Varun Sivaram

Energy, Security, and Climate

CFR experts examine the science and foreign policy surrounding climate change, energy, and nuclear security.

Print Print Cite Cite
Style: MLA APA Chicago Close


Does Expensive Oil Inevitably Cause Recession?

by Michael Levi
October 6, 2011

There is a popular belief that once U.S. petroleum expenditures exceed some threshold, recession results. Writing at the Harvard Business Review blog, Chris Nelder and Gregor MacDonald present this position clearly:


“The connection between oil shocks and recessions has been understood for decades. We have ample historical evidence that when petroleum expenditures reach 5% of GDP, recession typically follows. Annual energy expenditures rose from 6.2% of U.S. GDP in 2002 to a painful 9.8% in 2008, which was immediately followed by an economic crash. And now oil is sending energy expenditures back above 9% of GDP, just as we see fresh indications that the recession persists. This is not a coincidence.”


Some variation on this theme is a consistent feature of both peak oil writings and more moderate warnings about the economic threats posed by expensive oil. Indeed it would not be an exaggeration to say that this sort of worry motivates a large slice of energy policy thinking.


If you scratch the surface, though, claims of a threshold beyond which the economy goes into recession turn out to be pretty shaky.


Let’s start with a basic theoretical point: There is no fundamental reason to believe that 5 percent (or anything similar) should be a magic number. Petroleum expenditures were equal to 4.5, 4.5, and 4.3 percent of GDP in 1970, 1971, and 1972 respectively. Does anyone really believe that the U.S. economy would have gone into recession had that spending been half a percentage point higher?


There’s also an empirical problem: there are many years – 1976, 1977, 1978, 1979, 1983, 1984, 1985, 2006 – in which petroleum expenditures have exceeded five percent that have not coincided with recessions.


In fact the five percent claim rests on only three data points: the two 1970s recessions and the 2007-2009 one.


There is a huge literature attempting to explain all three recessions. None of them, though, tend to be chalked up to high oil spending per se. What does appear to play a large role, particularly in the 1970s cases, is a rapid increase in oil costs that temporarily overwhelms the economy’s ability to adjust. The corollary, though, is that high oil costs reached through gradual increases probably won’t do the same sort of harm.


There is, however, a possible back door explanation for why high petroleum expenditures relative to GDP  seem to correlate with recessions even if they don’t do a good job explaining them: it is easier for petroleum expenditures to undergo big changes in short periods of time if they are starting from a high level. If, say, the price of oil rises 50% from a starting point where petroleum expenditures are 2% of GDP, the change in spending is 1% of GDP; in contrast, if the price of oil rises the same 50% from a starting point where petroleum expenditures are 6% of GDP, the change in spending is 3% of GDP. Whatever your transmission mechanism – supply side contraction, demand destruction, shifts in consumer preferences for durable goods – the 3% jump is going to be far more economically damaging than the 1% one. Indeed the years where oil spending was high but recession was absent generally come from a period where prices were fairly stable.


This is a subtle but important distinction from the oft asserted five percent rule. It suggests that while rising oil costs can lead to substantial economic harm, they do not necessarily need to.  Specifically, it points to the increasing importance of blunting both price volatility and its consequences so long as the world remains in expensive oil territory. Bob McNally and I discussed the volatility problem at some length in a recent Foreign Affairs essay. The compound danger of high and volatile oil prices makes focusing on remedies to that problem all the more important.

Post a Comment 16 Comments

  • Posted by Mason Inman

    Of course you can’t draw a line and say that if you cross it, you go into recession, and if you’re below it, everything is fine. It seems you’re creating a straw man to argue against.

    I think the key point is there does seem to be a feedback between oil prices and the health of the economy. This is something that seems well established, but which many economists and policy makers do not seem to take very seriously—despite, in recent years, repeated warnings from the IEA. (Search for Fatih Birol “danger zone.”)

    [ML: Saying that there’s feedback is pretty useless unless you can say how strong that feedback is.]

