Some comments on the Cambridge Energy analysis of future oil supplies

Most pundits had one of two reactions to the recent analysis by Cambridge Energy Research Associates about the prospects for global oil supplies over the next five to fifteen years. Some analysts took the CERA report as confirmation that concerns about peaking world oil production have been misplaced. Others dismissed the CERA findings as completely without merit. I would urge anyone who had either of these two reactions to take a second look at some of the issues.

Daniel Yergin of Cambridge Energy Research Associates recently
reported the results of a field-by-field analysis of oil production capacity
which concluded

CERA estimates for increased production. Source: Energy Stock Blog
Country Predicted capacity increase

Saudi Arabia +3 mbd

Caspian Sea +2.5 mbd

Angola +1.35 mbd

Canada +1.32 mbd

Nigera +1.27 mbd

Brazil +1.16 mbd

Russia +1.15 mbd

Iran +1.00 mbd

Iraq +1.00 mbd

Mexico -0.20 mbd

U.K. -0.30 mbd

Norway -0.33 mbd

U.S. -0.47 mbd

There will be a large, unprecedented buildup of oil supply in the next few years. Between 2004 and 2010, capacity to produce oil (not actual production) could grow by 16 million barrels a day — from 85 million barrels per day to 101 million barrels a day — a 20 percent increase. Such growth over the next few years would relieve the current pressure on supply and demand.

Where will this growth come from? It is pretty evenly divided between non-OPEC and OPEC. The largest non-OPEC growth is projected for Canada, Kazakhstan, Brazil, Azerbaijan, Angola and Russia. In the OPEC countries, significant growth is expected to occur in Saudi Arabia, Nigeria, Algeria and Libya, among others. Our estimate for growth in Iraq is quite modest — only 1 million barrels a day — reflecting the high degree of uncertainty there. In the forecast, the United States remains almost level, with development in the deep-water areas of the Gulf of Mexico compensating for declines elsewhere.

While questions can be raised about specific countries, this forecast is not speculative. It is based on what is already unfolding. The oil industry is governed by a “law of long lead times.” Much of the new capacity that will become available between now and 2010 is under development. Many of the projects that embody this new capacity were approved in the 2001-03 period, based on price expectations much lower than current prices.

Further details about this study can be found at Energy Stock Blog and

Noise Filter

The Oil Drum noted that a less exhaustive field-by-field count by the Oil Depletion Analysis Centre came up with a slightly more pessimistic 12.5 million barrel/day increase in new supply by 2010. The Oil Drum suggested that the reason that ODAC saw 12.5 mbd as implying an extremely tight market whereas CERA saw 16 mbd as overabundance might be due to different assumptions about the rate at which oil production would drop from existing fields due to depletion. Exxon-Mobil figures that a 4-5% drop on average in each year’s production from existing fields might be reasonable to assume in the present environment, which would mean that more than 16 mbd in new sources would need to be found just to keep total production from falling. Arguing against the Oil Drum’s interpretation is the fact that the production declines over the next five years assumed by CERA in the countries detailed in the table above appear to be consistent with the declines actually observed over the last five years, which are reported by the Energy Information Administration to have been -0.52 mbd for the U.K., -0.34 mbd for Norway, and -0.33 mbd for the U.S. However, the question of depletion rates would seem to be a critical issue. I have not seen a detailed analysis of it elsewhere, and would appreciate hearing from any readers who may have further information about this.

A second concern that has been raised about the CERA report is that 3 mbd of that 16 mbd in new production is imputed to increased production from Saudi Arabia. Even if you completely discount legitimate questions as to whether the facts on the ground in Saudi Arabia are as Aramco is reporting, it is far from clear that it would be in Saudi Arabia’s financial interest to increase production if the global supply and demand situation otherwise remains as tight as it is presently. They repeatedly said they would increase production during the last year but so far have done little. Furthermore, as Yergin points out, regardless of the physical production possibilities, above-ground threats of a supply disruption owing to political factors are very real, and indeed, the market’s concern that such factors could produce a decrease rather than an increase in Saudi production over the near term could be one reason why oil prices have been making new highs. Of course there are pressing above-ground concerns about other key producers including Iraq, Nigeria, and Russia.

