IEA Monthly Report for December

Ok, you happy plateau watchers. The IEA is out with the latest word (thanks Halfin). Quotes:

Global demand increased by some 1.3% in 2005 and should grow by 2.2% in 2006 as demand rebounds in the US and China.

Global oil supply reached 85.0 mb/d in December, up by 0.6 mb/d from November. Non-OPEC supply for 2005 is revised down by 90 kb/d on weak OECD output to 50.1 mb/d, unchanged from 2004. Non-OPEC growth accelerates to 1.3 mb/d in 2006.

So non-OPEC supply is flat over 2004, but never fear, they're sure it will go up next year.

The graph at right includes the latest information. The key point is that although the "initial claim" line stays at the November value of 85mbpd, the "revised" line for November is down to 84.4mbpd, and is not yet back to it's May peak. So Freddy Hutter will have to hold the schadenfreude at least for a little while longer.

Average monthly oil production from various estimates. Click to enlarge. Believed to be all liquids. Graph is not zero-scaled. Source: IEA, and EIA. The IEA raw line is what they initially state each month. The IEA corrected line is calculated from the month-on-month production change quoted the following month.
Update [2006-1-18 2:26:39 by Stuart Staniford]: Here's the percentage change in each month relative to the same month in the previous year.

Percentage change from same month in prior year. Average monthly oil production from various estimates. Click to enlarge. Believed to be all liquids. Source: IEA, and EIA. The IEA corrected line is calculated from the month-on-month production change quoted the following month. The smooth curves are quadratic fits to each data set.

Update [2006-1-19 4:0:57 by Stuart Staniford]:

Here's the contextual plot requested down in comments. This is the history of annual global oil production from 1930 to 2004. The little yellow box shows the small piece of recent history we are plotting in the top picture that introduced the piece, and is also the timescale for the year-on-year growth plot above.

Average annual oil production from various estimates. Click to enlarge. Believed to be all liquids. EIA line includes refinery gains, others do not. Sources: ASPO, BP, and EIA.

Had to look that one up.
It's the same feeling a lot of peak-oilers get when oil goes up $2 in a day :-)
Schadenfreude is a very german word...
... and means to be happy/glad on behalf of your fellow beeing unhappy/having bad luck.
"Schade"=harm "Freude"=happiness
I would feel a lot more Schadenfreude if my oil and gold mining options hadn't tanked today too.  What's up with that?

So, I left with nothing but Schade, Schade, and more Schade.  

I am most confused by the fact that they say OPEC supply declined by 280 thousand bpd in December yet global supply increased by 600 thousand bdp. They should check their calculators for shorts.

"OPEC crude supply declined 280 kb/d in December to 29.3 mb/d. Iraqi supply fell to 1.55 mb/d from November's 1.7 mb/d. Nigerian production was also disrupted in December and January."

And yet production increases substantially Nov to Dec?
Can't wait for the revised figures next month.

Well that probably have to do with the "Others" production.

Sometimes these IEA spreadsheets look like some sort of horror movie catalog.

"The Others' production" coming to theatre near you.

Do you have IEA spreadsheets? I need them. I'm serious. I love horror movies too, by the way.
I'm sorry to disapoint you, I don't. Searching here you only get graphs.

It was just a form of speech, I was referring to the Oil Market Report, where others is recurrent.

"The others" is also referred by Deffeyes as a way IEA has been using to hide the real numbers. That's why I asked Stuart for the O&GJ data.

The full market report is available about two weeks after the summary for nonsubscribers. It has much more data, but probably not the level of detail you are hoping for. Right now we can only get the full report pub in Dec.
You're "player hating" us folks with small monitors.
I got a big one coming any day, but for now I'm left out in the cold.:-(
Sorry - I checked it on my laptop and it was ok. How small do I need to worry about?
my screen is set at 1024x768 and I only got left 3/4's.

BTW, where is this data coming from exactly on the IEA site? Is it their Oil Market Report? I couldn't find it. Is it free or do you have to subscribe?

Does it break out individual countries? If so, what do they give for US, Saudi, Angola, Kazakstan for December?

Why are their numbers 1.5mbpd higher than the EIA's?

