The Stanford Computational Antitrust Project announces the publication of “Computational Rule of Reason Analysis in Antitrust” by Keith N. Hylton and Karina Popowycz. The article appears in Volume 6 of Stanford Computational Antitrust (pp. 288-307).

Courts applying the rule of reason say they balance anticompetitive harms against procompetitive benefits. Few of them measure either side, and the court in Microsoft III did not. The authors ask what that balancing would look like if it were done with numbers. They build a framework in which each term of the test can be estimated and compared.

The framework weighs two kinds of effects. The static effect sets the harm to consumers and the deadweight loss from monopoly pricing against the operational efficiencies the monopolizing act may generate, such as lower supply costs. The dynamic effect adds the surplus a firm creates when it invests to build the market it later monopolizes. From this model, the authors derive the optimal antitrust penalty. When innovation does not respond to punishment, the penalty should internalize consumer harm. When monopolization does not respond, the optimal policy becomes a subsidy reflecting the innovation surplus. Most cases fall in between, and simulations show how the optimal penalty moves as consumer harm and innovation surplus vary.

The authors then turn to the decision courts actually make, which is binary. They treat an injunction as a penalty that removes the monopolist’s entire expected gain. Where static harm is positive and innovation trivial, an injunction is optimal. Where efficiencies exceed static costs, it never is. In the hardest cases, some punishment is desirable but an injunction overshoots, and the authors show how to compute whether the cost of overenforcement outweighs the gain from deterrence.

The article closes on implementation. Welfare components can be estimated from market data, and recent machine learning methods help estimate markups without strong assumptions about firm conduct. The authors are candid about the hardest part. The firm knows its own costs better than any court, so the estimation process must give it reasons to disclose them.

Thibault Schrepel, founder of the Stanford Computational Antitrust Project, stated:

“Courts have applied the rule of reason for more than a century without saying how to weigh one side against the other. Keith Hylton and Karina Popowycz write down the balance and show that each of its terms can be estimated. They also bring innovation into the calculation, which changes the answer in many cases, sometimes to the point of favoring a subsidy over a penalty. This is how a legal standard becomes something an agency can compute.”

The article is available for download on the Stanford Computational Antitrust project’s page.