Seventy years of oil shocks
11 episodes in 72 years, measured by the barrels that left the market rather than by what the price did. On that measure the largest oil shock ever recorded is one that no chart of oil prices has ever shown, and the thing that separates the episodes that did lasting damage from the ones that did not is neither the size of the shock nor the size of the price move.
Measure an oil shock by its price and you cannot see the largest one ever recorded, because in 1956 there was no price to see it with.
- 1 The biggest supply shock in the record produced no price signal at all. In 1956 the Suez closure took 1.7 million barrels a day off the market — 10.1% of world crude production, the largest share ever lost. The posted price did not move, because it was published by the oil majors rather than discovered in a market. Any study that screens on price change is blind to this entire era.
- 2 Shock size does not predict the equity outcome. Among the five largest physical shocks — all between 7.8% and 10.1% of world supply — the associated S&P 500 decline ranges from 48% at one extreme to none at all at the other. 2026 is the second-largest supply loss ever measured and equities reached a record high in the same year.
- 3 What does line up is whether anyone could respond. 4 of the 11 episodes landed on a world with no coordinated stocks and no spare capacity outside the producers doing the cutting. Those are the only ones in which the oil price repriced and stayed repriced. In all 7 episodes with a buffer, the price round-tripped within months.
- 4 And the same barrel costs less than it used to. Oil per real dollar of US output is down between 50% and 70% since 1973, energy has gone from about 28% of the S&P 500 to 3.5%, and in real terms 2026’s high is below the annual high of 13 earlier years.
Why price is the wrong unit
The conventional way to study this question is to screen on the oil price — crude rose some percentage from a recent low — and then ask what equities did next. The screen is easy to run and it fails in two separate ways, both of which matter more than the result it produces.
The first failure is historical. For the first twenty years of this record there was no oil price in the sense the screen assumes. Crude was sold at a price the major oil companies set and published, and that price moved when they or later OPEC decided it should, not when supply and demand changed. It sat at $1.80 a barrel, unchanged, from 1961 to 1970. The suez closure in 1956 is the proof of what that hides: the largest share of world production ever lost in a single month, and a price series that recorded nothing whatsoever.
The second failure is that a price screen cannot tell apart two events with opposite meanings. Oil rising because a war has closed a strait and oil rising because the world economy is growing faster than the oil industry can supply it are not the same event, and they do not have the same consequences for equities. Pooling them is what produces the mush that studies of this question usually arrive at.
So the unit here is the physical shock: barrels lost, as a share of contemporaneous world supply. That is the measure Hamilton uses, it is available back to the 1950s, and it is indifferent to whether a price existed to report it.
Ranked by what actually left the market
Put the episodes in order of the share of world supply they removed and the ranking is not the one the price history suggests. The top five span 7.8% to 10.1% — close enough together that they are, as physical events, much the same size.
Two entries deserve their placing explained. 2026 is the largest disruption ever recorded in absolute barrels, at 10.1 million a day, but the world now consumes roughly 107 million barrels a day against 56 million in 1973 — so as a share it is second, not first. And 2004 appears with a supply loss of zero, because nothing was lost: Brent went from $30 to $140 without a single barrel being taken by war, embargo or strike.
And the equity outcome does not follow from it
Plot the physical shock against what the S&P 500 did and there is no relationship to find. The two largest shocks in the record sit at opposite ends of the equity axis. The worst equity outcome of the whole set belongs to an episode that ranks fifth by barrels. The episode with the largest price move of all — 2003–08 — sits on the zero line of the supply axis.
This is the point at which a study that stopped here would conclude that oil shocks and equity markets have little to do with one another, and leave it. The more useful reading is that the scatter is telling you the axes are wrong — that something other than the size of the shock is doing the work.
Four kinds of oil shock
Kilian separates an oil price move into its causes, and Kilian and Park show that the causes do not act alike on equities. A price rise driven by global expansion lifts stock prices. One driven by oil-specific precautionary demand pushes them down. A physical supply disruption has the weakest effect of the three.
That ordering is the opposite of the folk model, in which oil going up is simply bad for shares. It is also what the record here shows once the episodes are sorted by kind rather than by size. The classification below is this site’s reading of each episode, not Kilian’s — he decomposes a time series, he does not label events.
