« Save The UN! | Main | Life Expectancy And The Budget »

Forecasting Fun

I myself have been known to sound the alarms over future oil prices, but sure the good people at The New York Times' graphics department can come up with something better than the chart (reproduced below) that accompanies this article:

Why would you do a straight-line projection of a trend that, based on the historical data, has no such pattern? And why even chart this fact? US consumption versus US production has nothing to do with anything. It's a fungible, tradeable product. The issue is global production and global consumption. Oh well.

January 3, 2005 | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Forecasting Fun:

» US Oil Supply vs Consumption from CommonSenseDesk
Matt Yglesias links to a NYTs graphic that clearly should cause some concern. [Read More]

Tracked on Jan 4, 2005 11:31:55 PM


If you squint, the graph looks like a crocodile. Other than that, it doesn't show sh**.

Posted by: def | Jan 3, 2005 2:14:44 PM

"US consumption versus US production has nothing to do with anything."

1) It has something to do with the U.S. trade deficit. In most years, the bill for imported oil is about equal to the total trade deficit.

2) The trend lines over the past 2 to 3 years are anomalous because of rapid price rise in that period.

Posted by: liberal elitist | Jan 3, 2005 2:23:55 PM

Elitist -- that's just the point, when the price rises, the upward trend in consumption slows or even (as in the early 1980s) reverses. You can't just do a straight projection. Demand is self-limiting thanks to the price mechanism.

Posted by: Matthew Yglesias | Jan 3, 2005 2:30:31 PM

I'm disappointed. Based on your title, I thought you were going to be forecasting fun. You know, something along the lines of

Based on linear extrapolations from the past three decades, we confidently predict that the world will be 2.8 times as much fun in the year 2040 than it is today...

Posted by: Daryl McCullough | Jan 3, 2005 2:33:05 PM

Sheesh, what did they do, take the winter uptick and project it into the infinite future? If anything the 97-04 part of the graph looks like our consumption has pretty much levelled off--which surprises me considering the popularity of the SUV.

Posted by: Sebastian Holsclaw | Jan 3, 2005 2:35:57 PM

Matthew, maybe my eyesight is poor, but it looks to me like the period from 1981-2004 is, remarkably, essentially a straight line, at roughly the same angle as the projected consumption line. Indeed, that 25ish year period is the bulk of the time charted, so saying that the "historical data" has "no such [straight line] pattern" makes little sense.

You're right that the bottom portion of the chart is kinda senseless, but I don't see anything wrong at all with the shape of their projection.

Posted by: J | Jan 3, 2005 2:47:39 PM

Is the EIA to blame here?

Posted by: praktike | Jan 3, 2005 2:50:56 PM

There is some good information to be gained from the graph.

The peak in 1980 followed by the dip is entirely due to conservation efforts in the US. The CAFE standards of Ford and Carter (fuel economy) kicked in and lowered gasoline consumption by about 20%. This also put gasoline consumption below refining capacity. This is the real reason why the US did not increase its refining capacity between 1980 and 2000. There was a surplus. Note that consumption only returns to 1980 levels in the year 2000. Let me think. Gasoline prices hit $2 per gallon in 2000 the summer before the election and Al Gore suggested releasing oil from the reserve.

Since the peak of the CAFE effects in the mid 1980s, US fleet fuel efficiency has deteriorated. A big factor in the good economy of the Reagan era was cheap energy prices that were driven by energy conservation efforts under Carter and Ford.

Would a renewed commitment to energy conservation benefit the US economy?? Under Mr. Bush we are unlikely to learn the answer.

Posted by: bakho | Jan 3, 2005 2:51:19 PM

If anything the 97-04 part of the graph looks like our consumption has pretty much levelled off--which surprises me considering the popularity of the SUV.

Autos are only one component of oil consumption. Energy cost as a percentage of production cost has gone steadily down. Where else would you expect it to go? Technology improves efficiency, and efficient systems are preferred by those making rational decisions, such as businesses trying to improve their profit margin. I imagine another factor is just that we've offshored so much manufacturing that energy imports to factories in Asia are effectively taken off our books.

Consumers who think it's cool to drive a Hummer on their 30 mile single-occupant commute are not making a rational decision. OK, OK, they are maximizing "vehicular masculinity," an important component of their personal utility function--but if higher gasoline prices put a little more discipline in their behavior, you won't see me complaining. I agree that there are no grounds to extrapolate the future linearly, and probably significant grounds to expect a leveling off.

Posted by: Paul Callahan | Jan 3, 2005 2:54:18 PM

the bottom portion of the chart is kinda senseless

Why, doesn't it make sense to graph together what you use and what you produce? I know it's a global market, but still - every nation has its own balance of produced/consumed.

And what's the price elasticity of demand for oil - is it really that high? I doubt it - whatever the price is, you'll still need to drive and to heat your house, don't you.

Posted by: abb1 | Jan 3, 2005 2:58:42 PM

I mean, event during the oil crisis of the 70's - the drop in consumption what mostly due to gas not being physically available rather than too expensive - correct?

Posted by: abb1 | Jan 3, 2005 3:02:12 PM

Oh bullshit Matt, ignoring the bottom line, the top line is a very reasonable fit to over twenty years of pretty much straight line growth.

And what happened in 1980? Conserving, sweater loving Carter was replaced with Ronald Reagan.

You can certainly argue that there are better curves, that consumption will accelerate or flatten or responds to prices, but you can't argue that a straight line projection isn't a reasonable fit.