A third question that has been raised about the CERA optimism has to do with the increasingly important role assigned to less conventional supply sources such as the Canadian oil sands and production from very deep water, whose contribution CERA projects to rise from the current 16% of global production to 30-35% by 2020. However, there is a reason that these sources are only now beginning to be tapped– they are far more difficult to extract, requiring much more in the way of both capital and energy to convert into usable oil. Via Oil Drum and Green Car Congress, Bloomberg reported that Shell Canada has raised the estimated capital cost of its 100,000 barrel/day capacity expansion for Alberta oil-sands production from C$4 billion to C$7.3 billion.
Peak Oil Optimist
is hoping that alternatives to the intensive use of natural gas to produce oil from this resource will be developed, but the substantial energy requirements suggest that costs will continue to rise as energy prices go up and that the effective contribution to net energy supplies from this resource is significantly overstated by just calculating barrels of oil produced. Energy Outlook cautioned that development of the Canadian fields on this level may be inconsistent with that country’s agreement to the Kyoto Protocol limiting carbon dioxide emissions.


One thing that the CERA and ODAC reports have done is to shift the discussion of global oil depletion from an abstract theoretical calculation of how long “ultimately recoverable reserves” might last, a concept inherently difficult to evaluate and with a very poor predictive track record, to a more practical discussion of how much oil are we going to have available over the next five years. Interest in the latter is in turn prompted by the ultimate here-and-now question of why supply was not adequate to keep prices from shooting up so dramatically over the last two years. Part of the answer to the latter question is that the recent surge in demand has been quite impressive, and part of the answer is that global spending on oil exploration dropped $2 billion or 18% between 1997 and 2002. The explanation for this drop in exploration spending is in turn the collapse in oil prices in 1997, which I attribute to the dramatic reevaluation of expected global demand growth rates that followed the economic crises in southeast Asia. The subsequent experience has been that Asian demand in fact resumed its very strong growth. Given the long lead times of which Yergin speaks, that left us with today’s very tight supply situation.

That leads me to highlight another very important claim that Yergin is making: CERA says that there will not be a sharp peak in oil production, but rather a series of undulating plateaus. As the reserves that are easiest to develop and in the politically most secure locations are exhausted, the price will rise, serving both to reduce demand and create the incentives necessary to develop more costly and riskier supplies, allowing production to rebound after an initial decline. I would not be at all surprised to see lower world oil consumption in 2006 than in 2005 as a result of the gradual changes in energy use that are already under way in response to the current prices, in other words, a production peak in 2005. I also would not be at all surprised to see demand subsequently rebound as world incomes continue to grow, met with supply coming on line from the factors identified by CERA. Nor would a sudden calamitous drop in that production, precipitated by above-ground events in the Middle East, be a surprising thing to see between now and 2010.

Sounds to me a whole lot like an undulating plateau, very much caused by the exhaustion of the easiest to obtain and least risky petroleum reservoirs. In other words, peak oil may be here right now. But not in the form that those who have popularized the expression have typically envisioned.

Technorati Tags: ,


39 thoughts on “Some comments on the Cambridge Energy analysis of future oil supplies

  1. Avo

    Matt Simmons, raiser of the “legitimate questions” you cite, is not impressed with the CERA analysis:
    “Now, I’ve read carefully through Daniel Yergin’s detailed field-by-field bottom-up report, and basically, it is a really flawed piece of analysis in my opinion. But the fact that they obviously believe it’s correct– they’re doing talk shows– shows you the depth of limitation of people that really understand how serious this is. Cambridge Energy Research Associates also, in 2001, were unbelievably pooh-poohing the idea that the United States had now entered a major natural gas crisis. But by 2004 they got the religion. I expect by 2009 they’ll issue a magnificent tome saying, ‘gosh! it looks like the world is now past sustainable peak oil supply.’”

