There is always an easy solution to every human problem — neat, plausible, and wrong.
— H.L. Mencken
Forecasts come and forecasts go1. Every year the Energy Information Administration (EIA) serves up the Annual Energy Outlook (AEO), a rosy estimate of the world oil supply and price which gives rise to a fresh wave of optimism that “business as usual” will continue. How reliable is EIA forecasting? Is it based on sound methods? Only an investigation into how it is done allows us to answer these questions. Some digging reveals that there is little reason to trust EIA prognostications.
Note — Click on the graphic (below, left) to open a new window containing Graphs 1 through 5 used in this article. Keep the window open for reference. [EB: You may have to hold down the Shift or Control keys when you click on the graphic to get the graphic to open in a new window.]
The EIA uses the National Energy Modeling System (NEMS) to make its AEO forecasts (Graph 1). The NEMS model estimates all American energy needs, but only the International Energy module (IEM), and secondarily, the Petroleum Market module (PMM), are of interest here. These components are used to model the future world and U.S. oil supply and demand curves, respectively. Understanding how NEMS works is best approached by examining the EIA’s model assumptions and the IEM model documentation. All quotations below are from these documents. NEMS is a market clearing equilibrium model. (A more technical description can be found here.) The assumptions document explains the EIA’s methodology —
The International Energy Module (IEM) performs two tasks in all NEMS runs. First, the module reads exogenously derived supply curves, initial price paths and international regional supply and demand levels into NEMS. These quantities are not modeled directly in NEMS because NEMS is not an international model…
The second task of the IEM is to interact with the PMM module during NEMS runs to determine changes in the world oil price and the supply prices of crude oils and petroleum products for import to the United States in response to changes in U.S. import requirements. A market clearing method is used to determine the price at which worldwide demand for oil is equal to the worldwide supply.
The EIA’s market clearing algorithm works iteratively to produce a clearing price which balances supply and demand (Graph 5). The procedure (Graph 3) shows how “world oil prices are computed as a function of OPEC production decisions, availability of non-OPEC supplies, and worldwide economic growth.” Steps #1 through #7 explain how the algorithm works.
- The World Oil Market (WOM) component of the IEM “forecasts international crude oil market conditions, including demand, price and supply availability, and the effects of the U.S. petroleum market on the world.” The WOM determines non-OPEC supply and demand (excluding the U.S.) using only “the current oil price, Gross Domestic Product (GDP) growth rates, and last year’s supplies and demands”, both as set initially, and as set by the model for each forecast year based on the previous year’s results.
- The model assumes price-taking behavior in a perfect competition model for all countries outside of the OPEC cartel.
- U.S. crude oil supply and demand is set within the Petroleum Marketing module separately from the WOM. See the documentation for details.
- Given both non-OPEC and U.S. supply and demand values, the model adds the “Call on OPEC.” Supply from OPEC is exogenous — determined outside of — the model in all scenarios. It is assumed that OPEC “produces marginal supply at prices that inhibit any significant market penetration of new technologies (i.e. substitutes for oil). In the AEO 2007, OPEC does not include Angola.
- Given a trial price, the specified OPEC, non-OPEC and U.S. supply and demand values, the model asks this question: “does demand for OPEC output equal exogenous specified supply? (i.e world excess demand = 0).” If “Yes”, then STOP. At this point, world oil market is in balance at the indicated price.
- If the answer to the question posed in #5 is “No”, then make a pricing decision. If world excess demand > 0, raise the trial price to “dampen demand and stimulate non-OPEC oil production.” If world excess demand < 0, lower the trial price to “stimulate oil demand and reduce non-OPEC production.”
- Run the procedure again at the new trial price, which produces new non-OPEC and U.S. suppy and demand values. Continue until the loop halts.
What does this algorithm really mean? OPEC production is specified as growing at a fixed rate over time in the reference price case (Graph 2), which “assumes that OPEC producers will continue to demonstrate a disciplined production approach.” There are two others price cases (Graph 5). The low price case assumes a more open, competitive market in which the OPEC cartel plays a less powerful role in regulating world oil supply. The high price case includes scenarios where OPEC uses a “more cohesive and market-assertive” approach to restrict supply. All cases are expressed in 2005 dollars. Special provisions are also made about OPEC member Iraq. The available non-OPEC supply (Graph 4) is mainly a function of price, as step #6 above makes clear. From the model assumptions document —
Iraq oil production is assumed to not return to pre-conflict volumes until 2009. By 2030, Iraq is expected to increase production capacity to more than 5.5 million barrels per day with likely investment help from foreign sources. Non-OPEC oil production is expected to increase 1.3 percent per year over the forecast period, as advances in both exploration and extraction technologies result in an upward trend.