Supply disruption
Barrels physically removed from the market by war, embargo, strike or blockade.
1956 · 1973 · 1979 · 1980 · 2002 · 2011 · 2026
Precautionary demand
Buyers paying for barrels they may not need, against supply they may not get. The fear premium drains as fast as it arrives.
1990 · 2022
Aggregate demand
Oil rising because the world economy is growing into a supply ceiling. A symptom of strength, not a tax on it.
2004
Demand collapse
The mirror image: consumption falls away and the price goes with it.
2020
The two episodes that did the lasting damage — 1973–74 and 1978–79 — are both supply disruptions. But so are 2011 and 2026, which did none. Kind narrows the question; it does not settle it. What settles it is the next section.
The variable that actually sorts them
Sort the 11 episodes by whether a coordinated response existed when they landed, and the record divides cleanly in a way that none of the other cuts manage.
4 episodes — 1956, 1973, 1979, 1980 — met a world with nothing to draw on. There was no International Energy Agency before November 1974, and there was no American strategic reserve worth the name before the late 1970s. Both institutions exist because of the 1973 embargo; the IEA was founded in direct response to it. In those episodes the price repriced and stayed repriced: after 1979 the annual high did not fall back below $30 until 1986.
In the 7 episodes since, the buffer has been used six times and the price has round-tripped every time. The clearest pair is 1978–79 against 1990–91: 8.9% of world supply lost versus 8.8% — the same shock, within a rounding error — and opposite price outcomes. What changed in between was sixteen years of institution-building, and on 17 January 1991 the IEA used it for the first time.
The machinery is also symmetric, which is easy to miss. Against a supply shock the tool is a release. Against a demand shock it is a cut — and the largest single intervention in the history of the oil market was 2020’s OPEC+ reduction of 9.7 mb/d, larger than the supply lost in 1973, 1979, 1980 or 1990.
The episodes
Each panel carries the physical shock, what the price did, what the equity market did and when it turned, what response existed, and what the episode is actually evidence of. Where the equity decline began before the oil event, the panel says so — which is why no “median lead time” is quoted anywhere on this page. In several of these cases the lead is negative.
Suez, and the shock the price could not report
October 1956 to March 1957 · recession August 1957 to April 1958
Egypt nationalised the Suez Canal in July 1956. Britain, France and Israel attacked in October, the canal was blocked by scuttled ships, and the Syrian pumping stations on the IPC line were sabotaged. Middle East production fell 1.7 million barrels a day between October and November — 10.1% of world crude output, the largest share ever lost in a single month.
The posted price did not move. Crude was sold at a price the major oil companies set and published, not at a price a market discovered, and that price stayed at $1.93 a barrel through the entire episode. Europe rationed petrol; the quoted price of oil recorded nothing at all.
This is why a study that screens on price change cannot see the first twenty years of this record. It is also why the largest physical oil shock in history is absent from every chart of oil prices ever drawn.
The 21.6% equity decline in the panel is dated from 2 August 1956, two months before the canal closed and days after Egypt announced the nationalisation. It is conventionally read as a monetary episode — the Federal Reserve was tightening into an investment boom — and the recession NBER dates from August 1957 began ten months after the oil came back. The decline is shown here because it is the nearest one, not because Suez caused it.
The largest supply loss on record produced no price signal, because in 1956 there was no price to signal with.
- Supply lost
- 1.7 mb/d
- Share of world supply
- 10.1%
- Oil price
- Unchanged
- S&P 500
- -21.6%
Response capacity
- US majors reroute Gulf crude around the Cape
- Texas Railroad Commission lifts proration to raise US output
Sources
The Arab embargo meets an unbuffered world
October 1973 to March 1974 · recession November 1973 to March 1975
OPEC raised the posted price 70% on 16 October 1973 and the Arab members cut output and embargoed the United States and the Netherlands the next day. Gross supply off the market peaked near 4.3 million barrels a day, 7.8% of world production. The posted price went from $2.90 in September to $11.65 by January.
It never went back. Of the nine episodes here, 1973 and the Iranian revolution are the only two where the price repriced and stayed repriced — and they are the only two that landed on a world with no coordinated stocks to draw on and no spare capacity outside the cartel doing the cutting.
