Optimal Legal Standards and Accuracy in Antitrust Enforcement Giovanni Immordino Universit di Salerno and CSEF Michele Polo Universit Bocconi and IGIER June 2, 2011 Abstract: This paper analyzes optimal legal standards and accuracy in antitrust enforce- ment. The enforcer chooses the legal standard, the ne schedule and, if the decision rule entails errors, the level of accuracy. The policy problem is applied to traditional industries, where rms choose within a given set of feasible practices, and innovative industries, where the rm can enlarge the set of practices by investing in research. Marginal deterrence is central in traditional environments, while in innovative ones also the e/ect of enforcement on research investment - average deterrence - matters. For unlimited nes a discriminating rule implements the rst best in traditional industries opting for minimum accuracy. Instead, for innovative industries the need to sustain research leads to select a per-se legality rule when the probability of social harm is low; the optimal legal standard becomes, when harm is more likely, a discriminating rule and type-I accuracy is improved to limit over-deterrence and sustain research. When nes are limited, in traditional industries a discriminating rule is initially adopted together with type-II accuracy to improve marginal deterrence, while per-se illegality is optimal when the social damage is more likely. Finally, limited nes in an innovative industry leads for increasing likelihood of social harm rst to per-se legality, then to a discriminating rule and a more and more symmetric level of accuracy, and nally to per-se illegality. Keywords: legal standards, accuracy, antitrust, innovative activity, enforcement. JEL classication: D73, K21, K42, L51. Acknowledgments: Giovanni Immordino Universit di Salerno and CSEF, 84084 Fis- ciano (SA), Italy, giimmo@tin.it. Michele Polo, Universit Bocconi, Via Sarfatti 25, 20136 Milan, Italy, michele.polo@unibocconi.it. We are indebted to Nuno Garoupa, Yannis Kat- soulakos, Dilip Mookherjee, Massimo Motta, Marco Pagano, Patrick Rey, Matteo Rizzolli, Lars-Hendrik Rller, Maarten Schinkel and Giancarlo Spagnolo for helpful discussions. All usual disclaimers apply.