It's their oil market report. I had to trawl through and extract it out of all the past issues. As to the lack of agreement - welcome to the problem. View it as a crude estimate of the uncertainty in the production stats (though the corrected IEA series and the EIA series are very rarely that far apart).
I've just finished a fairly extensive compilation of monthly world supply comprised of the 75 countries that make up 99.8% of production.

I used EIA's and BP's number alomg with my own calculations to determine monthly total liquids from raw (crude+lease condensate numbers).

I put down the US for about 7.5 mbp for Dec up a million from Nov. I've got Iraq at 1.3mbp, Angola at 1.5mbpd for the hell of it.

I get a total of 83.090 with a probable error rate of +/- .300.

I don't even know half the time if EIA's data matches itself internally. Take a look at their world oil balance numbers, the data is completely different for the exact same time periods from what they come up with when they calculate supply and demand separately. If you look at their demand tables - the numbers are higher than the corresponding supply.

How do you find US up 1 million from Nov to Dec? MMS numbers suggest something like 200,000 - 300,000 b/d further hurricane recovery, and EIA would suggest a total of about 500,000 b/d more domestic oil production in Dec than Nov. Just curious.
The numbers straight from EIA for July,Aug are 8.128 and 8.173 - count that as pre-hurricane. Sep and Oct are 6.532 and 6.584.

At this point, to make things easy, I'm not using a particular number for November since we are already into Jan and I'd like to wait for a revised, published number.

I arbitrarily made December 7.500 as a mark on the road to hurricane recovery which will eventually take us back to the 8.1 level. I realize this is probably premature and I will be using your input to revise my table. Thanks.

You can obviously see the fun involved in trying to come up with accurate, up-to-date info.

Thanks for the reply. The difficulty in getting currnet numbers is why I asked, because I know it's a struggle. I appreciate your efforts.  
Ok - if you've can't resize it small enough, that's a problem with the minimum width (which I think Super G has been playing with. It's not something I directly control in writing the piece.
I'm on Firefox 1.5 on Linux with 1024x768 and there's no problem.

What browser and OS are you using?

XPsp2 and IE. I just tried Firefox, works fine. For some reason IE makes the image large, like 5 inches high, so it misses right-hand edge.
Can anyone with a big screen and IE on XP describe the behavior of the top image as you vary the width of the window from very small to very large?

I just tried IE on my Mac, and there is no resize restriction on either the permalink or the front page. The layout doesn't look great with very narrow window widths, but there's no problem getting things to fit across the width of the window.

I generally use Firefox.  In IE with my 19" monitor (1280 by 1024) I tried reducing the size of the window.

First the column of text to the left of the chart squeeezed until it got to a certain, fairly narrow width, and then the column on the right (about us, personnel, etc.) started to move over the chart from right to left thus effectively "cropping" it progressively as the size of the window was reduced.

Yikes.  It looks really, really bad in IE.  (I usually use Firefox, but tested it in IE because you asked.)

The text is shrunk to a skinny, unreadable little column, and the image is also cropped as the screen size gets smaller.

Guess I should have given my system specs.  :)

Windows XP, 21" LCD monitor, set to 1280 x 960.  In IE 6, TOD looks okay at full size, but quickly becomes an unreadable mess as the screen shrinks.  

Firefox handles it a lot better, shrinking the image rather than cropping it.  Though it still does ugly things to the text.

Ok - I checked it on an XP/IE setup and fooled around a little bit. The problem seems to be that IE ignores the width qualifier in the tag when the image is inside a table. I couldn't find a fix. So I won't use that table layout in the future (too bad - it was a nice way of getting a portrait image and some text into a story lead without taking up too much real estate on the front page - and it looked great in Safari).
No biggie

Also, what program do you use for your graphs?
I have been looking at some programs to make my own graphs, but yours look the best so far.
Is the program spendy?

Most of my graphs, and both of the ones above, are just made with Excel (which is very widely available but is somewhat expensive if you don't already have it).  The look of my graphs has steadily improved over the months as I've learnt more tricks with it.  Recently, I have incorporated a few things from Mathematica (which is very spendy unless you're a student).
I also usually use Firefox, but did just try TOD in IE. I get the same result: the table overlaps with the right column. This happens at full window size and only gets worse as I reduce it. When the window is half the screen size, the table can not be seen. I am using a laptop with XP and a 15" monitor.
Stuart - Why not fit a curve each time you post this data?
The plateau is still here. Let's see what's next.