  2. Stuart Staniford

    It seems in general that the depletion rate is the absolutely critical issue, both in understanding how much it offsets the effects of the new projects, but also in understanding how bad the backside of peak oil will be. A few percent annual depletion will be easily handled for a long time by ramping up alternatives (oil sands, coal, nuclear, renewables), switching to more efficient technologies on the demand side, etc. However, depletion up near 10% would be almost impossible to handle via market substitution, due to the lifetime of existing equipment (capital stock turns over much less than 10% a year), thus would have to result in ongoing economic contraction. I don’t have a good way to estimate the precise crossover (the sustained depletion rate above which there must be sustained economic contraction till we are mainly reliant on things other than conventional oil for transport). My guess is somewhere in the 3-7% range.
    Good analyses of future likely depletion rates seem to be non-existent as far as I can tell so far. Older oil domains such as the US have historically depleted at a few percent annually. Peak oil thinkers such as Colin Campbell have historically assumed depletion going forward would be in that range. However, in the last few years, it’s become clear that UK oil depletion is well over 10% a year. Australia is similar. Mexico has stated that they are peaking now and expect to deplete at 10%-15% a year. Heading Out at The Oil Drum recently asserted that the existing well infrastructure in Saudia Arabia is depleting at 1m bpd (around 11% annually), which needs to be offset with new production (I’m not sure where he got this number) and Iran is similar. Iran and Saudia Arabia still have projects to do, where UK and Mexico have, at least for the time being, run out of adequate offsetting options (arguably, this may be due at least in part to lack of investment during times of low oil price).
    The issue seems to be that modern methods of oil extraction appear to keep production high for longer, but then it falls fast on the back side. Eg, horizontal multilateral wells at the top of the oil layer keep up much higher production than vertical wells, until the oil is pretty much gone, and then it starts to fall very fast. Similarly, regular seismic imaging of the oil in place allows rapid exploitation of pockets of left behind oil until there are none, then it’s over.
    As far as I’m able to tell, the use of this kind of technology is now widespread. So there is an argument that global depletion will be much faster than Hubbert, Campbell, Deffeyes et al have predicted. This is the basis for Matt Simmons saying things like “If we don’t address this, no scenario is too dark”.
    I would be very interested in anyone who has a compelling argument for why global depletion of conventional oil (the kind that flows by itself) will be slow once it gets rolling (of course it will be slow at the outset, and I agree the peak may well be bumpy – I think many peak oil folks expect a bumpy peak).

  3. TCO

    I agree that depletion and production from new fields seem to be confounded in some of the discussion. Stu, if understanding depletion is key, then it is simply a matter of understanding reserves. We know consumption. But that puts us back at the “Saudis might be lyiung about their reserves”. Although, why every “econ-hater” seems to think that the market does not consider this possibility given that it is in the press…

  4. Bubba

    Stuart has hit the nail on the head. My understanding is that the depletion from the North Sea is much greater than first forecast. In addition, the Mexicans have been very upfront about the depletion expectations from Cantarell field in the coming years, but I know from experience that predicting declines is a difficult thing. A couple percentage points either way, and you are way off in your prediction by 3 years out. As the tendency is to be optimistic, most forecasts underpredict field declines, especially in their early phases.
    The last thing I want to say is in regards to ultra-deepwater production. Most of what is being discovered now in the Gulf of Mexico, West Africa, and Brazil is not of the same caliber and quality of accumulation that was discovered in the 80′s and 90′s. The reservoirs tend to have many or all of the following characteristics a) low permeability b) are in very deep water (>2500 m) c) contain poor quality, viscous crude d) tend to be compartmentalized into small volume compartments, and e) are small in total volume. Even if the big oil companies were using $50/bbl price forecasts (which they aren’t) development of these reservoirs is technically uncertain – period.

  5. Bubba

    Another thing I would like to say is that, although the CERA forecast is getting a lot of airtime – there are other forecasts that have been done by equally well respected companies that are much more pessimistic. The one I can point to is from
    PFC Energy, a highly-respected Washington DC-based consulting firm. Their prediction, derived also from a bottoms-up forecast (unfortunately the study is proprietary and only for their customers) is that “It is difficult to envision a world where oil production exceeds 100 million barrels per day”.
    I personally am using CERA’s forecast as the upper bound forecast for predicting the future relative to oil supply.

  6. T.R. Elliott

    Hence my argument we should call it “plateau oil”, or perhaps “undulating plateau oil.”

  7. RayJ

    A couple of additional approaches taken to gain some insight into the supply demand in the near term are:
    1) A recent article in EnergyBulletin ( reports:
    “Chris Skrebowski, a board member of Odac, has analysed all planned oil field projects worldwide with reserves of more than 500 million barrels and concluded that, on current timetables, output from new fields will be insufficient to offset more major oil producers moving into net production decline.” In other words, if the project has not been announced and initiated by now, don’t expect any production before 2010.
    2) Herold has performed a company by compnay analysis that looks at when the major oil companies are likely to peak. Their analysis shows most companies peaking by 2010 based on their reserve and projects underway. This manily covers non-OPEC oil.
    Personally, I think the period between now and 2010 will probably be Yergin’s undulating plateau but with a downturn in supply thereafter.
    This is because the supply from unconventional sources (tar sands, deep water, coal-to-oil, etc.) is going to be less than predicted because of the low EROEI.

