Everybody notices it. Oil jumps and the pump price moves the same week; oil slumps and the pump takes a month to notice. Drivers assume this is a story they tell themselves out of irritation. It is not. Economists named it decades ago: rockets and feathers, and it has been measured repeatedly, on real data, in more than one country. The more interesting question is not whether it happens, but why it survives even where every price is published in real time by law.
Short answers
Why do fuel prices go up fast and come down slowly?
The pattern is called rockets and feathers. In South Korea, pump prices passed on 58% of a rise in spot prices within four weeks, but only 38% of a fall. Researchers point to tacit coordination between stations and to drivers not checking: a station has little reason to cut quickly when its customers refuel by habit.
The size of the effect
A study published in Sustainability in February 2023 examined South Korea using eleven years of weekly data, from February 2009 to May 2020, tracking prices through three stages: spot to wholesale, wholesale to retail, and spot to retail.
Its retail finding is the clearest number available on this. When spot prices rose by one won per litre, retail prices captured 58% of the increase within four weeks. When spot prices fell by the same amount, retail passed on only 38% of the decrease over the same four weeks.
Given long enough, the gap mostly closes: after twenty weeks, 92% of increases had reached the pump against 87% of decreases. So the money is not kept forever. It is kept for a while, and "a while" here means several months, during which every tank a driver buys is priced on the way down more slowly than it was priced on the way up.
Why the delay costs real money even though prices do eventually fall: a driver does not buy fuel once at the end of a cycle. They buy it every week or two, throughout. A four-week lag on the way down, repeated through every downward move over a year, is not a rounding error that cancels out. It is a persistent tilt in who captures the benefit of falling crude.
The experiment Germany accidentally ran
The obvious remedy is transparency. If everyone can see every price, surely a retailer holding its price high while its costs fall would be undercut immediately.
Germany has been running exactly that experiment for over a decade. The Market Transparency Unit for Fuels, part of the Bundeskartellamt, began operating on 31 August 2013. Every company operating a public filling station, or controlling its prices, must report price changes in real time, covering Super E5, Super E10 and diesel. The unit does not publish to the public directly; it passes the data to registered consumer information providers, who feed the apps and navigation systems drivers actually use.
That is a mature, comprehensive, mandatory transparency regime, more than a decade old, national in scope, running years before Britain's equivalent.
So researchers looked. Emmanuel Asane-Otoo and Bernhard Dannemann examined daily petrol and diesel prices from virtually every filling station in Germany across 1 January 2014 to 31 December 2018, five years of the transparency era, at essentially full coverage.
Their conclusion is blunt: asymmetric pricing is, in their words, "the norm rather than exception". The rockets-and-feathers pattern persisted. They point to two candidate causes: tacit coordination among competitors, and uneven search intensity on the part of consumers.
The half that is about us
That second cause deserves more attention than it usually gets, because it is the one drivers can actually move.
Publishing a price does not make anybody look at it. If most drivers refuel by habit (the station on the way to work, the one by the supermarket) then a retailer's incentive to cut quickly is weak even under perfect disclosure. It only pays to drop your price fast if a meaningful number of customers are checking and will switch. Transparency creates the possibility of a competitive response; search behaviour decides whether one actually arrives.
The Korean study reaches a compatible conclusion from a different direction, attributing the asymmetry primarily to search-cost explanations rather than to oligopolistic coordination.
Both of them are describing the same thing from opposite ends: the binding constraint is often not what retailers are permitted to hide, but how much effort it takes a driver to find out.
What follows for a driver
- Expect the lag and use it. After a fall in crude, the cheapest forecourts move first and the captive ones move last. The days after a drop are when local price gaps are widest, which makes it the moment when checking pays best.
- Do not read a stubborn price as proof of profiteering. A station that has not moved may be selling fuel it bought at the old cost. This is the honest half of the retailers' argument, and the data supports it existing. It just does not explain why the same logic fails to slow price rises.
- Judge on the way down, not the way up. Everyone raises prices at similar speed. The difference between forecourts shows up in how quickly they follow costs downwards, and that is worth remembering about your regular stop.
- Habit is the expensive part. If the research says search intensity is a cause, then always using the same station is not a neutral choice. It is the behaviour the pattern feeds on.
- Do not expect transparency alone to fix it. Germany has had real-time mandatory reporting since 2013 and the asymmetry stayed. Data changes what is knowable; it does not change what people do with it.
Why we find this uncomfortable, and publish it anyway
The two guides before this one made a case that better price data is worth having. This research complicates that case, and it would be dishonest to leave it out.
Germany's decade of full mandatory transparency did not end rockets and feathers. A map is not a market remedy, and we should not pretend otherwise. Anybody claiming their app will make fuel cheaper for everybody is claiming something that a national regulator with statutory powers has not achieved.
What the same research does say is where the remaining leverage sits: on the search side, with drivers. If uneven search intensity is part of what sustains the pattern, then the useful thing a tool can do is make looking cheap: fast, free, no account, no app, and honest about how old each number is. That does not fix the market. It does put a driver on the better side of a lag that is otherwise working quietly against them.
Prices will keep rising like rockets. Whether they fall like feathers depends, in part, on how many people bother to check.