The International Competition Network presents a scoping paper on “Big Data and Cartels” analyzing the impact of digitalization in cartel enforcement

First, big data and algorithms can be a new type of vehicle for collusion. In that respect, the paper offers some practical definitions of these often used terms (see §10-17 of the paper). There are different instances in which big data and algorithms can affect explicit or tacit collusion, taking into account the fact that from the legal point of view, tacit collusion generally escapes the application of competition rules.

Algorithms can affect two structural factors of collusion, i.e. the frequency of interaction and market transparency. Simultaneous use of algorithms, by competitors, without prior contact between them, raises several questions. This type of use may fall outside the scope of legal provisions prohibiting agreements between rivals, or be part of a larger anticompetitive coordination.

Competition enforcers may face several challenges, should big data and algorithms be a means for illicit collusion between competitors.

When there is no explicit concurrence of wills between the firms which use digital tools to set and adjust prices, two questions will be raised: could the monitoring and adaptation capacities provided by big data and algorithms be taken into account to demonstrate a concurrence of wills? How could the level of market transparency allowed by these technologies change methods to analyze documentary evidence by competition authorities? There may be issues concerning the intentional element and the evidentiary standard (e.g., is consciousness of the collusive outcome sufficient to fall within the provisions of anti-competitive agreement? Is there room to apply presumptions?)

Concerning the restrictive aspects of agreements, the Scoping paper looks at the role that market structures might play. For instance, should they be integrated as economic context, or as potential evidence to demonstrate the manifestation of will?

Regarding the liability issue, several questions that may be raised in relation to the control over an algorithm are listed: when a firm applies an algorithm’s autonomous decision, will the algorithm be deemed to remain under the firm’s control? May the company that developed the algorithm and the beneficiary firm be found jointly and severally liable?

The issue of monitoring capacities and duration continuity of the practice is also flagged. The paper looks, for instance, at the case in which an IT company knowingly provides a group of competitors with algorithms or data management services allowing illegal coordination, where this IT company might be looked at as a facilitator.

Big data can also be considered in cases that go beyond explicit collusion. Some forms of parallel use of algorithms may give rise to identical prices and commercial terms by different economic operators, in particular when the algorithm includes predictive features.

The second part of the paper looks at the different ways in which big data and algorithms can be used by competition agencies as a tool to detect cartels. The challenges to digital cartels detection and investigation are listed: dealing with large volumes of data, localization of digital information (“access approach” or “location approach”), cartel screening and fight against bid-rigging.

The paper finally looks at possible fruitful cooperation with actors outside competition enforcers, such as public procurement authorities, anticorruption / fraud bodies, or criminal enforcement authorities. With these agencies, the use of certain data sets that they have gathered might be useful, as well as sharing techniques and methodologies concerning data analytics.

In conclusion, the interest of advocacy in this domain is underlined, for example by disseminating a competition culture to those who design and use tenders, or by encouraging the development of electronic procurement policies by national authorities