The equity bear market had begun nine months earlier, in January 1973, with the Fed above 10% and inflation already rising. Oil did not start it. Oil is why it became the worst since 1929.
The first of two episodes to meet a world with no buffer — and one of only two where the price never came back.
- Supply lost
- 4.3 mb/d
- Share of world supply
- 7.8%
- Oil price
- 302%
- S&P 500
- -48.2%
Response capacity
- US price controls and retail allocation
- IEA founded, November 1974 — after the fact
Sources
Revolution in Iran
November 1978 to April 1979 · recession January to July 1980
Strikes in the Iranian oil fields from late 1978 and then the revolution took Iranian output down by nearly 90% by January 1979. The gross peak loss was 5.6 million barrels a day, 8.9% of world supply — larger in share terms than 1973. Panic buying did the rest: refiners and governments bid for cargoes they did not need, against a shortfall that was real but smaller than the price implied.
The second unbuffered episode, and the second permanent repricing. The annual high stayed above $30 every year from 1980 to 1985, against $13.66 in 1978.
Equities took six weeks to fall 17% in early 1980 — the decline in the panel — and had recovered the whole of it by July. The damage came later and is filed under the next episode, because what did it was the policy response to the inflation rather than the oil itself.
A bigger share of world supply than 1973, and the same outcome, for the same reason: nothing to draw on.
- Supply lost
- 5.6 mb/d
- Share of world supply
- 8.9%
- Oil price
- 166%
- S&P 500
- -17.1%
Response capacity
- IEA collective action
- US pledge to cut consumption by up to 5%
Sources
- Hamilton, "What is an Oil Shock?", Table 4 — exogenous disruptions in world petroleum supply, dated by the month of the largest observed drop
- IEA, Energy Supply Security 2014 — gross peak supply losses by episode
- US Energy Information Administration — historical supply disruptions
- Federal Reserve History — the oil shocks of 1973–74 and 1978–79
The Iran–Iraq war, and then Volcker
October 1980 to January 1981 · recession July 1981 to November 1982
Iraq invaded Iran in September 1980 and both countries’ exports collapsed: 4.1 million barrels a day off the market, 7.2% of world supply. Saudi Arabia pushed its own output to roughly 10 million barrels a day to fill the gap, which is the first time in this record that spare capacity, rather than a stock release, did the absorbing.
The equity bear market that followed — 27% from November 1980 to August 1982 — is the hardest in this set to lay at oil’s door. Paul Volcker had taken the funds rate close to 20% to break a 13% inflation rate, and short rates ran six points above long rates. Oil was the inflation that provoked the policy, and the policy was what broke the market.
This episode is kept separate from 1979 deliberately. Treating 1979-to-1982 as one chain credits oil with a bear market that Volcker caused.
- Supply lost
- 4.1 mb/d
- Share of world supply
- 7.2%
- Oil price
- 27.1%
- S&P 500
- -27.1%
Response capacity
- Saudi Arabia raises output to about 10 mb/d
Sources
Iraq invades Kuwait — and the buffer is used for the first time
August 1990 to February 1991 · recession July 1990 to March 1991
Iraq took Kuwait on 2 August 1990 and 4.3 million barrels a day of combined Iraqi and Kuwaiti output left the market — 8.8% of world supply, within a rounding error of 1979. Brent more than doubled in eight weeks.
Then, on 17 January 1991, the IEA took the first collective action in its history: a plan to put 2.5 million barrels a day into the market, with the US selling 17.3 million barrels from the reserve. The air war began the same day. Brent fell by a third in a single session and was back under $20 inside four months.
The equity drawdown — 19.9%, just short of a bear market — had started two weeks before the invasion, in a recession that NBER dates from July and a credit squeeze that had nothing to do with oil.
Same physical shock as 1979, opposite price outcome. The difference was 16 years of institution-building.
