Why do small supply changes move oil prices so much?
A 1% shortfall in the supply of almost anything else would be a footnote. In oil it can be a 20% price move. The reason is not sentiment or speculation — it is a property of the demand curve, and it can be written down as a division.
Both sides of the market are stubborn
Demand barely responds in the short run. If you commute by car, a 20% rise at the pump does not stop you commuting this week — you have a job, a route, and the same vehicle. Airlines have schedules sold months ahead. Petrochemical plants have contracts. Short-run price elasticity of oil demand is commonly estimated in the range of −0.02 to −0.1: a 10% price rise reduces consumption by well under 1% within a year. Over a decade, elasticity is far larger — people buy different cars, cities build transit, industry re-engineers — but a decade is not a trading horizon.
Supply barely responds either. A deepwater project takes five to eight years from sanction to first oil. Even shale, the fastest large source ever developed, needs months from decision to production and is capped by rig availability, crew availability, sand, water and pipeline capacity.
When both curves are close to vertical in the short run, a small imbalance in quantity requires a large move in price to clear it.
The arithmetic
World demand is roughly 103 million barrels a day. Lose 1 million barrels a day of supply — under 1% — and that quantity has to be rationed away somehow. With a demand elasticity of −0.05:
Required price change = 1% ÷ 0.05 = 20%.
One percent of the barrels, twenty percent of the price. Run it the other way: a 2% surplus with the same elasticity implies a 40% price fall to force enough extra consumption. That single division is the entire personality of energy markets, and it is why oil has always been a market of long quiet stretches punctuated by violence.
It also explains something that confuses newcomers: the same headline can be enormous or irrelevant depending on the buffers.
The buffers that absorb the shock
Two things stand between an imbalance and that arithmetic.
Inventories. Commercial crude and product stocks, plus government strategic reserves such as the US Strategic Petroleum Reserve, can supply barrels for a while. A market with comfortable inventories absorbs a disruption quietly. A market with thin ones has nothing between the news and the price.
Spare production capacity. Some producers can raise output within roughly 30 to 90 days — capacity that is deliberately held back. It has historically been concentrated in a very small number of countries. When spare capacity is several million barrels a day, an outage is a shrug. When it is under a million, the same outage is a crisis, because the market knows nobody can replace the missing barrels.
This is why practitioners follow spare capacity and inventory levels as closely as the headline balance. The balance says how big the gap is; the buffers say how much the price has to do about it.
The weekly ritual
In the US the EIA Weekly Petroleum Status Report lands Wednesdays at 10:30 ET with crude and product stocks, refinery utilisation and implied demand. It is the closest thing oil has to an earnings release, and it is a scheduled volatility event for the same reason the gas storage number is: it updates the buffer.
What elasticity does not tell you
Be precise about the claim. Elasticity explains why moves in energy are large. It says nothing whatsoever about direction — a tight market can be tight for a year and then not be. The same arithmetic that produces a 20% rally on a supply loss produces a 40% collapse on a surplus, and neither is predicted by the framework. What you have gained is the ability to read a disruption headline and ask the right question: how thin are the buffers right now?
Try it now
- The full crude history this page holds is below, as candles. Switch it to Weekly and look for the bars that are several times taller than their neighbours. They do not arrive evenly: they cluster into a handful of episodes separated by long quiet stretches. That clustering is what "inelastic" looks like from the outside.
- Measure across the widest of those episodes, high to low, and write down both the percentage move and the number of bars it took. Then redo the elasticity arithmetic with a 0.5 million b/d loss and an elasticity of −0.04. (0.485% of demand ÷ 0.04 ≈ 12%.) Then with an elasticity of −0.1. (≈ 5%.) The assumption about elasticity does most of the work — which is why estimates of it are argued about so fiercely.
- State in one neutral sentence why the same 1 million b/d outage can be a non-event one year and a shock the next.