Competition Bureau takes baby steps on its Big Data learning curve

  • December 14, 2017
  • Doug Beazley

Big DataEvery conversation about competition law and big data seems to start with Google.

The internet behemoth’s parent company took in $78.65 billion over the past three quarters as its stock hit all-time highs. So it probably didn’t break a sweat when the European Union slapped it with a record US$2.7 billion antitrust fine in June — punishment for abusing its dominance of the search engine market by placing its own shopping services above those of competitors on its web search queues.

In the U.S., Missouri recently signalled what could be the start of an antitrust push against Google with an investigation into whether the company has manipulated search results to frustrate competition. And Canada? In April 2016, the Competition Bureau wrapped up an abuse-of-dominance investigation of Google with a report that largely absolved the company of misdeeds (although it did get Google to suspend the use of anti-competitive clauses in its advertising terms and conditions for five years).

That apparent gap between the approaches taken by Canadian and EU authorities might explain in part why the Bureau recently issued a discussion paper to invite discussion on the implications for competition law of big data — information culled from individuals online that can be turned into a sellable product, or used to design or refine other online products, such as advertising.

The paper opens with two caveats. First, the Bureau argues that the current Competition Act is robust enough now to navigate the uncharted waters of big data. Second, it restates a principle: The Bureau’s job is to protect competition, not outcomes. Overly-aggressive intervention in big data marketplaces could chill innovation, it argues — and not every dominant company gets to the top by bending the rules: “Competition policy in Canada does not, and should not, assume that ‘big is bad’.”

All of which comes as something of a relief, if not a surprise, to many people practising competition law in Canada. “This is not a guidance document. It’s the beginning of a conversation. There’s an element of playing-it-by-ear here,” says Chris Hersh, a partner at Cassels Brock.

“I think it’s appropriate to suggest the Bureau is still at a point on the learning curve here. They’re taking a cautious approach.”

Caution seems to be the paper’s theme; throughout it the Bureau argues that it must approach big data competition cases cold, without tying its hands in advance with a lot of minutely-detailed policy. Lawyers practising competition law, meanwhile, say they’d like to see a little deeper into how the Bureau intends to approach big data in future.

Take the paper’s section on market definition – who’s selling what, to whom. Because big data firms often operate in “multi-sided” markets, collecting data from customers in exchange for “free” services (like a search engine), then commercializing the data collected, it can be hard to define the market because there are at least two sets of customers.

In such cases, the Bureau says it may have to rely on “alternative methods” to define the market. — or skip the step altogether. In its submission on the discussion paper, the Competition Law Section says it would like to know what kind of alternative methods the Bureau has in mind.

In the paper’s section on market power, the Bureau suggests costs imposed on customers for switching their data from one firm to another could frustrate less-established players trying to get into the market, constituting a form of abuse of dominance. Competition lawyers would like to know what the Bureau might consider as proof that switching costs aren’t frustrating market access. Would evidence that customers are bouncing between platforms, or using more than one, indicate there’s no dominance problem?

The Bureau suggests that it might distinguish between legitimate and illegitimate competition in big data cases through the use of the “no economic sense” test. The test determines whether a company could recover the costs of a market action — such as preventing a rival from accessing data — purely from profits that don’t stem from the action being evaluated. But the paper also says the complex nature of big data business models might prevent the use of the test. Competition lawyers want clarity on this point.

They also want assurances. At one point in the paper, the Bureau suggests forcing a company to share data with a rival may be “an acceptable structural remedy,” since big data is “non-rivalrous” — more than one company can use it at the same time. The Section says this should be a  nuclear option, and point out that since data is often collected using proprietary technology, divestiture could violate a company’s right to its intellectual property.

“It’s an extraordinary remedy for extraordinary circumstances,” says Hersh. “Ultimately, we want the Bureau to be really clear on what those circumstances might be.”

The Bureau’s decision to cite privacy as a non-price dimension of competition raised eyebrows as well. “Competition,” says the paper, “can drive big data firms to provide more privacy than legally required … and ”mergers between big data firms could eliminate the incentive to offer customers extra privacy guarantees.” Competition lawyers argue that adding privacy to the mix risks adulterating competition law with “public interest” considerations — that privacy isn’t really part of the Bureau’s brief.

The paper touches on cartels long enough to assert that big data is a powerful tool for manipulating markets — but it doesn’t really change anything from an enforcement perspective. “Despite the increasing sophistication of the tools, the (cartel) offence is still rooted in the agreement itself.”

As with the section on privacy, the paper’s section on deceptive marketing practices has some competition lawyers wondering if the Bureau isn’t straying from its turf somewhat. The paper cites instances where customers downloading apps may not know that they’re also delivering data to the app provider. That’s a task for federal privacy law, say competition lawyers.

All in all, the Bureau’s paper reads like what it is: a first draft of an approach to applying competition law to new data tools of unprecedented power. Given how little case law there is on big data and competition — and given the broad difference between the approaches being taken by European and North American authorities — the Bureau probably can be forgiven for taking baby steps.

“We would have appreciated more guidance, some analytical tools or approaches, yes, that would have been helpful,” says Hersh. “But I understand why they’re being cautious. Early days yet.”

Doug Beazley is a frequent contributor to PracticeLink.