Monday, November 23, 2009

US Economic Recovery in the Era of Inelastic Oil



Ok, suppose you'd never heard of peak oil.

Or you didn't believe in it.

What could you conclude from the recent history of price and global oil supply?


 The figure below gives the key data - a measure of total global oil production and also of oil price (adjusted for inflation) over the last five years.





Global liquid fuel production, 2005-2009, left axis, together with spot oil prices, right axis. Source: Supply is average of EIA, IEA, and OPEC estimates. Price is spot price of West Texas Intermediate according to EIA, adjusted for inflation using the CPI to Jan 2008 dollars.


I think what you'd have to conclude is that, for whatever reason, we are in an era when oil supply is not very responsive to price. Both axes are zero-scaled, and the most noticeable thing on that scale is that price has fluctuated enormously while the amount of oil supplied and consumed has fluctuated very little. Huge price increases from 2005 - 2008 produced only a few percent increase in the quantity supplied. Similarly, huge price decreases from the beginning of 2008 to the end of that year produced only tiny decreases in oil supply.

Whatever you believe caused the price spike - speculators, Nigerian rebels, peak oil, Chinese drivers - one thing is clear - oil producers were not willing and able to ramp up production very much in order to take advantage of the best price for oil they'd ever been offered.

The cause of the price collapse starting in late 2008 is pretty uncontroversial - massive global recession - but still it's striking that oil producers only had to cut back a few percent of total production to arrest the price fall and get things moving upward again (a cutback that was mostly done deliberately by OPEC).

To see those fluctuations more clearly, I've shown the same figure with the y-axis blown up hugely so you can see the changes better.





Global liquid fuel production, 2005-2009, left axis, together with spot oil prices, right axis. Axes are not zero-scaled to better show the changes and correlation. Heavy lines are a five month centered moving average. Source: Supply is average of EIA, IEA, and OPEC estimates. Price is spot price of West Texas Intermediate according to EIA , adjusted for inflation using the CPI to Jan 2008 dollars.


At this scale, you can see that oil supply definitely does respond to price! There's been quite a strong correlation between the two recently - albeit with variable lags. The R2 (a measure of correlation between two variables) between the two thick lines over this period is 85%. That is, 85% of the variance in oil supply is explained by current prices (or vice versa). Clearly, if you model oil supply as "constant plus a small variation proportional to the current price", you've explained the great bulk of what's been happening lately. For what it's worth, the ratio of the two scales in the graph above is 25$/mbd (dollars per million barrels per day), and an actual regression gives the coefficient as $24.70 - implying a relationship in which we, the world's oil consumers, need to pay oil producers an extra $25/barrel or so in order for them to collectively kick in an extra million barrels per day.

An economist's preferred way of looking at it would be the elasticity - the ratio of the percentage change in the quantity supplied/demanded to the percentage change in the price. Expressed in those terms, the data show a price elasticity of about -0.03 to -0.04 - generally considered to be an extremely inelastic value (ie very large changes in price produce only tiny changes in quantity).

What does it mean if this continues to be the case over the next five years? Ie, if we assume the near future works in a similar way to the recent past?

Well, presumably it means that global oil production is going to continue to be within a few percent of its present value, unless the price gets very extreme by recent standards - over $200 say, or under $30. If the world economy does well, then prices will be high and oil production may grow a little. If the credit squeeze lingers and the world economy does poorly, OPEC may reduce production a little more to maintain prices. But either way, the evidence of the last five years suggests that global oil production will not change very much in the next five years.

In particular, the linear relationship I mentioned above implies that it would take a price of $200 to get 90 million barrels per day (up from the 84-87 mbd range we've mostly been in for the last few years). If we figure that people probably aren't willing to pay that much for oil while in the process of coming out of a severe recession, that implies we will, as a planetary civilization, have to get by on a little less than 90mbd. In fact, most likely, total global oil supply is going to be pretty close to its current value for a while, and we may as well assume for simplicity that it's constant - neither increases or decreases.

To get a better handle on the implications of this, lets look at the behavior of different regions consuming oil historically. In particular, I want to look at past eras of high oil prices so I'm going to go back all the way to 1965 now. I'm going to divide the world's oil consumers into five categories
  • OECD (ie. developed) countries (the US, most European countries, Japan, Australia, etc), 
  • OPEC oil producers (Saudia Arabia, Iran, Iraq, Venezuela etc), 
  • Asian countries not in the OECD (China, India, Thailand, etc), 
  • the former Soviet Union (Russia, Ukraine, etc), 
  • the rest of the world (eg non-OPEC Latin America and Africa). 
 I've plotted the annual oil consumption for each of these groups since 1965 in this next chart:





Oil consumption for various groups of countries 1965-2008. Source: BP.


