# Macro effects of oil shocks– what should we be looking for next?

I recently prepared an entry on the macroeconomic effects of oil shocks for the new edition of the Palgrave Dictionary of Economics. Here I sketch some of the material from that essay and explore the implications for where the economy may be headed next.

Consider the example of an individual airline that is wondering whether it should cancel some flights in order to save on fuel costs. For any given flight, the airline should compare how much revenue would be lost from cancelling that flight with how much costs could be reduced. The company offers a range of flights, some of which are much more profitable than others. If the lost revenue from cancelling a given flight would exceed the cost saving, it makes sense to keep the flight going. If the lost revenue is less than the cost saving, it makes sense to cancel the flight. If we identify the marginal flight– the one the company most regretted cancelling, but that it nevertheless did decide to cancel given current fuel costs– the loss in revenue from canceling the marginal flight would be approximately equal to the reduction in costs. This is the familiar condition that profit maximization calls for setting marginal revenue equal to marginal cost.

Let Y be the number of passenger-miles the company provides, PY the price customers pay per mile, and Δ denote change. Then the lost revenue from canceled flights might be represented as PYΔY. If E denotes the quantity of energy consumed and PE the price per unit of energy, then the cost reduction would be given by PEΔE. Equality of marginal revenue and marginal cost in this case would require

PYΔY = PEΔE .

The above equation comes from analyzing the way that a particular firm’s choices determine the quantity of energy that it ends up using. But we can invert the equation to ask the following question. Suppose that as a result of a change in energy prices, firms are forced to reduce energy consumption by a particular amount ΔE. By how much would we expect to see their output decline? Assuming that this reduction in energy use is achieved through the market mechanism (i.e., all users get to choose whether or not to reduce energy consumption and by how much, so that the reduction is borne by the marginal activity), the predicted effect on output is given by

ΔY = PEΔE/PY .

Dividing both sides of the above equation by Y and multiplying numerator and denominator on the right hand side by E results in

ΔY/Y = [(PEE)/(PYY)] [ΔE/E] .

Let α denote energy costs as a fraction of total revenue, α = [(PEE)/(PYY)]. Then the above equation states that

Y = α %ΔE ,

so that the percent change in output resulting from a 1% reduction in energy use should be given by α, energy’s dollar share in total output. (For more details on this and the following discussion, see my entry for the forthcoming Palgrave Dictionary).

Date Event

Drop in world
production

Drop in U.S.
real GDP

Nov. 1956 Suez Crisis 10.1% "center"> -2.5%

Nov. 1973 Arab-Israeli War 7.8% "center"> -3.2%

Nov. 1978 Iranian Revolution 8.9% "center"> -0.6%

Oct. 1980 Iran-Iraq War 7.2% "center"> -0.5%

Aug. 1990 Persian Gulf War 8.8% "center"> -0.1%

If we thought of measuring α by the dollar value of U.S. crude oil expenditures as a fraction of GDP, we’d come up with a value below 4%, meaning that a 10% reduction in oil supplies should result in only a 0.4% drop in real GDP. The table at the right summarizes the drop in production associated with the 5 most dramatic historical oil supply disruptions along with the magnitude of the decline in U.S. real GDP that we observed in the U.S. between the oil shock and the trough of the subsequent recession, typically an interval of a little over a year. Since in a normal year we would expect to see GDP grow by 3.4% rather than fall, these numbers imply an effect on GDP that is an order of magnitude larger than the above calculation implied. Lutz Kilian has alternative calculations of the size of these disruptions that would lead to even smaller predicted values, increasing the puzzle.

My own interpretation is that energy disruptions only start to matter a great deal for the economy when utilization rates of other factors of production besides energy are observed to adjust. For example, in deciding to cancel flights, the airline is not just using less energy but also likely laying off workers. A typical pattern in the above episodes was that consumers suddenly became very apprehensive following the supply disruptions, postponing big ticket purchases such as automobiles. As automobile sales declined and workers were laid off in autos and the industries that sell to the auto makers, further cutbacks in spending by those affected led the economy into recession.

So where do we stand right now? In response to the rapid run-up in gasoline prices in August and the devastation from Katrina, the University of Michigan’s index of consumer sentiment fell from 96.5 in July to 76.9 in September. Consumption spending fell 0.5% in August, with sales of many SUV’s down 50% in September compared with the year earlier. And today Delphi, the largest U.S. auto parts supplier, filed for bankruptcy.