    If you want to pick nits… Nelder and Macdonald said “when petroleum expenditures reach 5% of GDP, recession typically follows.” They did not say that anytime that oil spending is above 5% of GDP, then the country is in recession. So they’re pointing out that when the price is rising and crosses this threshold, it causes big problems. Of course no one is really saying that the magic number is 5% or 4.5% or 4% (the latter figure being what economist James Hamilton points to). The broader point is that as oil prices get into this range, it causes pain and the economy tends to suffer.

    [ML: I count far more years where oil > 5% but there is no recession. That’s not “typically”.]

    And of course we can adapt; nothing is inevitable. But so far, the U.S. doesn’t seem to be very good at adapting to high oil prices.

    Also, you spelled the last names of both the bloggers. They’re Nelder and Macdonald.

    [ML: Thanks. Fixed.]

  • Posted by ANDY APPAN


  • Posted by Christopher Mims

    Glad to see the the level of discussion on this subject elevated. Very interesting that it’s volatility as much as anything that matters. Foreign Affairs is paywalled — what’s the remedy, if OPEC won’t (or can’t) maintain large buffers any longer?

  • Posted by Mason Inman

    To respond to Levi’s response:

    I think you misunderstood my point about what Nelder and Macdonald are saying. I don’t think that they’re saying simply that whenever you are above a certain threshold, then the economy will be in recession. They are saying that when prices are *rising*, and you *cross* that threshold, then you tend to go into recession.

    [ML: There are three modern data points for that. Without a model, you can’t turn three data points into a rule.]

    Let’s look at what was going on during those years you pointed to, in which oil spending was above 5% of GDP, and yet there was some economic growth. I’m not an expert on this history, but here’s a simplified version of what it seems to me that happened.

    After the 1973 OPEC embargo, oil prices quadrupled and this triggered a recession. There was a lot of demand destruction—consumption went down—which took away some of the bite of high oil prices. A few years later, some benefits of efficiency and switching power plants off of oil had eased the pain of high oil prices. So that’s why there was no recession in 1976 to 1979, despite spending on oil remaining above 5% of GDP. During those years where there was some economic growth, they were not exactly great times, though. Economic growth was quite low.

    And then economy did go back into recession in 1980-1982.

    In 1983-1986, the spending on oil had been falling for a few years, since the in 1980, and it seems the economy was able to grow despite oil spending being above 5% of GDP since it benefitted from these falling prices, and also from lowering of demand due to the high prices.

    This is all much clearer when looking at a graph by James Hamilton of recessions and spending on oil as a % of GDP:

    Two big contributors to lowering oil demand during this period (1973-1986) were, if my understanding is right, the introduction of fuel efficiency standards for cars, and the switch away from oil-burning power plants. But those were the really low-lying fruits. Ending oil-fired electricity was really a one-off, and not something we can repeat now, since most oil goes for transport, and hardly any for electricity or other uses where it could be easily substituted. It seems that making similar large cuts in oil consumption quickly would be much harder to achieve today.

    So I agree that we can adjust to some degree to spending a fair chunk of our GDP on oil. But it’s not easy, and it may be a lot harder now than in the 1970s and 80s.

    I still think it is important to recognize that there is a *significant* feedback between oil prices and the health of the economy, even if it is difficult to quantify. (Quantification is overrated, I think—and my background is in physics, and I work as a science journalist.)

    Anyway, that gets back to the issue that motivated Nelder and Macdonald’s point. Yergin and others who think like him argue that rising oil prices will motivate a search for more oil (or stuff like tar sands that we can turn into oil). But they don’t seem to think about whether people will actually be able to afford that more expensive oil. That is, they ignore the feedback of oil prices onto the health of the economy.

    [ML: That just isn’t true. Those people wouldn’t be writing about oil if they didn’t think it was central to the health of the economy.]

