28 February 2011

Oil Prices and Economic Growth

Last week I solicited perspectives on the relationship of oil prices and economic growth.  Thanks to all who emailed and commented.  This post shares some further thoughts.

First, there does appear to be a sense of conventional wisdom on this subject.  For instance, from last Friday's New York Times:
A sustained $10 increase in oil prices would shave about two-tenths of a percentage point off economic growth, according to Dean Maki, chief United States economist at Barclays Capital. The Federal Reserve had forecast last week that the United States economy would grow by 3.4 to 3.9 percent in 2011, up from 2.9 percent last year.
Similarly, from Friday's Financial Times:
According to published information on the Federal Reserve’s economic model, a sustained $10 rise in the oil price cuts growth by 0.2 percentage points and raises unemployment by 0.1 percentage points for each of the next two years.

Jan Hatzius, chief US economist for Goldman Sachs in New York, comes up with similar numbers ...
Today's New York Times sames something very similar:
Nariman Behravesh, senior economist at IHS Global Insight, said that every $10 increase in the price of a barrel of oil reduces economic growth by two-tenths of a percentage point after one year and a full percentage point over two years.
And adds:
As a rule, every 1-cent increase takes more than $1 billion out of consumers’ pockets a year.
Reuters offers some different numbers:
In 2004, the International Energy Agency calculated that an increase of $10 per barrel would reduce GDP growth in developed countries by 0.4 percent a year over the following two years. It would also add 0.5 percent to annual inflation. The impact was more severe in the developing world: in Asia, growth would be 0.8 percent lower and inflation 1.4 percent higher.

But the IEA’s estimates were made when the oil price was just $25 per barrel. While a $10 price increase today is lower in percentage terms, the absolute level is much higher: at the current price, oil consumption accounts for more than 5 percent of global GDP.
That these numbers seem unsatisfying 9at least they do to me) should not be surprising, as economists have devoted precious little attention to understanding the role of energy in economic growth.  David Stern of the Australian National University has a nice review paper titled "The Role of Energy in Economic Growth" that asserts:
The principal mainstream economic models used to explain the growth process (Aghion and Howitt, 2009) do not include energy as a factor that could constrain or enable economic growth, though significant attention is paid to the impact of oil prices on economic activity in the short run (Hamilton, 2009).
James Hamilton, cited above by Stern and a professor of economics at UCSD who blogs at Econbrowser, explains the math as follows:
Americans consume about 140 billion gallons of gasoline each year. I use the rough rule of thumb that a $10/barrel increase in the price of crude oil translates into a 25 cents per gallon increase in the price consumers will eventually pay for gasoline at the pump. Thus $10 more per barrel for crude will leave consumers with about $35 billion less to spend each year on other items, consistent with a decline in consumption spending on the order of 0.2% of GDP in a $15 trillion economy.
So much for that fancy Federal Reserve model, but I digress.  Hamilton has a new paper out on oil shocks here in PDF.

From this cursory review, it seems that the details of the relationship of energy prices and economic outcomes remains fairly cloudy in the economic literature, with conclusions resting significantly on assumptions and the specification of relationships.  Even so, the big picture is clear enough to draw some general conclusions, such as this prescient assertion put forward by Professor Hamilton in 2009 testimony before the U.S. Senate:
Even if we see significant short-run gains in global oil production capabilities, if demand from China and elsewhere returns to its previous rate of growth, it will not be too long before the same calculus that produced the oil price spike of 2007-08 will be back to haunt us again.
The conclusion that is draw is that regardless of the best way to represent oil prices and GDP in economic models, we need to work harder to make energy supply more reliable, abundant, diverse and less expensive.