  8. Pinon

    The “Predicted Capacity Increase” numbers look great but without being able to following individual well-head production rates, the figures and the deductions may all be meaningless.

  9. Fred Press

    Everything being discussed is only conjecture, and dependent on far too many unknown variables to produce an accurate short-term model. The best we can do is to produce mid-term and long-term models, which suggest that oil prices will revert back to numbers we have seen somewhere in the period of the last five to ten years before once again resuming a long-term uptrend. One thing we do know is that current soaring oil prices are highly correlated with recent record low interest rates and soaring real estate valuations. When Greenspan completes his correction of these anomolies, oil prices will follow, and we will soon enough be discussing the short-term oil surplus.

  10. Max

    JDH: “However, the question of depletion rates would seem to be a critical issue. I have not seen a detailed analysis of it elsewhere, and would appreciate hearing from any readers who may have further information about this.”
    Depletion can be derived mathematically from known data and a few assumptions. Assume the production curve is roughly a bell curve or inverted exponential of the form y = -x^n where
    y = depletion
    x = time
    We can solve for n because we know that after the production peak, depletion = production = consumption = Q = 31 billion barrels per year.
    Assuming for simplicity that 2005 is the peak year, integrate:
    Q = integral(-x^n)dx
    Q = – x^(n+1) / (n+1)
    Evaluate from x = 0 to 365 (days in a year):
    The solution is n = -1 (surprising?). Therefore,
    y = -x^-1
    y(365) – Y(0) = 0.0027
    I.e, 0.0027 billion barrels depleted per day
    or 2.7 million barrels per day (Mb/d)
    Current production = ~83 Mb/d
    2.7 Mb/d corresponds to a ~3.2% depletion rate
    Seems reasonable but perhaps a bit low if Saudi production falls off a cliff like Simmons suggests.
    And remember that the depletion COMPOUNDS year over year. 3.2% compounded over 10 years gives 37% depletion from the peak year of production. Rather sobering.

  11. Stuart Staniford

    Your math is incoherent. You say that the “production curve” is “y = -x^n”, which is “a bell curve or inverted exponential”. Actually it’s a power law. Also, you define y as depletion, which is inconsistent with this being a “production curve”. Either way, y is going to be negative with that formula, which makes no sense. Finally, there’s no way to compute the depletion rate only from the current production. You need to model the underlying processes that are causing less oil to be available.
    TCO: understanding depletion is not just a matter of understanding reserves (though good information about reservers would certainly help). At least not if you mean “we just need to know the total reserve nunber”. The existing well infrastructure is producing oil at a certain rate because (in most cases) the reservoir is under higher pressure than what it takes to push fluids to the surface, and at least some oil is incident on the surfaces of the wells that are open to the rock (probably some water too). Production of the set of wells the world has today is going to be less next year because of some combination of lower pressure and higher water to oil ratio in what is seeping out of the rock into the well. That must be offset by new production (more wells into existing fields, or bringing on new fields. Production capacity is long lived, long lead time infrastructure which can only be introduced or changed slowly.
    The form of the depletion curve depend dramatically on the nature of the rocks, the geometry of the wells in the field, and the availability of offsetting options. Eg, if you visualize a layer of rock with a 50′ layer of oil in the pores with water below, impermeable rock above, and injecting more water into the water layer to maintain pressure, then a set of vertical wells you might expectt to deplete *very roughly* in an declining exponential fashion: the ratio of oil to water is proportional to the remaining height of the oil column. However, if you have horizontal wells at the top of the oil layer, you would expect something that looks more like a step function: there’s hardly any water until there’s no oil layer left, and then it’s all water.
    This is a gross oversimplification of course, since the rocks are inhomogeneous, rock layers aren’t horizontal, etc, etc. But very roughly, this is what has happened in the giant fields in Saudia Arabia (switching from vertical wells to horizontal wells because the vertical wells were producing too much water).