    Yergin accuses peak oil people of ignoring economics. But it seems to me that many of them, such as Chris Skrebowski, have a much more sophisticated approach to economics than Yergin does. Skrebowski talks about global peak oil not as simply a geological phenomenon, but as a peak in the production of oil that people are willing and able to pay for.

    I think it would be a big step to simply get more people to recognize that there is a significant feedback between oil prices and the health of the economy.

  • Posted by David B. Benson

    It seems this would be a good place for a little statistics. Run the correlation cofficients to see how much of GDP is explained by petroleum prices.

    [ML: This has been a subject of active research for three decades. It’s not easy.]

  • Posted by Chris Nelder

    I generally agree with your explanation, Mason.

    A slight correction: The research I cited and linked, by David Murphy and Charles Hall, notes that the threshold is, more precisely, 5.5%. They have an updated paper (Aug 2011) here: Adjusting the economy to the new energy realities of the second half of the age of oil

    They explicitly do not attribute a simple causality to this relationship, as Levi seems to have inferred. Nor do we. The authors describe it as follows:

    Although this analysis of recessions and expansions may seem like simple economics, i.e. high prices lead to low demand and low prices lead to high demand, the exact mechanism connecting energy, economic growth, and business cycles is rather more complicated.

    Hall et al. (2009) and Murphy and Hall (2010) report that when energy prices increase, expenditures are re-allocated from areas that had previously added to GDP, mainly discretionary consumption, towards simply paying for the more expensive energy.
    In this way, higher energy prices lead to recessions by diverting money from the economy towards energy only. The data show that recessions occur when petroleum expenditures as a percent of GDP climb above a threshold of roughly 5.5%.

    Their chart here illustrates the point.

    This is straightforward empirical research, not a “popular belief.” There is nothing “shaky” about it. Further research is needed, particularly in the relatively new domain of biophysical economics. Indeed, the notion that adequate energy resources will always be available at an affordable price, regardless of geological or qualitative factors, is the popular belief which we call into question.

    The volatility angle may be interesting as well. I invite you to share with me any empirical research on that.

    [ML: You have N=3 for this analysis — it’s tough to claim that it’s a rule. That’s part of why most of the economics literature on oil prices and the macroeconomy is about changes in oil price (i.e. volatility) rather than on absolute levels. Search Google Scholar for “oil macroeconomy” and you’ll get a massive list of papers with empirical work.]

  • Posted by David L. Hagen

    Yes relative fluctuations are an issue. The more important question is why this change in relative fuel caused GDP fluctuations.

    The critical issue we address is whether we are now transitioning from cheap transport fuel based on readily available light crude oil, to expensive transport fuel from alternative fossil or renewable resources due to the plateauing and decline of light crude world wide. a.k.a. “Peak Oil”.

    The very large relative change required for this fuel transition and the inertia of inaction promise an economic “roller coaster” of systemically high and fluctuating fuel prices until a massive effort provides alternative fuels in sufficient volume to reduce relative prices.

    This challenge is popularly presented by Robert L. Hirsch et al in
    The Impending World Energy Mess
    ISBN-13: 978-1926837116;
    pdf or video at the ASPO 2011

    And in A thousand barrels a second Peter Tertzakian

    All the “green” focus on electricity is a misdirected diversion until this fuel transition is achieved.

    OPEC is already reaping a “tribute” of $ 1 trillion per year.

    For ways out, see The Plan, Edwin Black
    Turning Oil into Salt, Gal Luft and Anne Korin

    Ignoring the this reality will cause importing Non-OPEC countries to be gradually “boiled” with rapid transfer of wealth and economic sovereignty to OPEC countries.

  • Posted by Nick

    “I think it would be a big step to simply get more people to recognize that there is a significant feedback between oil prices and the health of the economy.”

    Couldn’t agree more Mason. From my research every major oil spike was followed by recession. I am not sure what the point of this article was, other than to muddy or confuse an issue which we ought to be clear on.

    [ML: And the clear, broadly agreed fact is…?]

    There’s also a very significant correlation today between the rise and fall in stock market indexes and oil prices. That indicates to me a lock-in.