Stuart (or anyone else) have you got the 2005 O&GJ data? Wasn't that soppused to be publish late December?

I don't at this point.
OPEC's December production slips

"Saudi Arabia cut production by 150,000 bpd to about 9.4 million bpd as buyers in the U.S. requested less crude ahead of shutdowns for spring refinery maintenance, according to the survey of consultants, shippers, industry and OPEC sources."

"Production from the 10 OPEC members not bound by quotas, excluding Iraq, fell 50,000 bpd to 28.23 million bpd as increased output from the United Arab Emirates partially compensated for the fall in Saudi output"

I wonder why, as the article states, the UAE would need to 'compensate' for lower Saudi production, if such a reduction was 'requested' due to refinery maintenance.

This will probably start a firestorm, but I believe Saudi production has peaked. The problem is, the powers that be do not want the situation to get out of hand, so any information reguarding Saudi production is fudged / distorted / suppressed. The disconcerting thing is that the post peak production fall off for Saudi production will likely be quite steep, as we are now finding with the North Sea. For those who are more skeptical about information suppression; What do you think would happen if CNN reported that Saudi production IS declining?

I think saudi have the capability to increase production a bit, but probably only of heavy, sour crudes which the world is currently tight on refining capacity for. I dug around for more detailed info on Saudi earlier today and found a surprisingly informative presentation from February 2004, see my post here:

They talk of increasing production to 12 mbpd by 2016, which seems realistic but too little, too late, though it's quite likely plans have changed in the 2 years since then. One very interesting figure I got from the presentation: they say the maximum ever field decline rate that they have had is 4.1% per year, the average is 2%. We'll see what happens when Ghawar starts to seriously decline, I'd bet its decline rate will top 4%.

But even if Ghawar is / does decline at 5% or more they can probably manage to hide that fact from the world for a couple of years. Until there is a real global supply shortfall and Saudi has to increase exports - or be found out - we are unlikely to know the real situation there.

Does anyone know how much incremental energy (in barrels or BTUs) it takes to refine heavy or sour oil into gasoline vs. sweet crude into gasoline? Ive seen a range of this stuff but  know that experts lurk.
How many points do I get if I tell you?

It'll depend on the type of heavy.  Give me an (the) API and sulfur content and I'll think about it.

yeah, it's weird. With prices continuing to rise, you would figure Saudi would be ramping to 10.5 mbpd this year. If you look at the numbers and you want to get another 1mbpd this year, Saudi is the only place you are going to find it.
Most applicable quotation for working with Saudi Arabian data...

"It is an old maxim of mine that when you have excluded the impossible, whatever remains, however improbable, must be the truth."

 - The Adventures of Sherlock Holmes, "The Beryl Coronet".

I wonder if this is enough time to clearly state oil production has plateau'd??

In my opinion, it's still too early.  There's plenty of evidence about extreme tightness in the availability of drilling rigs, tankers, refining capacity, etc. plus reduced production from Iraq and the Gulf.  So it's quite possible that we're seeing a plateau in production capability, and not the underlying oil peak.

My guess is that thanks to all these factors we'll be on an undulating plateau (as Jean Laherrère has predicted) for several years, right up to the time geology takes over and the true oil peak comes into play.

Still too early to say for sure, but that graph certainly is suggestive.  

If Hirsch is right, and the peak is sudden, sharp, and completely will be like all those countries that peaked.  We'll think it's just temporary technical difficulties, next year will be better.  Until five years later, we look back and realize yes, that was the peak.

Oil production has dropped in the past (more that three years ago).  While the graph is consistent with a peak, it certainly does not prove that we're at peak.
I am shocked that the IEA would make revisions. They are usually so reliable.
This graph looks extremely scary, but looks are deceiving.

We're used to seeing oil production, and we see a curve that goes up and falls back. With a nice tight fit. And then it crosses zero--yesterday! Eeek!

But this graph is year-on-year growth. That was quite negative in Jan. 2002. But we know that year-on-year growth has been mostly positive. No one was talking about a peak in Jan. '02. So it's clear that crossing zero does not mean peak.