  1. Most of the reasoning in the various quotes refers to the short term (1-2 years), and is mostly derived from some regression coefficient in an econometric model linking the price of oil and the increase in GDP. Other examples are cruder, like the one multiplying the increase in the gas gallon price by the number of gallons sold and deducing that as much money would not be available to consumers to spend in other things.
    Both "approaches" assume other prices and income levels remain constant. If the gallon price rises, it may cause rises in other commodities such as food, transportation and what not; and if my pay rises accordingly, the end situation might be the same as at the beginning.
    The real question is change in relative prices, i.e. oil (or more generally energy) becoming relatively more expensive in regard to other goods and services. And of course, after an exogenous shock such as a war in Kuwait or an uprising in Libya, prices may jump wildly but later settle down (as one may gather from looking at the chart at the start of this blog post). The wild ups and downs of oil prices since WWII have not caused a similar degree of ups and downs in economic outcome, although of course there are booms and recessions all the way through. This is because Man does not live on oil alone: other things also matter.

  2. "The conclusion that is draw is that regardless of the best way to represent oil prices and GDP in economic models, we need to work harder to make energy supply more reliable, abundant, diverse and less expensive. "

    I can't imagine that you'd find much disagreement with this statement Roger.

    The real question, of course, is how does 'work harder' translate into action? Is it efficiency regulations, alternative fuel mandates, RD&D, carbon taxes? I think I know how you'd anwser :)

    Wrt to the question of oil, the challenge as I see it (particularly in the N. American context), is that there is a great deal of inertia in the current system from a technical point of view (i.e. slow fleet turnover, refuelling infrastructure, built environment, lack of viable transit alternatives) that makes it very difficult to change to any sudden, but sustained, increase in the price of oil. This is why you have such a disparity between short-term and long-term elasticities for gasoline. In the short-term, your options are pretty much limited to doing the speed limit instead of 80 mph...

  3. For something so fundamental to the economy, energy gets curiously little attention from economists (in a way, economics is ultimately a study of energy).

    A lot of the oil price/recession arguments suffer from something of a correlation/causation problem. It' not always clear if an oil price spike caused or simply presaged a recession.

  4. Is our reasoning backwards here? Eyeballing that graph, it strikes me that there is as good a case for recessions sparking drops in oil prices as there is for rises in oil prices sparking recessions.

  5. "is that there is a great deal of inertia in the current system from a technical point of view"

    In 1976 the average weight of a US automobile was 4,079 pounds, by 1981 that had dropped to 3,202 pounds then slowly crept back up to 4,021 pounds in 2003.

  6. Putting $0.25/gallon in perspective for a family

    The average miles driven/car/year is about 13000m I think. Itmay have changed somewhat since I last looked it up
    The average fuel economy is 20MPG for the US fleet.
    13000/20 X 0.25= $162/ year

  7. I don't know who Stern is, but James D. Hamilton is the author of a widely used and almost encyclopedic, graduate time series textbook ("Time Series Analysis"). That said, I am not so certain there is a lot of insight in the effect of oil prices on economic growth. One, if not the, problem here is sorting out the exogenous from the endogenous changes in oil prices for a commodity whose demand curve may include the change in price as an argument.

    I presume that the estimates from the Federal Reserve model or IEA is simply about the effect of a purely exogenous oil price increase, other exogenous factors being constant.

    Profane's eyeball analysis probably has more wisdom than an estimate of a pure exogenous effect. Recall back in the 70s the effects were estimated at several times the current estimates, but the 80s (Reagan) recessions (where two recessions followed an oil price shock) were monetary policy induced confounded a macro-theory-neutral estimate of the effects. And the last run in oil prices seems almost certainly a case of endogenous price change from an increase in demand due to economic growth. Was the oil price increase a negative feedback? Sure, but so were a lot of things, the housing price bubble of course being the most important.

    I would be much less skeptical of estimates of the effects of oil price increase if those estimates where expressed in terms of the rate of change in oil prices and the relative volatility, or lack thereof, in the previous years of oil-using capital equipment.

    Finally, I disagree that we need to work harder to make oil more ... Harder means more dollars which have other good uses also applying to economic growth. It would be nice if we have more reliable substitutes -- particularly substitute processes that used other fuels or less oil -- but you need to show that investing in such things, with their inherently uncertain success, is cost effective, at least in an expected value sense.