    [ML: Separating correlation from causation isn’t easy.]

    Please don’t tell me higher energy prices are good for the economy. They are a by-product of growth, and tend to eventually rise at the expense of it.

  • Posted by D Murphy

    To echo the comments of Chris Nelder and Mason Inman, nowhere in my paper, or anywhere that I have seen (can you forward a ref?) has there been a “5% rule” espoused as the explanation of our recessions.

    [ML: I’m pretty sure I quoted the 5% claim in my original post, along with a link to the source.]

    Moreover, we performed a fairly simple empirical analysis that seemed to show that petroleum expenditures as a % of GDP related to the major recessions over the past 40 years.

    [ML: True. But there are only three data points for that claim.]

    That said, this hypothesis, much like all other hypotheses, is not “truth” and we would be remiss to think of it as a “rule” – which is the primary reason why we were careful in our words, saying that expenditures above 5.5% “tend” towards recession.
    On the other hand, there is some contradiction in the post by Dr. Levi. He states that:
    “There is a huge literature attempting to explain all three recessions. None of them, though, tend to be chalked up to high oil spending per se.”
    He is correct in stating that there are myriad papers discussing the relation between recessions and oil prices. But much of it uses oil price specifically as one of the main factors causing recessions. The work of James Hamilton ( comes to mind (as Mason indicates).

    [ML: Hamilton’s work, as with that of others, looks at *changes* in oil spending, not high oil spending in itself. That was my point. There is nothing — *nothing* — in the Hamilton stuff that claims to show consequences of high oil prices independent of whether those prices are changing.]

    There may be some differentiation in Dr. Levi’s use of “oil spending” and what he later refers to as “oil costs,” and in a strict economic sense there certainly is a difference. But as we are concerned, high oil prices have high oil costs, which is to say that to purchase the “high cost” oil will require high oil spending – i.e. they are all quite similar.
    There is clearly an asymmetry in demand responses to changes in oil prices depending upon whether the price is increasing, increasing to an all time high, or decreasing. It is also clear that rapid, large increases in the price of oil, a so-called oil price shock, have negative impacts on economic growth. 1973, 1979, and 2007-2008 are evidence of this.
    But there has been a fundamental shift in the oil-economy relationship in the past 5-10 years. Unlike ’73 and ’79, when oil companies around the world were able to find new and cheap sources of conventional oil, today, we have only high priced resources remaining (think tar sands). As Nelder and MacDonald indicate in their piece at HBR, numerous oil companies, governmental organizations, and independent energy agencies have acknowledged the coming age of expensive oil. The average real oil price during expansionary periods over the past 40 years was $37 per bbl, while it was $58 per bbl during recessions. The price-point for much of the new discoveries is above even the recessionary price during the past 40 years. The question then becomes, can the economy grow with sustained high oil prices?
    Which leads me to Dr. Levi’s comment that:
    “What does appear to play a large role, particularly in the 1970s cases, is a rapid increase in oil costs that temporarily overwhelms the economy’s ability to adjust. The corollary, though, is that high oil costs reached through gradual increases probably won’t do the same sort of harm.”

    The price of oil has increased gradually since the initial drop after the crash in 2008.

    [ML: An increase of nearly $100/bbl over 2-3 years isn’t gradual.]

    If Dr. Levi is correct, this should have little impact on economic growth. In the long run I would probably agree; consistently high oil prices will allow for substitution that will eventually decrease the dependence of our economy on oil and allow for growth despite persistently high prices. But I emphasize – long run – which is probably on the order of decades. But, here we are, two years after the last recession officially ended, oil prices are high, and have been high for some time, and the entire world thinks we are heading back to recession. Certainly oil prices are not the only cause of our current economic woes, but if we do go into a recession, at least we will have data point number 4.

  • Posted by D Murphy

    Re: [ML: An increase of nearly $100/bbl over 2-3 years isn’t gradual.]

