The Impact of Algorithms and AI on Merger Policy

Algorithms, especially those based on artificial intelligence (AI), play an increasingly important role in our economy. They bring cost savings, speed, precision, and sophistication to decision-making and planning processes, improving both day-to-day decisions and long-term innovation, strategy, and vision. As a result, they are used by market participants to make pricing, output, quality, and inventory decisions; to predict market entry, expansion, and exit; and to anticipate regulatory moves. This game-changing switch to (semi-)automated decision-making creates many benefits to users and consumers. At the same time, it also has the potential to reshape market dynamics in ways which might harm competition and consumer welfare. While the effect of algorithms on coordination between competitors has been a focus of attention, and scholarly work on their effects on unilateral conduct is beginning to accumulate, merger control issues have been undertreated. Accordingly, our recent article, ‘Algorithms, AI, and Mergers’, forthcoming in the Antitrust Law Journal, explores the effects of the use of algorithms on merger policy.

To do so, the article first identifies six main functions of algorithms that may affect market dynamics: collecting and ordering data; improving the ability to use existing data; reducing the need for data, for instance by generating synthetic data; monitoring; predicting how different types of conduct, including mergers, are likely to affect market conditions; and decision-making.

The article then demonstrates how such algorithms can exacerbate anti-competitive conduct with respect to both unilateral and coordinated effects. To this end, eight scenarios are explored: collusion, oligopolistic coordination, high unilateral prices, price discrimination, predation, selective pricing (in which a buyer offers a higher price to some suppliers in an aggressive bid for an input), tying, and reducing the interoperability of datasets. For each scenario, we analyze how the market conditions necessary for such conduct are affected by algorithms. We exemplify our claims by analyzing cases from around the world which attempt to deal with the effects of algorithms on competition (which are also useful for teaching competition in digital spheres).  Take, for example, predatory pricing, where a seller charges a price below some measure of its incremental cost of production, in the expectation that its less resilient competitor will not be able to sustain losses, and will either exit or choose not to enter the market. Algorithms can provide an efficient low-cost tool for predation: predators can use digital profiles to target only those consumers that would likely switch suppliers, and the use of algorithms can help firms finance their predatory actions by allowing them to maintain a profit-maximizing pricing scheme for inframarginal consumers.

These findings are then translated into merger policy. Algorithms are shown to affect substantive as well as institutional features of merger control. Algorithms also challenge some of the assumptions that are ingrained in merger control, suggesting that a more informed approach to some algorithmic-related mergers is appropriate.

To illustrate, consider the following two examples. First, algorithmic coordination affects the structural presumptions used to screen mergers. Concentration parameters (such as the Herfindahl-Hirschman Index) are given substantial weight in determining intervention thresholds, based on a general assumption that coordination is likely to take place where the market is highly concentrated. The current level at which these parameters are set assumes that mergers in markets with four or more firms are not likely to lead to coordination. Algorithmic coordination challenges these assumptions. Accordingly, where the risk of algorithms coordination is high, concentration parameters might need to be lowered.

Second, merger review gives much weight to market asymmetry. Yet, pricing algorithms can in some circumstances reduce the incentive of asymmetric firms to deviate. Similarly, while it is true that firms are better able to estimate the cost structure, production capacity, and other key supply conditions of symmetric competitors compared to asymmetric competitors, algorithmic predictive modeling can help firms to understand their asymmetric competitors and the prevailing demand conditions, thereby simplifying the process of establishing a supra-competitive equilibrium. Accordingly, the weight of asymmetry in the analysis should be educed where algorithms are involved.

The article makes numerous observations and suggestions. The importance of analyzing the effects of algorithms on merger policy is underscored by the fact that the consequences of algorithmic use are not efficiently addressed by alternative consumer protection or competition laws. We can thus not afford to disregard, or give insufficient weight to, the effects of algorithms on merger review.

Michal S. Gal is Professor and Director of the Center of Law and Technology, University of Haifa Faculty of Law, and the Former President of the International Association of Competition Law Scholars (ASCOLA).

Daniel Rubinfeld is Professor of Law at NYU and Robert L Bridges Professor of Law and Professor of Economics Emeritus at UC Berkeley.