Also, the fact that the graph was positive for most of the pre-'02 period means that the nice polynomial fit couldn't possibly continue if the graph were extended back to '00 or '98.

It'd be worth seeing the graph for the past ten or twenty years. We'd have a lot better idea of what zero-crossing meant and how common it was. I know Stuart had to go through a lot of back issues to get this much data, but the graph as it stands is misleading, since it shows a very close fit that we know can't continue even a few months to the left.


Well, crossing zero means a local peak. There's no way to know whether or not it's a global peak just by looking at that graph. Still, it seems illuminating to me because it removes any possible seasonal issues. The quadratics are just for the sake of aiding the eye in removing the noise. You are correct that they would not fit well very far to the left. They will only fit well very far to the right if oil production starts to decline pretty soon. I wouldn't rule that out, but it also seems possible to me that we'll eke out a small rise for a bit.
Yeah, I was kind of alarmed to see you fitting what seemed to be a parabola to this chart. Your recent work on Gaussians means that it should be a straight line. I'm sure if you zoomed out some more (i.e. included more years) you'd get closer to a straight line.

As Chris points out your parabolic fit can only be an illusion, it will never work when extended a bit farther to the left. This is a variant on the complaint I made previously, that in selecting this time window for the "plateau" graph you are giving a misleading impression. The fact that the % change graph fits a parabola rather than a straight line is more evidence of how anomalous this short time period is.

I'm certainly not seeking to mislead. What I am seeking to do is focus on a particular phenomenon - the inability of supply to quite keep up with demand in the last couple of years, leading to the present very high prices. This is a new phenomenon (at least in the last couple of decades). My time period is selected exactly to look at that issue - from the trough at the end of the demand-side, tech-crash problem to now. Thus we can debate whether it's a temporary tightness, as Freddy assures us, or a near-term peak, or something else again.

I don't see any deep reason for the parabolic fit to the y-on-y growth curve beyond that any peak in anything looks quadratic locally in the neighbourhood of the peak. We are fitting what in the grand centuries-long Hubbert linearization scheme of things would look like a little chunk of noise in the production curve. And asking is this the top piece of noise, or not?

I don't think this is invalid - this focussing in and out at various timescales. It's just important to remember the different contexts. Perhaps it would work better if I added a decades long graph and showed a box on it which this graph is the blow-up of. I'll try later and see if I can figure out a way to add a box to an Excel graph.

It's interesting that this local peak curve fitting clearly shows this local peak dead-on with the Deffeyes fitted global peak time of late '05 or early '06 (look at where the curves cross 0% production climb on the way down from this peak). You have a lot of other coincidences suggesting this may be the global peak, or at least one of a series on the plateau that won't keep up with the demand curve. Oil pricing has burst out of a fairly stable century long inflation/supply shock adjusted trading range in this '02-'06 time frame the way it did in a much more muted way back in the 70s demand surge. And you have had announcements begin for the first time in history during this time frame from Saudi Arabia that they can no longer play Federal Reserve for global oil supply by simply opening valves on existing capacity. Many things set this local peak apart from others that have occured.
I agree that it would be a good idea to show a larger curve and then a box which is a blow-up of the more recent data. It would let people see both the longer-term and shorter-term effects which are present.

I had thought that your analysis of Gaussian vs logistic curve was leading up to an analysis of world production levels using those same techniques. I have been waiting eagerly to see those results. If you could show the world percentage production change graph over the whole period, and fit a Gaussian curve to it, I think that would be very useful.

Then you could show this long-term graph with its fit, and the blow-up graph with both the blown-up long-term fit and also your short-term parabolic fit.

I believe the logistic curve shows linear production percentage declines in the vicinity of the peak, while the Gaussian of course shows linear declines throughout the whole curve. So I would expect either model to be showing a linear decline if we are near the peak now.

"I had thought that your analysis of Gaussian vs logistic curve was leading up to an analysis of world production levels..." Well, it is, but I don't necessarily go in a straight line. The goal of that analysis is to address how much/whether we should believe extrapolations of the linearization in the global case (you and others have raised questions about that). I want a really good understanding of what happened in the seventies before going deeper into that. I also need a longer time series for world production than I have right now. However, here's the Y-on-Y changes in the BP annual series from the Iran story the other day:


</center> The last sharp rise, ending in 2004, is the first half of the parabola above. You can contrast that with the US case: <center>

</center> (However, be aware that we know of no global equivalent to Alaska, which was known in 1970, unless you want to cast oil-sands/Orinoco/coal in that category - which is probably valid in the long term, but not in the short term).

