Criminal Antitrust and the Leniency Program: Cartels, Grand Juries, and Corporate Exposure

Document type: Article Practice area: Corporate — Antitrust Jurisdiction: United States (federal) Last reviewed: 5 September 2026

The race

Three companies fix prices. All three know it is unlawful. One of them — usually because a new general counsel arrives, or an employee leaves badly, or a compliance audit finds an email — decides to report it.

That company may receive complete immunity: no conviction, no criminal fine, and — critically — its cooperating executives are typically covered too. The others face felony convictions, fines calculated on the volume of affected commerce, and prison sentences for individuals.

There is one leniency spot, and it goes to the first company in. Everything about criminal antitrust enforcement follows from that structure. The program is designed to make cartels unstable by giving every member a powerful incentive to defect first, and it works: a substantial share of the Antitrust Division's cartel cases begin with a leniency application.

Which means the most important antitrust advice a company will ever receive is time-sensitive. A general counsel who discovers a possible cartel and spends three weeks investigating before deciding whether to report may find that a competitor spent three days.

What is criminal

Section 1 of the Sherman Act, 15 U.S.C. § 1 declares unlawful every contract, combination, or conspiracy in restraint of trade, and makes a violation a felony — punishable, for a corporation, by a fine, and for a person, by a fine and imprisonment.

But not every Section 1 violation is prosecuted criminally. The Antitrust Division reserves criminal enforcement for conduct that is per se unlawful — agreements so plainly anticompetitive that no inquiry into their reasonableness is required:

Price fixing — agreements among competitors on price, or on elements of price such as surcharges, discounts, credit terms, or fees.
Bid rigging — agreements about who will win, who will submit a losing bid, who will not bid, or how bids will be rotated.
Market allocation — agreements dividing customers, territories, or products.
Certain agreements affecting labor markets, including no-poach and wage-fixing agreements among employers, which the Division has prosecuted as market allocation and price fixing.

The classic statement of why these are treated categorically is United States v. Socony-Vacuum Oil Co., 310 U.S. 150 (1940), which held that agreements formed for the purpose and with the effect of raising, depressing, fixing, pegging, or stabilizing the price of a commodity are illegal per se, without regard to the reasonableness of the prices set or the good intentions of the participants.

Everything else is civil. Vertical restraints, monopolization under Section 2, most joint venture conduct, and information exchanges that fall short of agreement are analyzed under the rule of reason and are not charged criminally. Leegin Creative Leather Products, Inc. v. PSKS, Inc., 551 U.S. 877 (2007) confirmed rule of reason treatment for vertical price restraints, and American Needle, Inc. v. National Football League, 560 U.S. 183 (2010) addressed when concerted action exists at all among entities that might be seen as a single enterprise.

The line that matters in practice is between agreement and parallel conduct. Competitors may lawfully observe each other's public prices and respond. What they may not do is agree. The evidence that turns the former into the latter is usually a communication — a call, a meeting, a message — and that is what the investigation is looking for.

The penalty framework

Statutory maxima under § 1 are substantial, and they are frequently not the operative number.

18 U.S.C. § 3571(d) — the alternative fines provision — permits a fine of twice the gross gain to the defendants or twice the gross loss to the victims, in place of the statutory maximum. In a cartel affecting hundreds of millions of dollars in commerce, the alternative fine calculation produces numbers far exceeding any statutory cap, and it is how the largest antitrust fines are reached.

Sentencing. Corporate fines are calculated under the Sentencing Guidelines' antitrust provisions, which use the volume of affected commerce as the base and apply a culpability multiplier reflecting the organization's size, prior history, the involvement of high-level personnel, cooperation, and the existence of an effective compliance program. Individuals face imprisonment and fines under the same framework.

Individuals are prosecuted, and they go to prison. This is not a theoretical exposure. The Division charges individual executives as a matter of policy, sentences have lengthened over time, and foreign nationals have been extradited. For the executive in the room, the personal stakes exceed the company's by any measure that matters to them — which is the central fact in every joint defense and cooperation decision.

Limitations. The general federal criminal limitations period in 18 U.S.C. § 3282 applies — five years — but the conspiracy continues, for limitations purposes, as long as the conspirators receive payments or take acts in furtherance. That extends exposure well beyond the last meeting.

And the collateral criminal exposure is real. Obstruction under 18 U.S.C. § 1512, false statements under 18 U.S.C. § 1001, and conspiracy to defraud the United States under 18 U.S.C. § 371 are charged alongside — and sometimes instead of — the antitrust count. In more than one matter, the obstruction charge has been the one that produced the longest sentence.

The leniency program

The structure. The Antitrust Division's corporate leniency policy provides that a qualifying applicant receives complete protection from criminal conviction and fines for the reported conduct, and that its cooperating current directors, officers, and employees typically receive protection as well.

There is only one. Leniency is available to the first company to report. The second company through the door gets cooperation credit and a negotiated plea, not immunity.

The core conditions, in substance:

The applicant is first to report the conduct.
At the time of reporting, the Division had not already received information about the illegal activity from another source, or — under the alternative track — had information but not yet enough to sustain a conviction.
The applicant promptly and effectively terminated its participation.
The applicant reports candidly and completely, and provides full, continuing, and complete cooperation throughout the investigation.
The confession is a corporate act, not isolated confessions by individuals.
Where possible, the applicant makes restitution to injured parties.
The applicant did not coerce others to participate and was not the leader or originator of the activity.

