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
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
Dividing both sides of the above equation by Y and multiplying numerator and denominator on the right hand side by E results in
Let α denote energy costs as a fraction of total revenue, α = [(PEE)/(PYY)]. Then the above equation states that
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 |
| Drop in U.S. |
|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.