  8. The relationship is complex because price is the observed product of many factors. Isolating simply supply on the one hand, or demand on the other, you have a very different picture of the effects on commerce (namely far more deleterious as price increases are affected by the former- Chinese/Asian/EM demand, however, is clearly the latter).

    Likewise increased prices will have different effects over the short and long term. Likewise when the change in prices has ramifications for the energy mix, given that there are massive discrepancies in external costs and benefits across them, (from the better known, i.e. local, regional and global pollution, to the more commonly acknowledged, e.g. sponsorship of 'rogue states' and ultimately terrorism, to critical factors that are almost never discussed, i.e. the effect of the petroleum trade on highly deleterious international imbalances, which were so critical in setting the stage for the financial crisis).

    Roger, what you have itemized here as consensus is both narrow and highly misleading, in its glossing over of the relevant dynamics. I suggest you dig a little deeper if you are actually interested in this topic.

  9. There may be a big difference between the short term and long term effects of an oil price shock. One of the long term effects is likely to be increased substitution of alternative fuel sources. This already occurred once in the US, following the 1974 oil shock natural gas was substituted for oil in home heating. Following this shock, I predict widespread substitution of (natural/shale) gas for oil in transport, at least in commercial contexts. UPS is already buying gas powered trucks. Many are likely to follow.

  10. Why is there so much noise in the more recent portion of that graph than in the older portion? Does that noise represent real day-to-day or week-to-week fluctuations in the price of oil? Or is it an artifact of some kind?

  11. #11 Why is there so much noise in the more recent portion of that graph than in the older portion?

    It's a result of OPEC attempting to 'manage price' by managing production. There is always a time lag.

  12. Ok Roger. The first thing to note, as Stern has, is that the consensus you're reporting is the consensus answer to the very narrow question of what the implications are for increased oil prices on the just US economy and just in the very short term- typically around a year, (these models are brute forced estimated over the empirical data with all the eloquence that implies. Their forecasting skill, such as it is, diminishes rapidly). This is great to know if you're a monetary policy maker or a hedge fund manager (and the latter tend to be slightly important clients of the institutions employing these economists than your average policy wonk). Much less, however, is it useful as a guide to long-term policy or indicative of some larger economic truth.
    The second thing to note is that the agreement between the back of the envelope calculations by Dr Hamilton and the Fed model are far more coincidental than it appears you give them credit for. US based production is equal to about 50% of US demand (and total US 'ownership' of oil production revenues is probably higher). Increased costs to the US consumers of 140 billion gallons of gasoline are increased revenues for these domestic entities. The fact that the net effect in the short-term is negative is not simply a consequence of the considerable import fraction of this tab, but also a consequence of the considerable discrepancy between the marginal propensity to save of the respective parties. And this confounding influence does not include the effect of higher petroleum prices on the revenues of related and more domestically concentrated industries, e.g. refining, oil services, and on investment in the energy sector, both of which will factor into the ultimate realization of macroeconomic variables.
    All of that is not to mention the other considerable sources of petroleum demand in this country, e.g. industrial, utility, heating oil and other distillate, etc. nor is it to mention the fact that, even in the short-term, price rises result in some substitution effect which adds demand to other goods and services. And of course there is the knock-on effect of all and sundry in an economy where my demand accrues to your income and my savings affects your cost of capital. It's just not that simple. We do import around half the petroleum we consume and something like half of that does end up in the gas tanks of our vast automotive fleet, but it's just not that simple.