I've drawn your attention to three periods of serious oil shock with rose rectangles. The first was the mid-seventies oil shock caused by the Arab Oil Embargo. The second was the oil shock at the very end of the seventies caused by the Iranian revolution and Iran-Iraq war. The last one was the oil shock caused by tight supplies from 2004-2008 - what I'm calling the era of inelastic oil.

I've chosen these groups of countries because they respond in different ways to the oil shocks. Firstly, lets take the OECD developed countries. In each oil shock, the response was to use markedly less oil, through some mix of recession and conservation, with the most pronounced case being the early 1980s, when OECD oil consumption decreased so much from 1979 to 1983 that it took until the mid 1990s for it to surpass the level of 1979.

The OPEC oil consumer societies, as one might imagine, are benefited by the high prices of oil shocks, and thus their internal consumption continues to grow (aided by the fact that they can afford to heavily subsidize domestic oil consumption, thus insulating their citizens from high prices at the gas-pump. As you can see, in the orange circles, the OPEC line emerges from the oil shock periods higher than it went in.

Similarly, the rapidly industrializing region of non-OECD Asia generally grows through oil shocks. In particular, it's striking that Asian oil consumption has increased markedly from 2005-2008, despite oil prices peaking at almost $150/barrel during that time-frame, as well as the worst global recession since the second world war.

The FSU has its own dynamics - basically an oil exporting region, its oil consumption history is dominated by the fate of the communist system - rapid industrializing growth in the 1960s and 1970s, followed by collapse in the late eighties and nineties, and a very slow fitful recovery since.

The rest of the world behaves like a more moderate version of Asia - it is developing and industrializing and that process continues through periods of high oil prices; oil consumption growth may slow down, but it does not go into reverse in the presence of the historical oil shocks.

It seems likely that these trends will continue in the near future. China, as the most obvious example, is running an enormous trade surplus and will continue to be able to pay for more oil to allow more of its consumers to drive as the country continues to develop. OPEC countries now have enough pricing power that they can maintain prices in the $60+ range and thus will continue to grow.

This next graph projects those assumptions out through 2015 - the period when we would hope the US and global economies would be recovering from the 2008-2009 recession. What I did here was to assume global oil production was perfectly flat - neither growing or shrinking. Then for each of the country groups other than the OECD, I assumed that their oil consumption grew at the same rate as the last five years - ie that they would be able to grow at least as well in the next five years as they did through a major oil shock and major global recession.

Then the OECD oil consumption is just projected as the difference between flat global supply and rising consumption elsewhere.





Oil consumption for various groups of countries 1965-2008. Source: BP with author's projections to 2015 as discussed in the text.


As you can see, this set of assumptions results in a significant fall in oil consumption in the developed countries.  In fact, oil consumption would fall by an average of four percent per year from now through 2015.  On the graph, this looks like a fairly natural extrapolation of what has happened in the last few years.  However, what has happened in the last few years has been pretty painful, so that's not so good.

In general, as this next graph shows, based on history, we would expect the US to respond to oil price shocks more or less as other OECD countries do. In all the past oil shocks, the US has cut back on oil consumption at the same time as other developed countries have. So faced with a need for the developed world as a whole to conserve something like 4% a year, we would expect the US in particular to have to do more-or-less that.


Oil consumption for the US and the rest of the OECD, 1965-2008. Source: BP.

The US, of course, is a very car dependent country, and thus a very oil dependent country. Two thirds of our oil consumption goes for transportation, and 96% of our transportation is oil powered. This next graph shows the split of passenger miles by mode in the US from a 2002 Bureau of Transportation Statistics report.


US Passenger-Miles of Travel by Mode: 2002. Source: BTS

As you can see, just about all of our mobility depends on gasoline, diesel, or aviation fuel - mostly in cars and light trucks. In turn, economic activity depends on using all that oil powered transportation. Pretty much any time an American goes to work, heads to the mall to buy stuff, goes out for dinner, or even has a package delivered from Amazon.com, oil is consumed.  Because a lot of US development is laid out in a fairly low-density sprawling way, most people and businesses have limited choice about this in the short term.  For this reason, historically, there is a very close relationship in the US between economic output - GDP - and and the total vehicle miles traveled.  This next graph illustrates that by looking at the ratio of total economic output created per vehicle mile.





