#518 1993 · U.S. Department of Justice, Antitrust Division · Law enforcement / antitrust regulation
The Justice Department stopped trying to catch cartels from outside and started paying whoever inside confessed first
the problem
Secret price-fixing cartels leave almost no evidence visible from outside the room
background
A price-fixing cartel is built to be undetectable: the only people who ever see the agreement are the members who made it, and every one of them profits from the others' silence holding. DOJ's original 1978 Corporate Leniency Program already offered confessors reduced treatment, but the reward was a prosecutor's discretionary call, not a guarantee — a company weighing whether to confess could not calculate what it would actually get, only that it would be handing over a finished case against itself. The program drew about one application a year and never cracked a major cartel.
The standard tools for catching secretive crime — raids, subpoenas, wiretaps, longer sentences — all require already knowing roughly where to look, and cartels exist precisely to leave no such signal outside the conspiracy. DOJ's actual problem was not a shortage of investigators or evidence; every cartel member already held the evidence to convict every other member. The shortage was a reason for any single one of them to be first to hand it over.
what everyone would do
The standard answer to catching something that leaves no outside trace is to build a better outside trap — more raids, subpoenas, wiretaps, harsher sentences. DOJ already had all of these and, by its own account, the original 1978 leniency policy still drew only about one application a year, because a discretionary reward gave no rational executive a reason to trust that confessing wouldn't simply hand prosecutors a finished case.
what they saw
The evidence needed to convict a cartel was never missing — every member already held it about every other member. What was missing was a reason for any one of them to be the one who talks while the others might stay quiet. Making amnesty automatic by rule, and exclusive to whichever member arrives first, turns a cartel's shared silence — safe only for as long as it holds completely — into a race no member can afford to lose.
the move
In August 1993 the Antitrust Division rewrote its leniency policy into an automatic entitlement rather than a discretionary favor: full amnesty from prosecution, guaranteed by rule, to the first cartel member to confess before an investigation begins, with lesser guarantees for confessing once one is already underway — and protection extended to the confessing company's individual officers and employees, not just the corporation.
why it works
Guaranteeing amnesty by rule instead of discretion lets a company finally calculate the value of confessing instead of guessing at it. Making that guarantee exclusive to the first arrival flips each member's rational move: staying silent is only safe if every other member also stays silent, and once a no-risk exit exists for exactly one confessor, no member can verify the others won't take it. DOJ's own account describes the trigger as an empty chair at a scheduled cartel meeting — is the missing member late, or already at the prosecutor's door — and that fear is what produced the near-twentyfold jump in applications.
the payoff
Leniency applications rose nearly twentyfold after the 1993 revision, per the Division's own account, turning it into what DOJ calls its single most effective investigative tool. Since fiscal year 1996, cartel fines in the United States have exceeded $5 billion, with more than 90 percent of that total tied to investigations a leniency applicant helped initiate or advance.
where it breaks
Hammond's own account names three legs the mechanism stands on: a credible threat of severe sanction for staying silent, a real perceived risk of getting caught at all, and enough transparency that a company can predict how confessing will actually be treated. Remove any one leg and there is no race to win — amnesty is worthless to a group that doubts it will ever be caught, or that doubts the state will honor the deal. A separate strand of the game-theory literature (Miller 2009, citing Spagnolo 2000) also notes leniency is not unconditionally destabilizing: under some parameter regimes it can instead help a cartel police its own members' cheating, so the policy needs pairing with real enforcement risk to reliably work in the direction intended.
what came after
The revised program became, in DOJ's own words, the cornerstone of cartel enforcement worldwide: the number of jurisdictions running an effective leniency program went from essentially one in 1990 to more than fifty by 2010, after the European Commission (1996, revised 2002) and Canada (2000) adopted convergent versions. A 2009 American Economic Review study by Nathan Miller supplied the first independent empirical evidence, using twenty years of cartel indictment data, that the policy measurably increases both detection and deterrence.
references
- [1]The Evolution of Criminal Antitrust Enforcement Over the Last Two DecadesU.S. Department of Justice, Antitrust Division (Scott D. Hammond), 2010justice.gov
- [2]Strategic Leniency and Cartel EnforcementAmerican Economic Review, 2009aeaweb.org