In economic terms, the peak oil problem can be restated to say that the oil supply is becoming more and more price inelastic, which means that as prices go up, the quantity supplied by the market does not increase much over time, even when the time lag between price increases and available new supply is considered. After the peak or high plateau of production is reached, the world oil supply can not be increased beyond that level at any price. The EIA makes just the opposite underlying assumption — non-OPEC supply is highly sensitive to price changes in all cases. The EIA provides the formula used to compute the non-OPEC crude oil supply (p. 27-29 in the model documentation, graphic left for convenience).
EIA’s function determines both the conventional (Sc) and non-conventional (Su) supplies for its reference case (R) based on the price (P), demand (D), lag factors (d, g) and unspecified supply price elasticities for conventional (e) and unconventional (h) oil. It is not known how the values of the elasticity parameters are determined, but these values necessarily reflect the EIA’s view that higher oil prices can always stimulate future non-OPEC oil production. This assumption, along with the OPEC scenarios, lies at the heart of the EIA’s forecasts about future supply and prices.
EIA head Guy Caruso offered some cautious disclaimers when he presented the AEO 2007 to the Senate Committee on Energy and Natural Resources on March 1, 2007.
The AEO2007 is not meant to be an exact prediction of the future but represents a likely energy future, given technological and demographic trends, current laws and regulations, and consumer behavior as derived from known data. EIA recognizes that projections of energy markets are highly uncertain and subject to many random events that cannot be foreseen such as weather, political disruptions, and technological breakthroughs. In addition to these phenomena, long-term trends in technology development, demographics, economic growth, and energy resources may evolve along a different path than expected in the projections….
Geopolitical trends, the adequacy of investment and the availability of crude oil resources and the degree of access to them, are all inherently uncertain. To evaluate the implications of uncertainty about world crude oil prices, the AEO2007 includes two other price cases, a high price case and a low price case, based on alternative paths of investment in production capacity in key resource rich regions, access restrictions, and an assessment of the Organization of Petroleum Exporting Countries’ (OPEC) ability to influence prices during periods of volatility… The cases are designed to address the uncertainty about the market behavior of OPEC. Although the price cases reflect alternative long-term trends, they are not designed to short-term, year-to-year volatility in world oil markets, nor are they intended to span the full range of possible outcomes.
The problems are far worse than Caruso knows, or is willing to admit. Economics and geopolitics are not the whole story. Geologically determined constraints on future OPEC and non-OPEC oil supply expose the flaws in the EIA’s assumptions. Oil production may be viewed as a function of several interacting factors, including some — levels of investment and available technology for producing new oil or increasing recovery rates — in which the oil price is an important underlying driver of future supply. Others factors — reservoir geology, production declines in aging fields, volumetric reserves estimates and geopolitical conflicts — are not very sensitive to price increases, or are literally (and figuratively) set in stone.
No oil price will halt the ongoing production declines in the North Sea. Resistance movements in the Niger Delta will not quit disrupting West African oil production because the oil price — even the bunkered oil price — is too low to make it worth their while. Those waiting in Iraq for the Americans to leave so they can engage in civil war don’t care about the EIA’s assumptions about the divided state’s oil production in 2009, or 2030 for that matter. Much higher oil prices are unlikely to make significant improvements in production rates from Marathon’s North Dakota Bakken Shale play, no matter how large the proved reserves turn out to be. Saudi Arabia’s Ghawar is not getting any younger, as each passing year adds to the depletion of reserves there. Kuwait’s Burgan field, the world’s second largest, is exhausted. Such examples can be enumerated ad nauseum.
The EIA’s market clearing model describes an ideal world in which macroeconomics and OPEC’s policies determine the future oil supply. Alas, the world we live in is not so simple. There is a joke geologist Ken Duffeyes likes to tell which applies to the EIA’s 2007 forecast. The economists, Duffeyes says, all think that if you show up at the cashier’s cage with enough currency, God will put more oil in ground. It’s time to tell the EIA that NEMS does not model the way the world works. Despite what Guy Caruso says, it is unlikely that the AEO 2007 describes the future oil supply and price.
1. This is part 1 of an examination of the EIA’s Annual Energy Outlook. Part 2 will look at past EIA predictions for U.S. and world oil production.
Contact the author at dave DOT aspo AT gmail DOT com
ASPO-USA is a nonpartisan, proactive effort to encourage prudent energy management, constructive community transformation, and cooperative initiatives during an era of depleting petroleum resources.