US economic output per vehicle mile traveled 1960-2008. Source: FHWA for VMT data, BEA for GDP.  Update: that's real GDP in chained 2005 dollars. 

This ratio was fairly constant until recently, when it began to rise at a slow rate - about 1.5% per year from 1995-2008.  It seems most likely that this recent rise is due to the widespread deployment of the Internet making it possible to produce goods and services with less moving around.  This is an extremely helpful trend - however, 1.6% is still a lot less than the 4% decline in oil consumption projected earlier in this piece.

At the moment, a higher than usual fraction of US infrastructure is unused. For example, this Calculated Risk graph shows that 5.5% of the housing stock is vacant, versus a more historically typical 2.5%-4%.



Similarly, the office vacancy rate is now over 16%, versus a healthier 10-12%.  Correspondingly, unemployment is now over 10%, versus a healthier 5-7%.

Putting those people and resources to work will increase transportation demand - an unemployed office manager who gets a job will fire up her car to head to drive to that previously vacant office, and then later drive her family out to a restaurant to celebrate.

Thus, other things being equal, a US economic recovery that at least ends the under-utilization of existing resources will require a several percent increase in US oil consumption. Of course, a robust recovery with further economic growth would trigger the consumption of additional resources - new offices and houses and malls and hotels would be built - and require yet more oil to power movement between them.

In the meantime, as we discussed earlier, stagnant global oil supply and increasing Asian and Middle Eastern demand is going to require several percent per year decrease in US oil consumption.

The only way to square this circle is for the US economy to become more oil efficient, and to do so quite a bit faster than it has been.  Roughly speaking, the oil required to power the transportation for the economy can be thought of as a product of three factors:
  • How much total output there is
  • How many vehicle miles are required to create and deliver a given dollar of economic activity
  • How much oil is required to move the average vehicle a mile
In a post next week, I'll take up the data on fuel economy trends, and look at how fast we are going to have to improve in order to allow for the economy and employment to recover.

15 comments:

Datamunger said...

Looking at the data since 1973, I can't find a period where real GDP has grown while oil consumption has fallen 4%ish. (Except 1981 which just followed a recession in '80 and preceded something nasty).

Thanks for looking at vmt versus (Real?) GDP. Interesting.

Stuart Staniford said...

IIRC, fuel efficiency did increase around 4% a year in the late 70s early 80s. Given that we now also apparently have a lift from the GDP/VMT ratio improving, it seems like it probably would be possible to grow the economy despite that kind of decline in oil consumption. However, it would take more of a concerted effort than has been made so far.

Engineer-Poet said...

This is what I've been trying to say for some time, but I could never have mustered the concise data to prove it to skeptics.  Kudos!

And welcome back to blogging.  It is a compulsion, isn't it?

KLR said...

Stuart - fantastic that you're blogging on these matters again, as we're really at a major turning point in history with peak oil, climate change, and massive shifts in the transportation industry afoot.

Rapier tried to suss out whether grains and sugars were doing the US energy trade balance situation any good: R-Squared Energy Blog: Ethanol and Petroleum Imports Stupid downturn in demand buried that signal. Much of the commentary was niggling over btus; I've tried to chart the potential impact of the RFS over the next decade - forget energy content, on a volume basis ethanol should simply displace gasoline, with these issues of fuel line corrosion and decreased performance just a headache we drivers can deal with, no? Or is there a ceiling on non-pipeline transportation for ethanol that we'll run into before we're out of arable land?

This is a nice intro to this subject you've posted - I take it you'll start to break down oil into its various components in the next installment? Anyone who read your TOD pieces on transportation remembers that gasoline consumption had rebounded in the US from its '79 high in '86, after all.

I was struck by the data on Corporate Average Fuel Economy in this dedicated wiki; like you they've sussed out actuals for fuel economy, by what metric I'm not sure. This may be of use, I'm very keen on what's happening with that 4% efficiency wedge, given Obama's big promises. Dave Cohen points out that these are mostly carrots instead of sticks: Obama Tackles the Liquid Fuels Problem :: ASPO-USA: Association for the Study of Peak Oil and Gas.

Again, welcome back!

Datamunger said...

Yes, there are lots of fuel efficiency gains to had!

However, the last couple of years gasoline consumption per vmt hasn't dropped much (diesel has fallen much more)which may indicate that businesses are doing much better than households on the efficiency front.

[But I can wait for your next post]

Anyway... I'm wondering if without further policy action the economy is structured such that it responds by giving up jobs rather than having consumers become oil efficient. i.e. people without jobs are forced to sharply curtail consumption while people with jobs carry on as usual enjoying (along with the developing world) the lower oil prices the recession & jobless bring.