    Ok. I would agree, but it was not as quick of an increase as 2007-2008, ’79, or ’73. What would constitute a gradual change in oil prices? Has there ever been one?

    Re: [ML: Hamilton’s work, as with that of others, looks at *changes* in oil spending, not high oil spending in itself. That was my point. There is nothing — *nothing* — in the Hamilton stuff that claims to show consequences of high oil prices independent of whether those prices are changing.]

    I think is just another way of saying long vs short term. In the short term high oil prices are detrimental to economic growth, but over the long term the economy is expected to adjust.

    So I guess the question is this: do you think the economy can grow with a stable oil price of $100?

    I would argue that it depends on how quickly our economy can adjust, and when it comes to oil, I think this will take some time. In the near and mid-term (1-3 yrs and 3-10 yrs ish, respectively) I would expect to see very slow growth or even recessionary trends. I would also argue that we are trying to do that right now and it is failing.


  • Posted by P. Oil

    Please see BP’s inflation-adjusted oil price data going back to the 1860s-70s, take the 3-, 5-, and 10-year avg. prices, and then regress against Geary-Khamis real US GDP per capita for 5 and 10 years.

    You will discover that US 10-yr. avg. real GDP per capita is set to contract further and remain negative for the decade, if not longer. The additional implication is that gov’t spending must contract at least 30%, and 50% or more in real terms per capita over 10 years or more.

    Moreover, debt-money assets will contract 50% (more in real terms) over the course of the decade.

    Further, on the same basis, global real GDP per capita peaked with Peak Oil in ’05-’08 and will turn negative over the course of the decade.

    Conclusion: The Oil Age growth era is over, and contraction is inevitable.

  • Posted by Roderick Beck


    In economic prices are not exogenous. They are endogenous and are determined by supply and demand. Strictly speaking it is meaningless to say high oil prices caused X to happen. You have to know what caused the change in oil prices – how supply and demand changed. An increase in oil prices could be due to greater demand or falling supply. Consequences in each scenario are different. r

    Oil prices and productioin are procyclical – that suggests that the demand for oil is the key driver over the business cycle.

    Clearly the 197/74 recession was caused by the declining supply of oil relative to demand. A classic commodity price shock.

    However, it is downright silly to think the 2008/2009 recession was caused by high oil prices. It was the bursting of an asset bubble (real estate)} that caused it. Similarly, the 1982 recession was not due to rising prices, but a double digit Federal funds rate.

    There is no plausible monolithic explanation for the business cycle.

    Finally, to return to my first point. In a traditional model where aggregate demand fluctuations generate a business cycle, oil prices will rise during expansion and decline during the recession. In other words, the economy drives oil prices and not oil prices the economy.

  • Posted by Nichol

    Should there not be some basic economic calculus to treat such questions? When is it ‘volatility’, ie a quick shock .. and when is it a slow ramp-up that we can addapt to? That, naturally, depends on the time-scales involved in the adaptations we can make.

    It is clear that if we can increase efficiency, we will be able to afford a higher oil price, since what is important is how much we spend on energy, in total, as percentage of GDP. If that becomes too high, it is like a high tax, that we pay to a foreign country (so no local jobs, or economic stimulus). There should be some economic knowledge on how much of our GDP we can give away without too much trouble.

    On much longer time-scales than replacing inefficient vehicles, we can also adapt our usage pattern, by travelling less, and living closer to our work. But that demands a total redesign of cities and transport systems.

    Demand for oil is quite inelastic, precisely because the low-hanging fruit for improving efficiency have already been picked.

    And now oil use in developing countries is becoming significant, spread over much higher numbers of small users. Little diesel-generators, motor bikes and minibusses. Those are all essential to development, not really optional, and as such that demand is even less elastic than in the developed world. We can only hope that solar panels become the new alternative to diesel generators, all over the developing world, very soon.

  • Posted by Steven Kopits

    Here’s a paper on both the 4% rule and oil price volatility. (

    In 2007, the US went into recession when oil expenditures were 4.6% of GDP. If you want to take RAC as the appropriate measure, then US crude oil consumption ranged 5.0-5.3% of GDP from March to July 2011. It was 4.6% in September.

