. Then apply EIA and IEA’s conversion factors and see the difference. Do the same exercise for other countries that measure their oil in tons. Sum them up and compare the difference. You will be surprised!
The misuse of conversion factors is not limited to primary oil. Similar problems exist for refined oil products. Although barrel-ton conversion factors vary by country and by time for supply, for demand (which is calculated by using the oil products), the general tendency is to use generic conversion factors. However, there are many petroleum products whose conversion factors might have big impact in oil balance tables due especially to their share in total. Not surprisingly, these generic conversion factors again differ among the internationally recognised sources. (By the way, the world has switched from leaded gasoline to unleaded gasoline but conversion factor from ton to barrels remains unchanged.)
Assume that IEA’s world oil demand in ton is stone set for a given year. Would you believe for example that in converting from ton to barrels by using the OPEC (OAPEC and MEES) barrel ton conversion factor for gas-diesel oil instead of the IEA’s one, the world oil demand in barrels would have been reduced by more than 5%? You didn’t like it, no? Then replace the IEA’s conversion factors for heavy oil with OPEC (BP). You would increase world oil demand by about 5%. I did this exercise a few years ago. Very entertaining.
Missing Demand Item - US Military Oil Consumption Overseas
According to my estimates roughly 350 kb/d of oil is missing in world oil demand statistics. See more on this here.
Data Quality in Non-OECD Area: Although both IEA and EIA acknowledge the increasing importance of developing countries in their energy outlooks and oil market analyses, they are myopic when it comes to the quality of data in those countries. If, as stated by Lynch (2002), the industry is becoming more sensitive to the marginal barrel, then more care and attention should be paid to the data quality in constructing world oil balance and hence its interpretation. This is one of the reasons of the data discrepancies observed in the vast sources of information made available to the industry players.
World oil statistics, in scope and accuracy, are still far from perfect. They are of poor quality, imprecise, mostly contradictory, and subject to sometimes radical but mostly not noticed revisions. Therefore, they can easily lead to misguided conclusions regarding the state of market fundamentals. Without proper attention directed at statistic caveats, the ensuing interpretation of oil market data opens the door to unnecessary volatility, and can distort perception of market fundamentals.
Meanwhile, world oil production and demand centers are shifting from developed countries to developing countries where statistics are extremely poor.
Given restructed resources, the institutions monitoring the oil market fundamentals (especially the IEA energy statistics division headed by a proactive leader) do their very best in order to deliver the most correct, timely and coherent information to people involved in the market. However, there is still a lot of room for improvements.
Auerback, M. (2002), The Secular Case for Higher Oil Prices
Horsnell, P. (1998), The Strange Case of the Missing Barrels, OIES Monthly Comment, December 1998.
Karbuz, S. (2004), Conversion Factors and Oil Statistics, Energy Policy, 32, 41-45.
Laherrere, J. (2001), Estimates of Oil Reserves, paper presented at the EMF/IEA/IEW meeting, IIASA, Austria,.
Littell, G.S. (1999), World Crude Production: Bad Statistics Produce Poor Conclusions, World Oil, June 1999.
Littell, G.S. (2000), Bad Data Problems Worsens as Oil Prices Skyrocket, World Oil, November 2000.
Littell, G.S. (2001), Volatile Oil Prices Result from Bad Data and Misguided Attention to Inventories, World Oil, October 2001.
Lynch, M.C. (2002), Causes of Oil Price Volatility, paper presented at the 8th International Energy Forum, Osaka, September 2002.
Mabro, R. (2001a), Does Oil Price Volatility Matter? OIES Monthly Comment, June 2001.
Mabro, R. (2001b), Transparency in Oil Markets and Other Myths, OIES Monthly Comment, February 2001.
Simmons, M.R. (1998), An “Arithmetic Mystery”: Missing Oil, Simmons & Company International.
Simmons, M.R. (2002), A Reality Check on the World’s Oil Supply and Demand: Are the Current Perceptions of Weak Demand and Ample Excess Capacity Correct? Canadian Energy Research Institute, Alberta, January 2002.
Dr. Sohbet Karbuz is the former head of Non-OECD Energy Statistics Section of the International Energy Agency.