  13. The second thing to note is that the oil price spikes of the past as depicted by the Times graphic were driven by supply shocks and analogous spikes in what's known as the convenience yield (relatedly, there has clearly been a 'hand-in-hand' effect of speculation, pronounced in the late seventies and early eighties, and now... and how**). This type of 'price shock' is actually stagflationary at a global level- new equilibrium levels of output will be lower at higher prices- which has important implications for the economy when you consider the significance of the good, the inelasticity of its demand and implications for credit in the context of Minskian business cycle dynamics (IMO the only kind worth discussing).
    What we have been experiencing of late is, however, more demand driven. This is not to say supply is unconstrained, and certainly at historical production/distribution cost- there is, at a minimum, some truth to 'peak oil'. However, demand growth has been the story as quite literally hundreds of millions of people have been lifted out of poverty in the past decade and a half. This type of price shock is not stagflationary. It is inflationary, depending on the policy response, but it should result in higher levels of global income. This means that type of thing is not in any way a bad development at the global level ceteris paribus.
    At the US level and in the short-term, that is a bit more ambiguous. However, one must still balance the cost of weakened consumer purchasing power with the benefit of robustly growing trading partners, with all its implications for US workers/consumers (or, as happened in the last decade, with outflows recycled back into the US bond market making for a free money party, bumper residential investment and one hell of a nasty hangover).

  14. The third thing to note is that long-term considerations of what happens in the petroleum market are much more benign than they are in the short term. The extreme inelasticity of demand to price is not the case as capital is given time to accumulate and labor augmenting technological growth given time to act over a different solution set. This paper, which ostensibly speaks to the economics of climate change, but whose implications are broader, is an example of what the literature has to offer in terms of describing these dynamics which are IMO far more relevant to policy than what you've cited here: http://www.econstor.eu/dspace/bitstream/10419/24483/1/dp0162.pdf
    The final thing to note is that externalities of energy production and consumption are, of course, not priced. They are also massive and vary quite widely. Each source of energy, even the renewables (not least ethanol), have considerable environmental impact. It is certainly fair to say though that, from an environmental standpoint at least, and assuming that some day some way the nuclear fuel cycle could be resolved, or the waste problem at least more convincingly managed, that fossil fuel energy and especially coal are substantially worse than the alternatives. That relative difference has considerable implications for the effect of price changes on the mix of energy consumption for obvious reasons (i.e. society can be made poorer from an income account perspective but richer from a utility perspective with higher prices on fossil fuels). There is plenty of scope for this along this dimension alone.
    However, when it comes to fossil fuels vs. renewables, environmental impact is only part of the story. Oil and gas production tends to be concentrated in countries that are antagonistic to our own, e.g. Venezuela, Iran and Russia (and thanks to bungled policy and their exploration efforts, now Brazil). It is also concentrated in the hands of despotic Middle Eastern regimes that ultimately fund and support the worst and most dangerous international terrorism, (if not directly or by the consent of the autocrats, in any case). Any energy policy that continues to emphasize oil and gas will continue to support the global markets wherein these enemies of our state and our closest allies make their living. That is the case whether we drill for more oil in ANWAR or the Gulf or not- there's simply not enough oil and gas there to alter the picture one iota. The scale of the resources we expend on these conflicts give some indication of how considerable this externality really is.

  15. And there is more. Amongst other ways in which it is wanting, the savings glut hypothesis of our persistent and destabilizing trade and current account deficits ignores the fact that we simply are not globally competitive in the industries in which our imports are concentrated, e.g. autos, textiles, toys, electronics, etc. Our annual energy imports constitute a considerable and intractable component of this deficit. So while increased energy prices would be a net negative for output, to the extent industry response results in more domestically produced energy, not least whole industries creating solar thermal, pv, wind, nuke, etc. power plants for export, we will have improved this highly ominous and destabilizing picture (one that was, for several reasons, critical to the build up of financial risk culminated in the financial crisis). Mitigation of risk, to say nothing of the wealth destroying power of the boom/bust cycle, is likewise not priced on the NYMEX. Of course, as the renewable industry goes, Germany, China and Japan continue to race further ahead of our own country, with GE doing its best to hold down the fort. That boat is setting sail.
    How's that for deeper?
    ** Speculation is a big issue here. The oil price is clearly highly sensitive to liquidity conditions, as its behavior in 2007 and 2008 respectively clearly illustrated. Of course, the real economy is as well, but this is hardly the full story- for example, the oil peaks corresponded rather well with the peak in gold in the late seventies and eighties, and I don't know of many industrial uses for that commodity. Clearly there is some symbiosis here between real price discovery, speculative investment behavior, speculative underwriting and the monetary policy backdrop, I would argue with particular emphasis on the political considerations of policy makers in the types of macroeconomic environments that give rise to price dislocations.