  12. Max

    “Your math is incoherent. You say that the “production curve” is “y = -x^n”, which is “a bell curve or inverted exponential”. Actually it’s a power law. Also, you define y as depletion, which is inconsistent with this being a “production curve”. Either way, y is going to be negative with that formula, which makes no sense. Finally, there’s no way to compute the depletion rate only from the current production. You need to model the underlying processes that are causing less oil to be available.”
    Stuart: I’d draw you a picture if it were possible in this forum. There’s nothing wrong with y being negative, since it’s defined as a drop from a peak…the sign doesn’t matter, so change it if you like. Yeah, you’re right –should have said “power” instead of “exponential,” but it doesn’t affect the argument. A bell curve can be roughly approximetaed by an upside down power function –at least near the peak where it looks like an inverted parabola, which is our focus for this argument. Of course you can compute depletion from current production — I just did. I don’t think you follow the argument or understand the assumptions.

  13. STS

    I think I get the picture you’re trying to draw, but the math is at best distractingly off. Even if you want the production curve to have roughly the shape of the right half of a bell curve you can’t use a function like y = const – x^n because it will hit zero in a hurry. Production is likely to tail off and can’t go negative anyway.
    After that, you proceed to integrate x^n as a polynomial, then decide n = -1, which integrates to log, not a power of x.
    But doing any of this without a model of the underlying production engineering is apt to be no better than a random guess.

  14. Ian

    I understand your argument (but it does need to be explained a little better). Isn’t it interesting that if you use the power law to describe the depletion curve (which given the applicability to similar natural processes seems reasonable) that you get n=1, or Zipf’s distribution! Of course your argument relies on the fact that we are at peak production … Even if we are not it provides a good lower limit estimate.

  15. TCO

    Ok, Stu: The shape of individual feild depletion curves tells you the pressure drop info? (Even here I would think you need more info, because some fields are probably pressure limited, some are limited for other reasons, some are purposefully not pumped at capacity (SA)?

  16. Max

    STS: “I think I get the picture you’re trying to draw, but the math is at best distractingly off. Even if you want the production curve to have roughly the shape of the right half of a bell curve you can’t use a function like y = const – x^n because it will hit zero in a hurry. Production is likely to tail off and can’t go negative anyway.”
    I agree, which is why I approximated depletion just 1 year off the peak. You can use the distribution function but you’ll need to make some guesses:
    STS: “After that, you proceed to integrate x^n as a polynomial, then decide n = -1, which integrates to log, not a power of x.”
    That was the “surprise” I mentioned in my original post. I’m being deliberately sloppy because the details of the function aren’t known except that it must approximate a falling off a peak. The point is, what is the falloff rate, i.e., n? Is n 1? I showed how to solve for n assuming one possible expression that uses n.
    I say my method is at least as valid as efforts to draw statistical inferences from flawed global reserve data. Debating about rock formations is kind of futile in this context.

  17. STS

    Max, your reasoning looks pretty circular to me. You mention only two data points, present production and some unexplained quantity from “after the peak”, then solve for the exponent of a power law to fit these two data points, then use this to extrapolate 1 year past peak. Seems like the choice of 1 number from the future was more or less begging the question.
    The presentation cited above:
    notes that North America has been post “plateau” for 20 years. The production data for NA should provide a sensible basis for estimating a functional form for the depletion curve and its parameters. Aggregating over a large region would probably help to smooth over well-specific variation in the depletion patterns allowing some abstraction from the geological details.
    I wonder if anyone has done that type of analysis.

  18. STS

    Estimating the duration of the plateau itself would be of considerable interest as well, since that varies widely region to region.