- Supply lost
- 4.3 mb/d
- Share of world supply
- 8.8%
- Oil price
- 108%
- S&P 500
- -19.9%
Response capacity
- IEA collective action — first in its history, 17 January 1991 — 2.5 mb/d plan
- US Strategic Petroleum Reserve sale — 17.3 m bbl
- Saudi and other OPEC spare capacity — about 3 mb/d
Sources
- Hamilton, "What is an Oil Shock?", Table 4 — exogenous disruptions in world petroleum supply, dated by the month of the largest observed drop
- IEA, Energy Supply Security 2014 — gross peak supply losses by episode
- IEA — history of collective actions (1991, 2005, 2011, 2022 × 2, 2026)
- US Department of Energy — history of SPR releases (1991: 17.3 m bbl; 2005: 11.0; 2011: 30.6; 2022: 180.0)
The Venezuelan strike
December 2002 to February 2003
A general strike against Hugo Chávez shut roughly two-thirds of Venezuela’s 3 million barrels a day from December 2002, overlapping the invasion of Iraq in March 2003. Brent rose about a third.
The episode is here as a control. It is a real supply disruption of real size, and the equity bear market that was running at the time — the dot-com bust — bottomed in October 2002, during it. Nothing about the oil shock marked either the low or the recovery.
A genuine supply loss that coincided with the END of a bear market.
- Supply lost
- 2.1 mb/d
- Share of world supply
- 2.7%
- Oil price
- 31.6%
- S&P 500
- No decline of 15%
Response capacity
- OPEC quota increase — 1.5 mb/d
Sources
China, and a price spike with no outage at all
January 2004 to July 2008 · recession December 2007 to June 2009
Between 2003 and 2008 Brent went from $30 to $140 without a single barrel being lost to war, embargo or strike. Chinese and emerging-market demand grew into a supply ceiling, OPEC’s spare capacity fell to around a million barrels a day, and the price did what a price does when supply cannot answer.
Equities rose with it. The S&P 500 gained more than 80% from the 2002 low to October 2007 while oil quadrupled — which is exactly what Kilian and Park predict for an oil rise driven by global expansion, and the opposite of the folk model.
The crash that followed was a credit crash. Oil peaked in July 2008, nine months after equities had already turned.
The largest price move in the whole record, and the supply loss was zero.
- Supply lost
- None
- Share of world supply
- 0%
- Oil price
- 363%
- S&P 500
- -56.8%
Response capacity
- No collective action — no supply was lost
Sources
Libya, and a release that worked
February to October 2011
Libya’s roughly 1.6 million barrels a day went almost entirely offline in the civil war. Brent rose about a third. The IEA released 60 million barrels in June, half of it from the US reserve, and the price was back near $100 by October.
Equities fell 19.4% over the same months, which had to do with the euro crisis and the US credit downgrade in August. The oil release and the equity low are three months apart and unrelated.
A small shock, a proportionate release, and a full price round trip.
- Supply lost
- 1.6 mb/d
- Share of world supply
- 1.8%
- Oil price
- 32.9%
- S&P 500
- -19.4%
Response capacity
- IEA collective action, 23 June 2011 — 60 m bbl
- US Strategic Petroleum Reserve sale — 30.6 m bbl
Sources
Demand collapse, and the largest production cut ever made
February to June 2020 · recession February to April 2020
The pandemic took roughly 20 million barrels a day of consumption out of the market in April 2020, and a price war between Saudi Arabia and Russia flooded it at the same moment. Brent fell 72% from January to the April low. The expiring May WTI contract settled at minus $37.63 on 20 April, because there was nowhere left to put the oil.
OPEC+ answered on 12 April with a cut of 9.7 million barrels a day from 1 May — the largest coordinated production cut ever agreed, more than twice the 4.2 million of 2008, and larger than the supply lost in 1973, 1979, 1980 or 1990. The IEA put the effective cut at 10.7 million barrels a day.
Equities fell 33.9% in 33 days, the fastest bear market on record, while oil fell with them. The folk model has oil and stocks moving against each other. Here they moved together, down, because the same event caused both.
The response machinery is symmetric. The largest ever intervention in the oil market was a cut, not a release.
