Antitrust Amnesty, Game Theory, and Cartel Stability

Article excerpt

I. INTRODUCTION

While most areas of antitrust law breed controversy over which types of business conduct should be illegal, there is one area of relatively broad consensus: price-fixing cartels should be condemned. Cartels injure consumers by raising prices. For example, a cartel of producers of the amino acid lysine increased prices by 70% within the cartel's first six months of operation and relatively quickly doubled prices.1 Cartels create allocative inefficiency by reducing production in order to raise market price, and they encourage productive inefficiencies by protecting inefficient manufacturers, which can increase the average production costs in an industry.2

Yet some disagreement remains about the most efficient mechanism for detecting, punishing, and deterring cartels. While some commentators have argued that cartels are inherently unstable and will inevitably dissolve without active antitrust enforcement, others have advocated a more aggressive approach. The primary focus of current cartel-fighting efforts has been to offer amnesty to the first cartel member to defect and expose the price-fixing conspiracy to federal prosecutors.

The Antitrust Division of the Department of Justice (DOJ) reformed its amnesty policies in 1993. Its Corporate Leniency Policy (amnesty program) has been the "most effective generator of cartel cases and is believed to be the most successful program in U.S. history for detecting large commercial crimes."3 Before the new program, the government received one application each year from firms willing to expose a cartel in exchange for leniency. Currently, the government is receiving three applications each month. The new policy has led to the exposure of international cartels in marine transportation services, graphite electrodes, bromines, and vitamins,4 resulting in over one billion dollars in fines.5 These fines represent a sea change in antitrust penalties. Individual firms whose cartel activities have been exposed through the amnesty program have paid fines over $100 million, including Hoffman-LaRoche's agreed-to fine of $500 million for its participation in the vitamin cartel.6

The purpose of my inquiry is to examine why the Corporate Leniency Policy has been so successful and, based on that discussion, to suggest changes that would increase the program's effectiveness even further. It is intuitive that if the government offers an incentive to confess, then confessions will increase. But the mechanism by which the amnesty program works is far more complicated and nuanced. Using a series of prisoner's dilemma models, this Article argues that government policies work by creating distrust among cartel co-conspirators. Understanding this mechanism will facilitate reforms to make antitrust law in general, and the DOJ's amnesty program in particular, work more effectively.7

This Article presents a paradox. The conventional wisdom holds that imposing high penalties should deter illegal cartelization and that rewarding minor players who confess their role in a price-fixing conspiracy can help unravel a cartel. Thus, it seems counterintuitive to expect that policies making it easier for the worst antitrust criminals to escape all criminal and most civil liability would enhance deterrence and destabilize existing cartels. Yet proper application of game theory models suggests that extending amnesty even to cartel ringleaders (who have often profited the most from the illegal activity, sometimes for decades) makes cartels more fragile. Perhaps even more counterintuitively, the theory shows that an antitrust leniency program would have a greater destabilizing effect on price-fixing cartels if the first firm to confess receives full amnesty even though antitrust prosecutors, through their own investigation prior to any confession, had already acquired sufficient evidence to convict the firm of criminal price-fixing.

Part II introduces the basic prisoner's dilemma model and explains its relevance to the prosecution of price-fixing. …