On Friday we further learned that U.S. nonfarm payroll employment fell by 35,000 jobs in September. Given that we’d normally expect to see a monthly increase in employment of 150,000 jobs, the September figure amounts to 185,000 jobs lost. Although this loss was evidently smaller than many other analysts had predicted it was very close to the 200,000 number that I suggested on Sept. 7 for the size of the effect that we might expect to see from Katrina itself. I also noted then that 200,000 lost jobs would amount to about 1/4 of a recession-inducing employment shock. See also excellent discussions of the latest job figures by Calculated Risk and Macroblog.

The key question now is very much the same one I raised a month ago, namely, how the Katrina-induced unemployment will interact with the other macroeconomic disruptions that are also incipient in the other September data discussed above. We’ll have a much better view of this in another month. The key indicators that I’ll be watching for are further declines in consumer sentiment and spending, the timing and magnitude of the layoffs in auto- and airline- related industries, whether investment or export spending can take the place of reduced consumption, and whether house prices and construction join in with the other negatives.

Will they or won’t they? Stay tuned– we’ll find out soon enough.

## 20 thoughts on “Macro effects of oil shocks– what should we be looking for next?”

1. freejack

A move by BushCo to prevent the opening of an International Oil Bourse in Iran which will trade oil for Euros rather than US dollars? (just my guess)

2. Joseph Somsel

Historically, our energy usage patterns have shifted over time and this might explain the GDP to oil disruption data above. A good source is the EIA here:
http://www.eia.doe.gov/emeu/aer/overview.html
Looking at the data from the EIA against the oil crises mentioned about, a critical value is the ratio of imported oil against total US energy consumption. Call this ratio imported oil intensity.
The Suez Crisis occurred at a time of continuous increase in imported oil intensity. We were using more oil and had to bid for global supplies.
The ’73 war occurred at a time of domestic oil peak. The US was making a big jump in imported oil intensity in spite of the boycotts.
The 1980 war was a time when US nuclear plants were coming on line and oil was losing market share in electric markets. Domestic natural gas was also making inroads in electric markets. From 1977 to 1986 saw imported oil intensity decline to pre-1972 levels.
The 1990 war saw a three year retreat or plateau in imported oil intensity.
I’d conclude that the GDP hits are worst when we’re increasing our use of imported oil while global oil supplies are tightening. We’re much less sensitive to oil disruptions GDP-wise when we’re stable or decreasing our oil importing.
Seems obvious but the data above from JDH and from EIA supports this. Since 1992, we’ve increased our imported oil intensity almost every year.
Since petroleum is today used most extensively and with fewer substitutes in transportation, one can look at the percentage of energy used for transport. Again, the EIA:
http://www.eia.doe.gov/emeu/aer/consump.html
The ratio of transport to total energy consumed is today at its highest (beginning from 1949) at 28%. Its slowly trended up from a low of 23% in 1966. But don’t put all the blame on SUVs! The ratio of jet fuel to motor fuel (gasoline) has increased from 13.9% in 1966 to 17.8% in 2004.
Somehow the airline industry consumption has been growing even faster than road travel. Backing off on jet travel and doing more video conferencing looks like a better idea than imposing a gasoline tax.

3. Hal

What about the possibility of inflation in response to increased oil prices? This is being discussed on theoildrum.com and elsewhere. Most people assume that increased energy costs will automatically lead to inflation, because everything takes energy to make or produce, both directly and indirectly. Do you have data showing the effects of energy price rises on general price levels, similar to the chart above for GDP?

4. Tom Leeman

According to Matthew Simmons Higher oil prices do not curtail consumption between 1969 and 1978 global demand increased by 20 million barrels a day this during a decade when oil prices increased 14 fold. (Remember oil went from \$1 a barrell in 1969 to \$40 during the Iran hostage crises) The fall in oil prices won’t last long, winter is coming and it’s not only crud but natural gas that will be taking a bite out of consumers pocket books. The catipillars coats are thick & fuzzy this fall we could be in for a long hard winter.

5. Joseph Somsel

One of my profs in engineering school in the ’70s was following this. He asserted that the price shock in ’73 would cause general price levels to increase so that oil and and general prices would equilibriate back to before the oil price increases.
By the end of the ’70s, he was right. In other words, they wanted more money than oil was worth so we gave them cheaper dollars. Inflation.
If you followed Stuart Staniford’s logic in an earlier thread, he would predict that as general EROEI increased, “stagflation” would result – increasing prices and lower or negative growth. Just one party charging monopoly rents is only a change in ownership – remember recycling of petrodollars?
I agree with Stuart and predict that peak oil (and peak natural gas) will result in more capital being diverted to energy sources with lower EROEI – the Hirsch report’s alternatives and more nuclear power are the leaders. That will stagnant growth as excess production now has to support maintaining energy inputs.
In other words, we have to work harder to maintain ourselves and will have less left over. So just increasing the price because you can has different effects than if EROEI is trending lower.
If you look at the capital requirements for the Hirsch alternatives, you see that we can compensate a 3% depletion rate on a 100 mbpd economy on \$100 billion a year with an assumed equal balance of Hirsch methods (CTL, GTL, oil sands, and heavy oil). I’ve a rough spreadsheet if anyone is interested – just email me.
That’s about what we’re spending on oil E&P investment now (I’m guessing) so we’re doubling petroleum capital requirements at 3%, at least at first. That’s cash and resources that don’t get invested elsewhere, and people choising careers as petroleum engineers rather than lawyers (we hope.)