Stuart, I didn't think, or mean to imply, that you were seeking to mislead. It's just that that particular graph was misleading to me.

Come to think of it, probably the visual impact wouldn't even be noticeable to the person who made it (you), since you were familiar with what the lines meant before you ever saw it.

The box-and-zoomed-graph would certainly clear up the confusion. A disclaimer on the parabola would also have helped; here, you're a victim of your own success, because so many of your fits are so meaningful and useful that I just assumed this one was intended to be meaningful too.


Ok - it's in an update.
A simple boxcar average (with the box centered) might be useful for smoothing the data without assuming any specific type of curve.  I'm assuming you are plotting monthly data, so a 3 or 5 month box might work well.
Still would like an explanation why seasonal variations have shrunk by 80% in the last 3 years. The summer minimums have kept rising sharply while the winter maximums have barely changed. The summer low of 04 was higher than the winter max of 03. It looks ominous.
Hm. The first thing that comes to mind is: Perhaps production can't be increased in winter (max) anymore, so it has to be increased in summer to make up the shortfall? This would imply that oil is being stockpiled for half a year, in some way that is inconvenient (or else it would've been used earlier) and may or may not be listed in standard statistics.

If this is the explanation, then yes, it seems very ominous that the low is basically gone. That would seem to imply that production can't now be increased any faster than the winter max has been increasing.


Just a suggestion.  Some correlation might be seen on the US component in API storage figures.
Hi Stuart,

Not that I think you need any more work to do, but I was thinking about the flat production curve for the last year or so and the rise in prices we've seen over similar sorts of time frames.

If you project increases in demand based on some sort of extrapolation of recent oil demand growth, flat production represents a virtual "loss" of supply that we would have had at a constant price.

I wonder how that "supply loss" and related price rise looks compared with your analysis the other day of the supply shocks and corresponding price increases of the 1973, 1979 oil shocks etc.


I did a quick scan of the 2004 and 2005 reports of the IEA

This is how they reported every month

Month    Year    Inc/Dec    Volume Reported
jan    2004    395000    82100000
feb    2004    -15000    82100000
mar    2004    60000    82000000
apr    2004    410000    82300000
may    2004    -440000    81500000
jun    2004    460000    82000000
jul    2004    790000    82500000
aug    2004    550000    83500000
sep    2004    300000    83600000
oct    2004    640000    84000000
nov    2004    890000    84600000
dec    2004    -40000    84400000

Total reported growth  4000000   
Total Growth (Dec-jan) 2300000
Exageration Factor + 1.7

Month    Year    Inc/Dec    Volume Reported
jan    2005    -45000    84400000
feb    2005    -645000    83600000
mar    2005    885000    84300000
apr    2005    365000    84200000
may    2005    435000    84500000
jun    2005    260000    84600000
jul    2005    -155000    84600000
aug    2005    250000    84700000
sep    2005    440000    84900000
oct    2005    -845000    83800000
nov    2005    865000    84400000
dec    2005    1300000    85000000
Total reported growth 3110000
Real growth (dec-jan) 600000
Exageration Factor +5.2

Interesting trend. If we take jan '04 as a baseline, then if we added all the reported growth numbers, we would today be at : 89.2 MB/day. Clearly there is a policy at work, because if this were a learning organisation there errors would level out over the years, but they don't.

"Missing Barrels" Reveal Bad Data, Says Matthew Simmons

From 2004 through 2005, the world produced almost 650 million barrels more oil than demand consumed, according to the accepted International Energy Agency model. But where are these barrels of oil?

"Little of this glut has shown up in observed petroleum stocks," writes Mr. Simmons, author of Twilight in the Desert: The Coming Saudi Oil Shock and the World Economy, in the January issue of World Energy Monthly Review. "In fact, as demand for oil continues to grow, daily usage rates indicate many key stock points are at historic lows."

Either bad data, or some govt(s) privately stockpiling.
Did anybody look under Switzerland?