Personalized pricing isn't the only regulatory concern around algorithms

29 October 2018 3:29pm

12 October 2018. By Simon Zekaria.

Some of the world’s biggest airlines, retailers and online booking companies will be relieved that the UK antitrust authority found scant evidence of computer algorithms being used to personalize customer pricing. But a slew of other issues related to algorithms suggests that they shouldn’t rest too easy.

Expedia, Tesco, Ryanair, Amazon, Apple and Zara all had their product-pricing tested by the Competition and Markets Authority for a study published this week, which analyzes how pricing algorithms are used by companies and how they could lead to competition concerns.

The focus of the research isn’t a surprise. Antitrust watchdogs are increasingly mindful about the impact of algorithms — sets of rules used by computers in data processing — on both competition and consumer choice. But this week’s study did raise the temperature of the discussion — and dragged big corporate names into the debate at the same time.

Algorithms can either cement a collusion strategy or themselves create unlawful coordination, ultimately resulting in competition infringements and consumer abuse in the form of higher prices, the CMA says. The watchdog classifies these behaviors as “explicit collusion” or “tacit coordination,” crossing a wide variety of online behavior.

The use of algorithms isn’t a problem it itself. Regulators across the world agree that they can bring benefits: cutting business costs that can be passed on to consumers as savings, smoothing buying transactions, and arming consumers with information on purchases. But they do raise a wide variety of concerns.

— Personalized pricing —

One of the most pressing worries is about personalized pricing — offering different prices to different consumers based on their data. On that front, the CMA found “limited” evidence of algorithms being used, despite a “widespread of use of algorithms to set prices” in the broader sense.

In other words, algorithms aren’t being used to personalize prices, even as they are used for the ranking of products and services, advertising and offers.

The CMA also said that it’s “unlikely” that tacit coordination and personalized pricing occur within the same market. For the regulator, price comparisons are easy for customers, and the increasing use of data and algorithms doesn't alter that.

The CMA’s finding on personalized pricing echoes that of its predecessor, the Office of Fair Trading, five years ago. In 2013, the OFT said there was no evidence of it in the UK.

— Other issues —

But while this particular finding gives succor to companies such as Amazon, whose power as a combined online platform and retailer is under intense scrutiny, it may not last long. The CMA’s paper also highlighted a slew of algorithmic practices on the regulatory radar that give it cause for concern.

First, it said that as companies get more and more data on consumers, and as algorithmic modelling software advances, the danger of personalized pricing escalates. The use of “data-scraping methods allows for real-time data collection on consumers and competitors” that foster algorithmic pricing behavior, the CMA said, and “raises the possibility” of personalized pricing.

Second, it reminded algorithms can be used to implement illegal price-fixing and form cartels.

On this matter, the CMA has been here before. In 2016, the authority issued its first director disqualification in banning Daniel Aston — managing director at online poster supplier Trod — from acting as a company director for five years.

Trod had colluded with rival GB Posters not to undercut each other’s prices for posters and frames sold on Amazon’s UK website through the use of automated software to monitor prices. The sellers used repricing software to automatically review and adjust their prices to stay in line with the other, the CMA says.

Another looming issue is the use of algorithms without human involvement. As computing power grows and machine learning becomes more potent, the possibility of fully automated collusion could grow, the CMA said.

— Theories of harm —

The CMA stresses that the paper is focused on economic rather than legal analysis. That allows the regulator ample breathing room to advance its theories of harm and scale up its enforcement approach.

It already has a heavy body of work exploring these matters. Online consumer services are increasingly attracting scrutiny and UK enforcers have already looked at a plethora of sectors, such as price-comparison services, hotel-booking sites, ticketing websites and online retail.

With a specialized antitrust enforcement unit up and running to increase surveillance of data and the digital economy, the CMA is stepping up its gaze. In the EU, too, pricing algorithms are likely to come under increased scrutiny from EU antitrust regulators under their own study.

The CMA’s work will also be helped by a UK government push for sector regulators, notably the Financial Conduct Authority, to develop performance metrics for companies and platforms.

And a government-led expert panel working to assess competition and consumer harm across digital technology markets, which will report back to the government, includes a focus on algorithms.

That initiative follows a wider UK government review into competition policy aimed at “modernizing consumer markets” including a focus on digital technology.

The CMA knows it has to be ahead of the curve. On consumer enforcement, it has asked for additional fining powers and more expansive inquiry timeframes. And a revamped toolkit will be become ever more important as complex antitrust cases involving digital industries that are currently reviewed in Brussels will fall into its lap after Brexit.

Companies whose business depends on elaborate online strategies may read the paper with a sigh of relief, but the feeling will be temporary. The regulatory scrutiny of algorithms has only just begun.

Joh Sung-Wook Interview