  19. Stuart Staniford

    My point is that future depletion depends in a dramatic way on how the field is being managed. A reliable study that projects future global oil supply, it seems to me, would need to to understand, on a field-by-field basis, what the likely depletion rate is, which would have to take into account the geology and petroleum engineering issues (like are the producing wells vertical or horizontal, how thick is the remaining oil layer, etc, etc).
    Chris Srebowski’s report applies a global depletion number. The CERA press releases make no mention at all of how they handle depletion (so it seems unlikely it was much of a focus). So my feeling is these projections are probably pretty worthless. It’s like trying to do a good job on projecting a company’s financial performance by carefully mapping out future revenue sources, while only using a rough single number guess for expenses.
    The cause for concern is that the massive depletion rates in the North Sea etc as a result of applying modern technology seem to have been a surprise to the industry, and even to the well known peak oil experts, who all have been projecting lower depletion rates. Eg Colin Campbell in 2002 [1] projected that the then 6% depletion rate for UK oil would continue out indefinitely, in fact it’s gone up dramatically since then.
    Peak Oil sceptics such as Michael Lynch have noted that Campbell has repeatedly miscalled the peak too early [2]. Lynch’s interpretation is essentially that the URR (ultimately recoverable resource) must really be much bigger than Campbell thinks (and indeed Campbell has been steadily increasing his URR estimates over the years).
    The worry is that Campbell was more-or-less right about the URR to begin with, but wrong that global oil production would have a roughly symmetric Hubbert peak [3] – instead, the increasing application of the modern oilfield technology is going to give rise to a ripsaw tooth peak (much steeper on the back side than the front side). As far as I can see at the moment, the data do not rule out this hypothesis, and that makes me extremely uncomfortable.
    [3] Hubbert basically assumed that cumulative global production would follow a logistic curve (an equation originally derived by Verhulst in the 19th century for population modeling, but since used for all kinds of things). If C is cumulative production as a fraction of the URR, and t is time, this has a differential equation dC/dt = kC(1-C). The C factor captures the idea that early on the rate of production is proportional to the total installed capacity of the industry. The (1-C) captures the idea that as you run out, the amount you can produce is proportional to how much is left (literally true for an idealized single vertical well in a depleting oil layer with a strong reservoir drive). This model gives rise to a sigmoid cumulative production, and a symmetric peak. Horizontal wells, 4D seismic etc, do not seem to me well modelled by that 1-C term.

  20. JDH

    Great comments, all. Can any defenders of CERA give us some more concrete details about what CERA is assuming for depletion rates and how these depletion rates were arrived at?

  21. bubba

    I don’t mean to insult anybody here, but reading many of the above posts reinforces my desire to be “vaguely right rather than precisely wrong”. There is really a danger of losing the forest for the trees here.
    Every year at my company we try to build a long term forecast of production starting with forecasting the decline of every producing well in every producing field. We layer on the wells which are not yet on production, then the ones that are not yet drilled, then the fields that are prospects that have not yet been explored, and lastly we layer on the fields that haven’t even been dreamt up yet. This involves the work of hundreds of engineers which is all summed together in the equivalent of a large spread sheet. Usually by the end of the first quarter of the following year it is grossly (1-2%)out of alignment with reality and heading badly in the wrong direction.
    Sometimes the problem is arithmetic errors. Sometimes the problem is pressure to be optimistic. Sometimes the problem is not actually discernable without deep analysis. In the end, a forecast fitting a curve to the previous years gross production would almost always have been more accurate.
    I have about as much faith in these bottoms up forecasts as I do in my own companies long term production forecasts. Without an obvious “step changing event” trends will continue.

  22. JDH

    I’m not sure I’m following you all the way to the conclusion here, Bubba. Doesn’t extrapolating last year’s trend mean that you’d predict that global production will keep going up forever because that’s what it’s always done so far?

  23. Stuart Staniford

    Fascinating. Are these internal forecasts always over-optimistic, or equally likely to err in either direction? Do the forecasts of depletion in the producing fields tend to be solid, or is that unreliable also? (I can see projecting the production of unexplored fields is bound to be an error-prone process).

  24. Edward Hugh

    I’m enjoying enormously reading all this. This is not an area where I have any background, so I am learning a lot. I just can’t help seeing the similarities between this debate and the demographic/aging one (which, of course, CSIS is also into). Stuart points out the common modelling origins in Verhulst etc, but also I can’t but be struck by the seeming temptation to try and buy the better case scenarios, and the short-term lock-ins as the time scale needed to change things (assuming some change is possible) are significant. There seems to be a general system theoretic issue here of just how adapted our political and administrative systems are to taking decisions over the pertinent time horizons, and especially in situations where there is a high level of inbuilt uncertainty. Incidentally Wolfgang Lutz and Sergei Sherbov have been developing a probabalistic methodology for trying to assess the “peak population” issue. A summary of this can be found here:

  25. SR

    You are not alone in your pessimism about “bottoms up forecasts”.
    You may find the article “Forecasting oil supply: theory and practice” (Lynch) interesting:
    It looks at the track record of some of forecasting models and argues that historically forecasting models have underpredicted oil supply and overpredicted oil prices: “there was a consistent tendency to predict a peak within a few years, and a decline. As this peak was surpassed, the correction was not to remove the peak, but to raise it and move it out, always a few years into the future. This resembles the price forecasting record, where throughout the 1980s, prices were predicted to rise from whatever the current level was, even though they were continually falling.”
    Here are some of the conclusions from this study:
    “Most oil supply forecasting has been done very badly in the past, with many models severely underspecified. Although geology is an important determinant of discovery rate, the tendency of some modelers to interpret all supply behavior as being geologically determined is impossible to justify. And even where the models appear to be correctly specified, the results still prove to be too pessimistic, suggesting there is some remaining bias at work.
    Indeed, one of the best pieces of evidence of pessimistic bias is simply the overwhelming number of forecasts that have been produced-many from models whose design appears accurate-but which proved not only wrong, but embarrassingly too low. That many of those modelers have generated new forecasts which are nearly identical to their old ones without explaining the cause of the previous errors should make even the most casual observer skeptical.”

  26. T.R. Elliott

    Bubba: I agree that forecasts are almost invariably wrong (though 2% isn’t too bad if you ask me). Though difficulty forecasting is not an argument to not forecast. It’s just an argument to be cautious with the results, to improve them, and hedge the forecasts with other sources of information or guidance, e.g. trends.

  27. bubba

    The trends I see are a continually increasing rate in the demand for oil, and a ever decreasing rate in both oil discoveries and deliverabilty of oil worldwide. When they cross depends on the details, which at the moment are uncertain, but less uncertain than last year.

  28. Bill Ellis

    It is my experience that in evaluating projections we must first examine the environment from which that study is spawned. When we examine the oil companies data they are subject to two pressures – the optimistic information that should prop up stock prices, and the realistic pressure to generate targets that are attainable because the markets punish over-optimism.
    However, there is a much greater economic community that is vested in pessimistic forecasts of supply. Through assuming long term increasing prices we can justify increased drilling activity to the economic benefit of the drilling operations. Acqusitions such as the purchase of Unocal by Chevron can be justified to the delight of Wall Street. Obviously the oil producing states are benefiting from the pessimism of supply forecasts and the optimism of consumption forecasts. Can we expect that these entities are likely to provide data to their economic detriment? Short term tight supplies are to the current benefit of every data provider.

  29. The Oil Drum

    More thoughts on the CERA report

    Econbrowser has kindly printed, and directed us to a source for the CERA predictions on supply growth. A quote from that source may be an appropriate start…

  30. Robert Schwartz

    If the CW (Peak Oil in this case) is always wrong, the Yergin is right. OTOH About 25 years ago, I spent good money on a book about energy policy by a bunch of wonks from Harvard Business School: Robert Stobaugh and Daniel Yergin, Eds., Energy Future: Report of the Energy Project at the Harvard Business School (New York: Random House, 1979); which confidently asserted that oil was headed straight for $100/bbl (about $240 in 2005 money). Of course, the bubble burst and oil went down for most of the next 23 years.
    Gentelmen, it is time to place your wagers on the table.

  31. Joseph Somsel

    Here’s a great piece on OPEC oil and gas production plans by a guy who sounds like he knows what he is talking about:
    In summary, there is no economic incentive to boost production, there is no excess capital, and the meeting the domestic needs for natural gas are more important to their countries. He throws very cold water on the EIA and IEA market projections and even touches on JDH’s point in the CERA thread about demand destruction from high prices.
    What he doesn’t address is whether there are oil prospects available, even if they wanted to increase oil production.

  32. TCO

    As a cartel, or the key parts of the cartel, or the swing producer of the cartel, they have no incentive to find more. But as individual players they do (to cheat). But that gets us back into the whole, is the cartel functioning or not question. Which I don’t feel is settled. I think there are still some game theoretic aspects to be looked at here. and that if we can do the right things to crack the cartel and bring abot free competition, that would be good for us consumers. (I don’t strongly beleive this…just think it’s possible enough to be checked out…)

  33. saif

    If canterell in mexico declines as expected by their chief that would mean a decline of about 300,000 barrels per year at least. Which would mean a negative 1.5 million for mexico alone. Anyone know how CERA ends up with just -0.2 million? Maybe they took the last 5 year depletion rates and that is really like Matthew simmons was saying. I am 65 and have not spent a single night in the hospital so it is likely I will not spend single night in the hospital for the next 65 years

Comments are closed.