Glad you are back btw. :-)

Khebab said...

Stuart,

First, welcome back! I think your analysis is spot on, the oil not consumed by the OECD will be absorbed by emerging economies and large net exporters providing subsidies to their domestic market. The chart showing spot oil prices and supply is quite striking clearly showing that the market is extremely tight. The question is how low can the OECD consumption go? a 4%/year decline in consumption seems extremely damaging for an eventual economic recovery. The IEA is predicting a floor around 40 mbpd in 2015 for the OECD consumption.

Stuart Staniford said...

Greetings and thanks for the comments guys. In response to Datamungers initial paranthetic question - yeah, it's real GDP (chained $2005 as done by the BEA themselves).

Phil Hart said...

nice Stuart.. insightful and convincing analysis. but in your new incarnation you are quite a moderate :-) Even the IEA agrees with you now:

International Energy Agency calls 'Peak' on OECD Oil Demand

You are optimistic that increasing GDP oil efficiency is due to internet. Is it possible that the trend was growth of financial sector and asset price bubble that is now in decline too? or is last data point recent enough to suggest otherwise?

Stuart Staniford said...

Hi Phil - good to connect with you again, and that's a great catch!

Michael Cain said...

An excellent "welcome back" post. If I might expand on a couple of things implied by this?

Allocation of available oil exports between countries happens at the margin. That is, who derives the most value from an incremental barrel. In the developed countries, the incremental value is low: say, the entertainment value of a teenager getting to the mall. In the developing countries, the incremental value is higher. The changes in your oil consumption chart are to be expected, and the effect on developed countries' GDP is likely to be small.

This also suggests to me where the "burden" is likely to fall in the US. Poor people will be hit hardest, and the rural poor in particular. The $100/hour consultant will be able to afford the commute from the suburb. The minimum wage rural worker, less so. Rural areas also provide fewer opportunities for adjustment. I think this is going to be the big "stealth" energy policy problem for the US: the rural areas can't be abandoned, but most of the usual adjustments for high oil prices aren't applicable.

Chris Nelder said...

Great to see you writing again Stuart! I've missed your perspective.

You may be interested in this new study from U of Utah scientist Stephen Schneider, in which he claims that a simple physical constant links global energy use to the world's accumulated economic productivity, adjusted for inflation, irrespective of population growth or standard of living.

Also, if I may make quick plug for my take on the IEA report, I believe they used the whole climate change angle as a stalking horse for their fears about peak oil.

Looking forward to your future work!

Cheers,
--C

kjmclark said...

Excellent. You've been missed.

I think there's more to the OECD story. The US is 41% of the consumption of the OECD, and there's a wide range of responses in the OECD countries to the high prices of the past few years. Since '05, the US has cut consumption by 6.7% (BP; '05-'08), but the rest of the OECD only cut by 2.8%. The next five highest after the US cut consumption by 3.2%. In the past year ('07->'08), the change ranges from a 6.1% increase for Switzerland to a 6.1% decrease for the US. Even Germany saw a 4.6% increase over the past year.

I suspect that supports what Michael said. The US will experience a significant portion of the pain. It sure doesn't look like China will be cutting back any time soon.

Datamunger said...

RE: U of Utah study mentioned by Chris

9.7!

The previous estimate was 42 (Adams 1979)

The study will vaporize a lot of minds if the author hasn't made an error. Link to the pdf can be found here.

KLR said...

Funny, my BP data worksheet has a sum of -6.35% for the US 2005-2008. They are very alone in this category for major oil producers, next on the list for contraction is Qatar at -3.39%. Most are showing double digit advances. Norway and Canada show minor advances.

But many major economies had years in the last decade where they decreased their consumption of oil and still grew their GDP - France, Germany, Japan. I recall JD crowing about this on his blog, with good reason. Would be interesting to have some data about buildout of MT; maybe Alan Drake has an .xls.

kjmclark said...

KLR, I got data from BP SWRE 2009 at the link Stuart provided. I used the spreadsheet link they provided, looked at the "Oil Consumption - barrels" tab. I get US 2008 - 19419 kbpd, 2005 - 20802 kbpd, difference -1383, change over 2005 -.0665 or -6.7%. I bet we're looking at different tabs or something.

I can't quite figure out how they get their "Change 2008 over 2007" value either. For instance, for Qatar, they have an increase of 11.7% When I do (2008-2007)/2007, I get 11.5%. Their numbers are values, not calculations, so you can't see how they get there.