  16. -Majorajam-

    Many thanks for the detailed reply ... A few responses:

    1. "It's just not that simple"

    Agreed, this is very much my view. Interestingly, a slew of emails I have received from economists have told me that it is in fact, that simple. A widely respected analyst sent me an email yesterday that began "The answer is simple. When the price of oil rises, we send more money overseas. . ."

    So, my turn over simplify, it does seem that there are economists who think that the issue is simple and others who think it complex. By nature I tend toward the latter view.

    2. You do not mention US QE from the Fed, surely that has contributed to inflationary pressures. I am very much in agreement on the demand-side argument (see Pielke, Green and Wigley 2008)

    3. Thanks for sharing the ZMAW paper, this conclusion is particularly important:

    "The current state-of-the-art modeling of (endogenous) technological change still
    relies heavily on ad-hoc assumptions."

    When we debate economic models of energy or emissions, we are often not simply debating model specification and their consequences, but rather the bedrock assumptions on which they rest.


  17. Thanks for the response. Just to clarify my take:

    1) Simplicity has its place and as that goes, I have no quibble with the simplification of your economist source, certainly in the short-run (and note approvingly his not reducing the issue all the way down to the retail price of gasoline). The forest is not obscured for the trees when that's done appropriately. The problem is that sometimes people don't realize that these things are simplifications, or invest too much in them in any case, to see where contours, nuance, etc. not captured could be relevant as the circumstances and the questions change.

    2) It was probably unnecessary of me to invoke inflation at all (though clearly it is related to 'liquidity conditions' that are relevant). I did it to better illustrate the difference between demand and supply driven oil price shocks. Now that you have raised the issue though, I will say that QE is of a piece with the, (IMO catastrophic), approach to monetary policy of the Fed since Greenspan.

    It marks a departure only to those who are not aware of the historical precedence or that believe in impeccably timed coincidences whereby for 25 years agency balance sheets consistently exploded northward in the middle of liquidity crises. That being the case, neither those policies nor QE has led to any significant inflation in this country. That's a matter of record extending to this day.

    This is not to say we haven't exported inflation courtesy of Bretton Woods II. We have and continue to and, in case you're wondering, that indeed is related to the oil price 'conundrum'. As was the other famous conundrum of US bond yields. This is a big topic.

    3) Agreed. I am encouraged though when I see economists get out of their freakynomic shells and out into the real world. So when Weitzman moved the ball forward on IAMs, this was a great development. I don't know that DT survives the academic debate so far as climate change goes, and I'm not sure it should, but he has shone light in areas that desperately needed it.

    Likewise when Loschel and others begin to incorporate the endogeneity of technological change, that is a very promising development for informing policy. As that goes, I don't know that it's really that much of a surprise that WWII saw remarkable technological progress. It was all on the line. Neither do I doubt that the space program was pretty good catalyst for the same. Neither does it surprise me that my resourcefulness, perseverance and faculties are better the more skin I have in the game.

    These things are reflections of behaviors and dynamics we already know to be true. When economic models ignore these realities, as they so often do, they aren't worth the bits they are coded in.

  18. The political commotion over the recent increase in prices of petroleum products has put the PPP-led government in a very difficult situation. If it withdraws this increase, it faces the spectre of a yawning fiscal deficit and failure to meet international financial institutions’ conditionalities for loans. And if it does not, it faces a hostile political reaction, which will further weaken its already weak position after the PPP’s ouster from the coalition in Punjab. It is not just Pakistan’s government that is facing this dilemma; other countries too are calculating the risks of the gap in demand and supply of petroleum. After disruption of oil supplies from Libya, Saudi Arabia has pledged to boost production, but it will take some days before it could stabilise the market. Also, there is no guarantee that this will continue in the long-term if the situation so demands. A country like Pakistan, with the economy teetering on the brink, has few choices and our political spear heads must reckon the international factor.