    If we use regular gasoline prices as a proxy (projected over total crude oil consumption), then the US fell into recession when the gasoline share reached 6.9% of GDP. This number ranged 7.2-7.6% of GDP from March through August 2011, and was 7.0% in September. (I use a gasoline index due to the question of how important Brent prices are versus WTI or RAC. Gasoline prices implicitly include the influence of Brent on US retail prices. Obviously, not all crude oil is consumed as gasoline.)

    By any of these measures, the US would have to be considered at risk of recession based on historical oil prices.

    While volatility might play a part, I could never make the statistics look right. In the paper, for example, you’ll see that using year-over-year price changes causes the impact to occur too late in economic cycle to be convincing as a cause. It didn’t work with three month prices changes either. And it does not explain the 1981 recession when prices were falling but still well above the 4% of GDP threshold.

    In the end, I was left with the GDP threshold as the more convincing explanation. Our threshold test suggested, in April of this year, a recession by the end of summer, and we advised our clients to be cautious with valuations and capital structuring after that time. (Good advice, as it turned out).

    ECRI and Soros say recession; Hamilton and Dueker say only slowdown. Hamilton’s analysis relies, in part on the “dead already” theory, that the economy is so far down in many sectors (housing, vehicles) that it will resist further depression. Further, other than government deficits, the economy has no visible imbalances. Neither stocks nor housing are over-valued. Thus, the reason for an adjustment, other than oil prices and deficits, is missing. So that could also play a part.

    We now have to wait for the other shoe to drop. By year end, we should know who was right. Those who have guessed right will feel vindicated; the others will go off to modify their models.

  • Posted by Freude Bud

    Three strikes and you’re out in my book.

    I’d guess that as disposable income shrinks (from ’72 on) for the vast majority of Americans that the threshold for demand destruction and tipping the country into general economic malaise is only getting smaller as well, for what it’s worth.

    (It would be a little silly to look at BP’s price series going back to the 1860s as suggested by P. Oil–it’s not a single crude for which prices are given and the end uses change quite dramatically over that period of time, that is, the notion that a severe shortage of oil in 1880 would have potentially made the entire economy collapse like it would today is, well, difficult to reconcile with the facts as we know them.)

  • Posted by steve from virginia

    The usual implication of oil price-recession correlation assumes a cash ‘exchange’ economy where fuel costs manifest themselves at the gas pump.

    The reality is more nuanced. During the period 2004-2007 the Federal Reserve allowed the funds rate to rise from 1% to 5.5% stipulating “inflation caused by rising fuel prices” as justification (by way of FOMC minutes).

    The question then becomes, ‘why did interest rates rise (why did lenders cease lending at low rates)?’ The answer then is the lack of return on investments including those issued by the government because of inflation risk … along with other reasons.

    One of those ‘other reasons’ was the improbability of many investments to earn under any set of circumstances, that is these enterprises were failing in a near-zero interest rate environment.

    It’s impossible to easily separate fuel price risk from other forms when all are linked in some way by credit.

    A more general way to look at risk is to see when fuel use infrastructure was put into service under what assumptions. Assuming +$100 fuel would never permit sprawl over millions of square miles, tens of thousands of highway lane miles along with millions of massive fuel guzzlers. Even with subsidy, these would be too costly … and are proving too costly right now!

    With the higher priced fuel all of the infrastructure created assuming low cost fuel is underwater. This in itself becomes a charge against GDP.

    Quantifying fuel costs isn’t hard if you decide to look a bit outside the box.

    And people wonder why there is a crisis?

Post a Comment

CFR seeks to foster civil and informed discussion of foreign policy issues. Opinions expressed on CFR blogs are solely those of the author or commenter, not of CFR, which takes no institutional positions. All comments must abide by CFR's guidelines and will be moderated prior to posting.

* Required