AI cartels: what does artificial intelligence mean for competition policy?

Adam Smith warned of firms within an industry colluding to charge higher prices. Such concerns are magnified in a time of online algorithms and instantaneous price adjustments.

People of the same trade seldom meet together, even for merriment and diversion, but the conversation ends in a conspiracy against the public, or in some contrivance to raise prices.’

This famous quotation from The Wealth of Nations was written at a time when cartels were organised in physical, usually smoke-filled, rooms between real people. There are good reasons to think that this is no longer the case.

Economists have long worried about the possibility of tacit collusion, whereby firms adjust their pricing without formally agreeing to do so. This was demonstrated by General Electric (GE) six decades ago. Having been convicted in 1960, along with Westinghouse, of explicit collusion in the market for turbine generators – and then having seen prices fall by a half in the following three years – GE began to post prices publicly. It released its pricing book, announced its pricing policy, sat back and watched Westinghouse follow suit. The result: the prices and profits of both firms rose.

This example of tacit collusion involved people making decisions. What would Smith make of the possibility of algorithmic collusion that involves no human beings?

This is the idea that firms can set prices online using algorithms that are able to respond to changes in the prices of competing products in real time. It is not difficult to understand why this might allow tacit collusion. Any firm can immediately respond to a competitor’s price cut, thus removing most of the incentive to lower prices in the first place.

Equally, the algorithm could experiment by hiking prices and seeing if competitors respond. If they do, great. If not, then the algorithm can bring the price straight back down, all within the blink of an eye.

This possibility is a nightmare for competition authorities. Firms reacting to the decisions of their rivals is the essence of price competition. But how do we know when we have too much of it? And what can we do about it? How might authorities stop firms reacting to their rivals’ price changes?

This is not a far-fetched scenario. The Trod/GB eye cartel case in 2016 – in which two online sellers were colluding around the sale of posters and frames on Amazon – was all about using online repricing software to monitor the prices of rivals and implement an agreed cartel.

In that case, there was an agreement between humans to create the cartel. But now even that may not be necessary. Recent experimental research shows that relatively simple algorithms based on artificial intelligence can lead to prices above what would be sustainable in a competitive market.

In 2017, Margrethe Vestager, the European Union’s commissioner for competition, said:

‘It's true that the idea of automated systems getting together and reaching a meeting of minds is still science fiction… But we do need to keep a close eye on how algorithms are developing… So that when science fiction becomes reality, we're ready to deal with it.’

I’m not sure we’re in the realm of science fiction any more.