- Supply lost
- None
- Share of world supply
- 0%
- Oil price
- -71.8%
- S&P 500
- -33.9%
Response capacity
- OPEC+ production cut from 1 May 2020 — largest in history — 9.7 mb/d
- IEA assessment of the effective cut, against high April output — 10.7 mb/d
- Tapered to 7.7 mb/d for H2 2020, then 5.8 mb/d to April 2022
Sources
Ukraine: a fear premium on a one-percent shortfall
February to October 2022
Russia invaded Ukraine on 24 February 2022 and Brent rose a third in two weeks. The barrels actually lost were about a million a day by April — close to 1% of world supply, the smallest physical shock of any episode here that moved the price this far. What was being priced was the three million a day the market feared might go.
The response was the largest to that date: 182.7 million barrels across two IEA collective actions, including 180 million from the US reserve, the biggest American release ever made. Brent was back under $80 by December.
Equities fell 25.4% — but from a peak on 3 January, seven weeks BEFORE the invasion, through a year in which the Fed raised rates faster than at any time since Volcker. There was no recession.
A 1% supply loss and a 34% price move. What moved was the fear premium, and fear premia drain.
- Supply lost
- 1.0 mb/d
- Share of world supply
- 1.0%
- Oil price
- 34.1%
- S&P 500
- -25.4%
Response capacity
- IEA collective actions, March and April 2022 — 182.7 m bbl
- US Strategic Petroleum Reserve — largest release in its history — 180 m bbl
Sources
Hormuz: the largest disruption in history
From 28 February 2026
Strikes on Iran from 28 February closed most tanker traffic through the Strait of Hormuz, which normally carries over 20 million barrels a day. World supply fell 10.1 million barrels a day in March — the largest absolute disruption ever recorded, and about 9.4% of world supply, second only to Suez as a share.
Brent rose roughly 51% in March on the narrowest reading and about 65% on the World Bank’s, its largest monthly gain since the series began. The IEA put 400 million barrels on the market from 11 March, its largest action ever, and Brent fell back to about $73 by June before climbing again as attacks on shipping resumed.
Equities dipped about 8% in March and then went on to records. In real terms the March peak was below 1980, 1981, 2008, 2011 and 2012 — the largest physical shock in the record produced a real price lower than five previous ones.
The largest supply loss ever measured, and the smallest equity response of any episode in this set.
- Supply lost
- 10.1 mb/d
- Share of world supply
- 9.4%
- Oil price
- 65.9%
- S&P 500
- No decline of 15%
Response capacity
- IEA collective action, 11 March 2026 — largest ever — 400 m bbl
- Member contributions as confirmed — 426 m bbl
- US share — 172.2 m bbl
- Released by early October — about 325 m bbl
- Further acceleration agreed, October 2026 — about 100 m bbl
Sources
- World Bank — Strait of Hormuz disruption, March 2026: 10.1 mb/d, "largest oil market disruption in history"
- IEA, "Member countries to carry out largest ever oil stock release", 11 March 2026
- US Department of Energy — history of SPR releases (1991: 17.3 m bbl; 2005: 11.0; 2011: 30.6; 2022: 180.0)
- CNBC — timeline of the 2026 Iran war and oil prices
The same barrel, less damage
One more thing has changed, and it is why a 9.4% supply loss in 2026 does not read like a 7.8% loss in 1973. Barrels of oil per real dollar of US output have fallen by somewhere between a half and seven-tenths since 1973. Oil is under 2% of consumer spending.
Oil burden is the oil bill as a share of GDP. Every completed shock whose burden passed about 4% of GDP was followed by a US recession within eighteen months. The 2026 shock peaked near 2.6%. The useful version of “how big is this shock” is therefore not the price and not even the barrels, but the oil bill as a share of output — and on that measure the largest physical disruption ever recorded came in at about two-thirds of the level that has historically preceded a recession.
The equity market has moved the same way for a mechanical reason. Energy was roughly 28% of the S&P 500 in 1980. It is about 3.5% now. An oil shock moves the index less than it used to partly because the index is barely in the oil business.
What this does not establish
11 episodes in 72 years is a case series, not a sample. Nothing on this page is a test of significance and no probability is quoted, because with eleven observations and four categories any conditional rate you could compute would be a description of these eleven events rather than an estimate of anything.