Posted by: jerry | Jan 3, 2005 3:06:47 PM

I think one reason the linear extrapolation is silly is that no such linear trend has ever been observed over such a long period of time. The one in the late 90s probably coincided with the economic boom. Even if oil supply were not an issue, it is not clear where the growth in consumption would come from within the US; you don't just burn oil for the heck of it, and over time technology yields equal output with equal or less enegy input. In fact, oil supply is slated to become an issue in the next few decades. Prices will rise and this will affect consumption. I would bet that we are more likely to see a new peak and a long decline.

As for forecasting fun, I project the supply of fun to be depleted by 2010 based on a lack of new discoveries and a steady decline in output of fun since late 2000.

Posted by: Paul Callahan | Jan 3, 2005 3:14:40 PM

I think, too, the point of the chart is that is no forseeable massive increase in American oil production, which is important to consider when the Bush Administration makes vague claims that the US can produce its way out of its oil issues.

Posted by: weboy | Jan 3, 2005 3:16:08 PM

Speaking of projection fun, did anybody ever hear the one about the man who--upon hearing that a certain wood stove would use fifty percent less fuel--bought two of them, figuring that way he wouldn't need any fuel at all.

Nyuk! Nyuk! Nyuk!

Posted by: Curly Joe | Jan 3, 2005 3:16:56 PM

Is the EIA to blame here?

I don't think so. Putting aside the question about whether the increase in consumption should be projected to increase in a straight line (there, in fact MAY BE good reasons for that to largely occur), Matthew's main objection to the chart is that, to be relevant to prices, the "issue is global production and global consumption". Exactly correct! But the EIA may not have intended the chart to give price-related information. In fact, it seems to me that the chart is quite useful for one important question: to show America's increasing dependence on oil imports. So, perhaps the idiot NYT used the chart for a purpose it was not intended for...

Posted by: Al | Jan 3, 2005 3:22:57 PM

Why would you do a straight-line projection of a trend that, based on the historical data, has no such pattern?

You don't.

Which is probably why neither the production nor consumption projections on that graph are straight lines. At least, they don't look like straight lines to me.

Posted by: cmdicely | Jan 3, 2005 3:23:30 PM

The other thing forgotten by Matt is that the national speed limit was repealed in the 1980s.

What would have been most interesting would have been showing the U.S. chart with charts for the EU, Japan, and China.

Posted by: Jeff I | Jan 3, 2005 3:23:33 PM

over twenty years of pretty much straight line growth.

Even if you look at '82-'02 as a straight line, it has a smaller slope than the projection. Using the "piece of paper against my monitor" technique, I would extrapolate to 25 million barrels in 2025, not 28 million.

But there is also not theoretical justification for linear growth. You might expect exponential growth, though thankfully this has not occurred. I expect prices to rise and efficiencies to improve. I think even SUVs are close to the limit given factors such as steering, finding a place to park, and the fact that Americans are neither growing longer legs nor having larger families. So where exactly is this growth expected to come from? (Oil consumption in China and elsewhere in Asia is another story.)

Posted by: Paul Callahan | Jan 3, 2005 3:23:43 PM

So where exactly is this growth expected to come from?

Population growth and economic growth?

Posted by: abb1 | Jan 3, 2005 3:27:55 PM

And what happened in 1980? Conserving, sweater loving Carter was replaced with Ronald Reagan.

That was 1981. Any trend you can show as having begun in 1980 began under Carter. I was just reading a dumb Republican article the other day where they found a trend that began in 1992 and tried to blame it on Clinton.

The other thing forgotten by Matt is that the national speed limit was repealed in the 1980s.

Huh? I know of only one such repeal, and it happened in 1995.

Posted by: digamma | Jan 3, 2005 3:30:12 PM

I don't think we will have sustained economic growth over the period of the extrapolation. Population growth is not all that high in the US either. I doubt we will see an 80% increase in population over the time of the projected 80% or so increase in oil consumption (from 15.5 million to 28 million barrels).

Posted by: Paul Callahan | Jan 3, 2005 3:30:20 PM

FYI, here's a short paper (PDF) on US elasticity of demand for oil. I dunno who the prof is, but it's worth a quick read. He concludes thusly:
This study shows that increases in oil prices of 1.6 - 4.5 percent a year are essential for sufficient growth of proven reserves to keep U.S. foreign oil dependence in check in the face of steady growth of economic activity and, hence, oil consumption. Bearing in mind that oil prices are not determined domestically, these findings accentuate the importance of alternative policies to ensure sufficient future domestic oil supply. Such policies may include less stringent leasing regulations for oil prospecting firms or fiscal measures (tax credits/subsidies) to boost drilling activity and to encourage research and development investments aimed at technological innovation and adoption in oil development and exploration. Rationale for these policies is reinforced if private firms undervalue the increased national oil security resulting from increased proven reserves.

Posted by: praktike | Jan 3, 2005 3:40:48 PM

Matt - your criticism of this NYTimes chart is very well placed. But note, this kind of nonsensical forecasting was also used by Vice President Cheney back in 2001 to support his pork ladened energy proposal.

Posted by: pgl | Jan 3, 2005 3:52:18 PM

On population: according to the Census Bureau, the US population grew from 226M to 276M from 1980 to 2000, and is projected to be about 350M in 2025. That's a 54% growth over the period, enough to explain a lot of the 80% projected consumption growth.

Still, I think this is a side issue to Matt's central point: forecasting demand like this is mad, because when prices rise, demand falls.

Posted by: william | Jan 3, 2005 3:59:07 PM

The comments to this entry are closed.