6. bubba

What I see is people making a lot of irrational choices that have no real explanation in conventional math or economics. This is similar to what happened after 9-11. The fear and uncertainty from an imagined threat (that being continuous and ongoing terrorist attacks throughout the US) paralyzed the US economy and brought on a worldwide recession.
Recently we lost a whole city to a flood. There is tremendous political pressure to rebuild it even without thought of whether or not it can be protected from such an occurence in the future. And right now, from where I sit in Houston, I don’t see a lot of jobs going back to New Orleans.
2.8 million people fled a hurricane in SE Texas which resulted in more people dying on the road than would have died if most had stayed home. The fear in the population was palpable. The gas and provision hording was disgusting. There were places where the feeling was everyone for themselves.
For most Americans, there is no understanding as to why the prices at the pump are so high. They think it is either 1) oil companies screwing them or 2) related to the two hurricanes. They cannot fathom that it is a structural worldwide problem where excess production capacity has disappeared and where excess refining capacity has disappeared (and on top of this we had two hurricanes!). They are going to be shocked this winter when they start getting their heating bills an they are 50% higher than last winter.
The first reaction will be to demand a political solution. Congress will be happy to oblige the populace with a non-solution.
In the meantime, people will stop buying conventional SUV’s but the wealthy will spend twice what they would have spent for a conventional SUV to buy a hybrid SUV that MIGHT save them some gas money over the next few years, depending upon the kind of driving they do. Their maintenance costs will triple, however.
Like I said, irrational behaviors. However, maybe that is what the Freakonomics Guys are onto.

7. Joseph Somsel

Bubba,
“The first reaction will be to demand a political solution. Congress will be happy to oblige the populace with a non-solution. ”
I’m seeing that Congress is getting very nervous about this winter. Heating costs will increase greatly, on top of gasoline prices post-hurricane season. A long cold, hard winter could result in natural gas shortages and maybe people dying in their unheated homes. With all our new gas-fired electrical generators, electric power shortages might make things that much worst. I’d give such a scenario (frozen voters and power outages) maybe a 10 to 20% probablity.
They did pass the Energy Bill this summer. It only did a few things right but it did break some logjams. Friday, I heard that Harry Reid and Pete Dominici (Senate Energy Committee) are negotiating changes for Yucca Mountain (\$100 billion to bury a trillion dollars of power plant fuel?) Plus, I suspect that the people looking to pay for the next generation of nukes are asking how the Federal government will take physical delivery of spent nuclear fuel for these new plants.
Jeb Bush also changed is mind about drilling offshore Florida. Will Arnold be next?
As an aside, if people are looking to lynch excess profit makers, why not look at the companies that own the LNG terminals? They have to be making it big time with domestic prices at \$14 and LNG landed costs at \$6.
Expect more push for new LNG terminals.

8. Bill Ellis

I presume that you recognize that the illustration of the airlines’ fuel cost vs passenger revenue is greatly over simplified. The cancelling of a flight may improve revenue because the reduced number of seats available on a route may result in higher loads on other flights on that route, and higher per seat revenue because it will not be necessary to discount as many tickets.
In addition, not all routes will react the same. Consider that Chicago to D.C. is generally a high percentage of business travelers, and except to the extent that there is less business, the route will continue to carry about as many passengers. (With Delphi’s attorneys based in Chicago, actual load may increase.) While one might expect vacation flights to decrease when faced with higher prices, there is probably some correlation between the cost of the destination and the change in travel plans. High income fliers continue to fly while low income travelers make alternate choices.
Bill

9. M1EK

“hybrid SUV that MIGHT save them some gas money over the next few years, depending upon the kind of driving they do. Their maintenance costs will triple, however.”
sigh.
No added maintenance costs; in fact, arguably less (brake pads, etc.). There’s already the hybrid cab driver in Vancouver to look at for the 200,000 mile scenario.

