Leniency policies and rewards for whistleblowers are being introduced in ever more fields of law enforcement, though their deterrence effects are often hard to observe, and the likely effect of changes in the specific features of these schemes can only be observed experimentally. This paper reports results from an experiment designed to examine the effects of fines, leniency programs, and reward schemes for whistleblowers on firms' decision to form cartels (cartel deterrence) and on their price choices. Our subjects play a repeated Bertrand price game with differentiated goods and uncertain duration, and we run several treatments different in the probability of cartels being caught, the level of fine, the possibility of self-reporting (and not paying a fine), the existence of a reward for reporting. We find that fines following successful investigations but without leniency have a deterrence effect (reduce the number of cartels formed) but also a pro-collusive effect (increase collusive prices in surviving cartels). Leniency programs might not be more efficient than standard antitrust enforcement, since in our experiment they do deter a significantly higher fraction of cartels from forming, but they also induce even higher prices in those cartels that are not reported, pushing average market price significantly up relative to treatments without antitrust enforcement. With rewards for whistle blowing, instead, cartels are systematically reported, which completely disrupts subjects' ability to form cartels and sustain high prices, and almost complete deterrence is achieved. We also analyze post-conviction behavior, finding that there is a strong expost deterrence (desistance) effect. Moreover post-conviction prices are on average lower than before even though the average prices within cartels are the same. Finally, we find a strong cultural effect comparing treatments in Stockholm with those in Rome, suggesting that optimal law enforcement institutions differ with culture.
