The price of gas is a puzzle. Monitor the average price in gas stations in a particular city and it will vary dramatically, sometimes in a matter of hours and often in ways that appear cyclical.
Economist have long scratched their heads over this kind of pattern. One explanation is that this behaviour emerges when two competing companies change their pricing strategy at each stage by reacting to the other. The resulting behaviours are known as Edgeworth Price Cycles.
The problem is that gas station prices are not controlled by two competing players but many competing retailers. It’s easy to assume that the many-body problem produces similar patterns but nobody has been able to show this.
Until now. Today Tiago Peixoto and Stefan Bornholdt, physicists at the University of Bremen in Germany, show how a more complicated model with many buyers and sellers reproduces this kind of behaviour.
But it also goes further. Peixoto and Bornholdt say that when condition are right, cartel-like behaviour emerges naturally without collusion between sellers.
Theirs is an agent-based model where all players are both buyers and sellers. The buying behaviour is determined by a value for many criteria. So players buy from a certain number of sellers but can change a seller at each step if they find one offering better value for money.
The sellers can also change their value for money parameter at each step by looking at other sellers. If they find one offering better value for money, they match it.
Peixoto and Bornholdt study this behaviour in a population of a million agents over time period of a billion iterations and more.
The results make interesting reading. It turns out that a crucial factor is the speed at which buyers and sellers react to the market. When buyers react quickest, sellers are forced to match the best possible value for money and prices tend to drop.
By contrast, when sellers react quickest, they are quick to copy others offering poor value for money. This reduces the number of sellers offering good value for money in a vicious cycle that drives prices as high as possible.
This is the emergence of a cartel and it happens in these guys’ model without any collusion between sellers. Instead, it is an emergent property of the market place that happens when the sellers outperform buyers in the way they react to market conditions.
“This cartel organization is not due to an explicit collusion among agents; instead it arises spontaneously from the maximization of the individual payoﬀs,” say Peixoto and Bornholdt.
These guys are clearly studying a parameter space displaying a rich variety of patters. And the cartel-like region of this space has its own patterns of behaviour. It is categorised by sudden and dramatic price variations, particularly moving suddenly upwards but decaying only slowly. These variations can also appear cyclical (but are actually aperiodic).
This more or less exactly matches the price behaviour at gas stations and many other economic areas, such as electricity and natural gas prices in Europe. It would be interesting to see if this kind of behaviour emerges in other markets such as eBay.
The big question of course is what to do. Cartels that form by collusion are illegal and clearly not in the interests of the general population.
But this work muddies the waters somewhat. If cartel-like behaviour is an emergent property of an ordinary market, how should it be controlled, regulated and punished?
The good news is that various strategies could easily be tested using this kind of agent-based model. The bad is that new strategies may themselves lead to emergent properties that are hard to spot in advance.
An interesting new puzzle for econophysicists.
Ref: http://arxiv.org/abs/1201.3798: No Need For Conspiracy: Self-Organized Cartel Formation In A Modiﬁed Trust Game
How AI is reinventing what computers are
Three key ways artificial intelligence is changing what it means to compute.
These weird virtual creatures evolve their bodies to solve problems
They show how intelligence and body plans are closely linked—and could unlock AI for robots.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
Surgeons have successfully tested a pig’s kidney in a human patient
The test, in a brain-dead patient, was very short but represents a milestone in the long quest to use animal organs in human transplants.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.