Three specific limits are worth stating plainly. The pre-1984 price is an administered posted price, so every price change in that era is a decision, not a market outcome, and the two are not comparable. Supply-loss estimates differ between the IEA, the EIA, Hamilton and the CIA by factors approaching three, and the choice between them is a choice of definition rather than of accuracy. And the classification of each episode by kind is a judgement; a reader who reclassifies 1990 or 2022 as pure supply shocks rather than precautionary ones would weaken the argument of section three without touching section four.
The buffer finding is the one that survives the most reclassification, because it does not depend on the shock taxonomy at all. Whether the IEA existed on a given date is a matter of record.
Every figure on this page
One row per episode, with the measurement convention for each column named in the header. This is the table the charts are drawn from.
| Episode | Kind | Supply lost mb/d |
World supply mb/d |
Share of world |
Oil move |
S&P 500 decline |
Buffer | Release m bbl |
|---|---|---|---|---|---|---|---|---|
| 1956 · the Suez closure | Supply | 1.7 | 17 | 10.1%H | — | -21.6% | none | — |
| 1973 · the Arab embargo | Supply | 4.3 | 56 | 7.8%H | 302% | -48.2% | none | — |
| 1979 · the Iranian revolution | Supply | 5.6 | 63 | 8.9%H | 166% | -17.1% | nominal | — |
| 1980 · the Iran–Iraq war | Supply | 4.1 | 60 | 7.2%H | 27.1% | -27.1% | nominal | — |
| 1990 · the invasion of Kuwait | Precaution | 4.3 | 66 | 8.8%H | 108% | -19.9% | coordinated | 17 |
| 2002 · the Venezuelan strike | Supply | 2.1 | 77 | 2.7%c | 31.6% | — | coordinated | — |
| 2004 · the China demand spike | Demand | 0 | 86 | 0% | 363% | -56.8% | coordinated | — |
| 2011 · the Libyan civil war | Supply | 1.6 | 89 | 1.8%c | 32.9% | -19.4% | coordinated | 60 |
| 2020 · the pandemic demand collapse | Collapse | 0 | 94 | 0% | -71.8% | -33.9% | coordinated | — |
| 2022 · the invasion of Ukraine | Precaution | 1.0 | 99 | 1.0%c | 34.1% | -25.4% | coordinated | 183 |
| 2026 · the Hormuz closure | Supply | 10.1 | 107 | 9.4%c | 65.9% | — | coordinated | 426 |
Sources
Where two of these disagree about the same event, both figures are given in the text rather than averaged.
- Hamilton, "What is an Oil Shock?", Table 4 — exogenous disruptions in world petroleum supply, dated by the month of the largest observed drop
- Hamilton, "Causes and Consequences of the Oil Shock of 2007–08"
- Kilian (2009), "Not All Oil Price Shocks Are Alike"; Kilian and Park (2009) on the differing effect of each shock type on stock returns
- IEA, Energy Supply Security 2014 — gross peak supply losses by episode
- IEA — history of collective actions (1991, 2005, 2011, 2022 × 2, 2026)
- IEA, "Governing Board concludes 2022 collective actions" — 182.7 m bbl
- IEA, "Member countries to carry out largest ever oil stock release", 11 March 2026
- IEA, Oil Market Report, April 2020 — effective OPEC+ cut of 10.7 mb/d
- OPEC+ agreement of 12 April 2020 — 9.7 mb/d from 1 May, tapering to 7.7 then 5.8
- US Department of Energy — history of SPR releases (1991: 17.3 m bbl; 2005: 11.0; 2011: 30.6; 2022: 180.0)
- US Energy Information Administration — historical supply disruptions
- US Energy Information Administration — Brent and WTI spot prices
- Federal Reserve History — the oil shocks of 1973–74 and 1978–79
- World Bank — Strait of Hormuz disruption, March 2026: 10.1 mb/d, "largest oil market disruption in history"
- CNBC — timeline of the 2026 Iran war and oil prices
- CME Group — energy’s weight in the S&P 500, 1980 to date
- Oil burden as a share of US GDP, and the 4%-of-GDP recession threshold
- Bessemer Trust — the fall in US oil intensity of GDP since 1973
Follow the research
New valuations and research notes, a few times a year. No promotional email, no sharing of your address, unsubscribe in one click.
Delivered by Buttondown. You can also subscribe by RSS.