10. Bill Ellis

I understand that the cost of a set of replacement batteries is about \$4,500. If so, and assuming that you might get 6 years life out of a set of batteries (twice any battery powered device I ever owned), then at the end of six years the vehicle will not be worth what it will cost to replace the batteries. No maintenace. Just like a toaster. When it quits working throw it away.
Bill

11. Barkley Rosser

Jim,
Well done. However, did you not exceed the 3500 word limit for these entries? Also, you should identify it as the Second Edition of the New Palgrave Dictionary.

12. odograph

Anyone interested in all the dope on Prius batteries and their replacement costs can search the PriusOnline forums for latest info:
http://www.priusonline.com/
I think they’ll last much longer than I’ve ever kept a car … but this isn’t really the place to discuss it.

Macroeconomic Oil Shock in Progress?

Jim Hamilton says: maybe. Econbrowser: Macro effects of oil shocks– what should we be looking for next? : If we thought of measuring %u03B1 by the dollar value of U.S. crude oil expenditures as a fraction of GDP, we’d come up with a value below 4%, me…

14. TI

I think it isn’t quite proper to treat hurricane effects simply as an oil shock. Here we have both oil and natural gas production disruption, but only locally, not globally. These disruptions are only temprorary and perceived as such. This means easy adapting by temporary meeasures. Long-term plans are not changing (may be in New Orleans…).
Besides there are lots of demand destruction – literally. Refinery capacity is lost so refineries need less crude. Strategic reserves are in use – they exist for these kind of eventualities – both in the US and Europe. No wonder oil prices are not reacting much.
These factors make comparison to the earlier oil shocks difficult and using models fitted to them may not be useful.
The table of the oil shocks and their impact on the US GDP seems to overstate the changes in the oil supply. BP statistics don’t give so big changes year-to year basis.
The natural gas supply problems hit probably worst. There is not enough LNG importing capacity and the decreasing production in North America is a permanent and worsening phenomenon. This problem cannot seen as merely a temporary one. The gas prices will hit many households and businesses (both power shortages and prices).
We could expect the strongest effect through housing markets. Rising heating and power costs plus more expensive commuting might break the back of many households that have taken mortagages up to their limits. This may lead to foreclosures which will take down real estate prices and pick the housing bubble. Power shortages may show in industrial production and jobs. But we will see quite soon.

15. Joseph Somsel

TI,
I agree that the natural gas disruptions will probably have the larger effect over the next 6 months. Oil and gasoline can be delivered by tanker.
The gas that didn’t get produced was supposed to flow into storage for this winter. When winter comes, a cold streak means that the pipelines from the gas fields are max’ed out and the local companies have to drawn down their storage to meet demand. If storage is not topped off before cold weather hits, then somebody gets cut off.
The most hopeful news is the NOAA long range forecasts for this winter here:
They predict the heartland will be warmer than usual but decline to call the Northeast and the Southeast. Of course, these predictions are not very reliable but its something!
The solution for declining natural gas production is more coal and more nuclear, so we can burn less gas for electricity and heat our homes with electric heat pumps. The last two weeks I’ve seen capital cost estimates for two new coal plants at \$1800 and \$2000 per kWe. They were 500 and 600 MW each so there are economies of scale for a 1500 MW, full size plant which would mean a lower \$/kW.
New nuclear at 1500 MW can cost maybe \$1500/kW but with lower operating costs, especially for fuel. However, coal doesn’t scale as efficently as nuclear so its a horserace!
Unfortunately, the easy way out is building LNG terminals. That’s just going to worsen our balance of payments and make us more dependent on the same set of characters to whom we’re dependent today (largely).

16. JDH

TI, the size of the quantity disruptions that appear in this table do not represent year-to-year drops in global production but rather peak-to-trough drops in production in the specific affected countries– details of where these numbers come from can be found in Table 4 here.
Also, I agree very much that Katrina is not just a pure energy shock, as I discuss here.

17. Stuart Staniford

So JDH, the theory that the change in GDP due a a change in oil availability should be proportional to the fraction of GDP being spent on oil before the shock is wrong by a factor of ten when compared to this data. I appreciate your honesty in pointing this out! Clearly, that theory is missing the main effects, whatever they are.
The very crude theory that says that change in (expected) GDP should be proportional to the change in oil supply is wrong by a factor of around 2 (8-10% reductions in oil supply, versus an expected few percent increase in oil supply create about a 0-3% reduction in GDP versus the expected 3% increase, ie a 3-6% percentage point difference from the expected). Ie, it’s a lot better but still not very good.
The next best theory, in my mind, would add in the elasticity to try and explain some of that factor of 2. The elasticity is effectively exploring the degree to which some uses of oil are more marginal than others, and the marginal ones go first – the elasticity is about how much marginal use there is, relative to use people aren’t willing to give up lightly. I think doing that analysis is getting beyond the scope of a comment – I’ll have to post on it over at Oil Drum.