The marker. Because the race is real, the Division permits an applicant to obtain a marker — holding its place in line for a limited period while counsel completes an internal investigation sufficient to make the full proffer. Obtaining a marker is frequently the single most consequential decision in the matter, and it can be made on far less information than a company would ordinarily want.

The tension this creates for counsel is genuine and should be understood before it arises. The prudent instinct is to investigate thoroughly before deciding. The structure rewards speed. A company that spends a month building certainty may find the leniency spot gone — and then be negotiating a plea while the competitor that acted in three days pays nothing.

Leniency is not costless. The applicant must cooperate completely, for years, which means producing documents, making executives available for interviews and testimony, and — where individuals do not qualify or do not cooperate — potentially watching its own people be prosecuted. It must terminate the conduct, which may have commercial consequences. And it remains exposed to civil damages, subject to the important qualification below.

The civil exposure, which usually exceeds the fine

A criminal antitrust resolution is the beginning of the financial consequences, not the end.

Treble damages. Section 4 of the Clayton Act, 15 U.S.C. § 15, gives any person injured in business or property by reason of an antitrust violation a claim for threefold the damages sustained, plus the cost of suit including a reasonable attorney's fee. The follow-on class actions are filed within days of a public resolution.

The government's judgment helps the plaintiffs. Section 5(a) of the Clayton Act, 15 U.S.C. § 16(a), makes a final judgment or decree in a government antitrust proceeding prima facie evidence against the defendant in a subsequent private action as to matters respecting which the judgment would be an estoppel — with an exception for consent judgments entered before any testimony is taken. A guilty plea is not that exception, which is why the plea's factual admissions are negotiated as carefully as the fine.

No contribution among conspirators. Texas Industries, Inc. v. Radcliff Materials, Inc., 451 U.S. 630 (1981) held that there is no right to contribution among antitrust wrongdoers. Combined with joint and several liability for the whole conspiracy's damages, this means any one defendant can be left holding the entire trebled exposure — which drives settlement dynamics far more than the merits.

The indirect purchaser structure. Illinois Brick Co. v. Illinois, 431 U.S. 720 (1977) limits federal damages actions to direct purchasers, and Hanover Shoe, Inc. v. United Shoe Machinery Corp., 392 U.S. 481 (1968) forecloses a pass-on defense against them. Many states permit indirect purchaser recovery under state law, so the practical result is parallel federal direct-purchaser and state indirect-purchaser litigation — and potential exposure exceeding the actual overcharge.

Pleading. Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007) requires enough factual matter, taken as true, to suggest that an agreement was made — parallel conduct alone is insufficient. That standard matters much less after a guilty plea, which supplies the agreement.

Venue is broad. Section 12 of the Clayton Act, 15 U.S.C. § 22, permits suit against a corporation in any district where it is an inhabitant, may be found, or transacts business, with worldwide service in some circumstances.

And the international dimension. Cartels are rarely national. A resolution in the United States is typically accompanied by proceedings in the European Union, the United Kingdom, Canada, Japan, Korea, Brazil, and elsewhere — each with its own leniency program, its own fines, its own timing, and, increasingly, its own follow-on damages litigation. Leniency in one jurisdiction does not confer it elsewhere, and applications must be coordinated globally, in parallel, on day one.

ACPERA, and why leniency is worth more than it looks

The leniency applicant still faces treble damages — except that Congress provided a specific benefit.

Under the Antitrust Criminal Penalty Enhancement and Reform Act, a successful leniency applicant that provides satisfactory cooperation to civil claimants is liable in the follow-on litigation only for single damages attributable to its own conduct, rather than treble damages and joint and several liability for the whole conspiracy.

The conditions are demanding. Cooperation must be timely and satisfactory, and typically includes providing a full account of the facts, producing documents, and making witnesses available to the plaintiffs — the same cooperation given to the government, extended to the people suing.

The court decides whether the cooperation was satisfactory, often late in the litigation, which means the applicant cooperates for years without certainty about the benefit.

But the arithmetic is powerful. Single damages on your own sales, versus treble damages with joint and several liability for the entire conspiracy and no right of contribution after Texas Industries, is frequently a difference of an order of magnitude. When counsel models the leniency decision, this is the line that usually decides it — and it is routinely omitted from the first analysis because it sits in the civil case rather than the criminal one.

A running example

Ferrocourt Alloys is a specialty metals producer with $1.1 billion in revenue and four significant competitors in a concentrated product category. Its general counsel is Solveig Baptiste-Nwankwo, three months into the job.

The discovery. A routine compliance audit of expense reports flagged a pattern: the vice president of commercial for one product line had attended six dinners over two years with counterparts from two competitors, none appearing on any trade association agenda. The audit was looking for expense fraud.

Day 1. Baptiste-Nwankwo read the expense reports and then, on a hunch, the executive's calendar. The dinners preceded the company's quarterly price announcements by two to nine days, consistently, for two years.

She did not investigate further before calling counsel. That instinct — to know more before escalating — is the one that loses the race. She engaged criminal antitrust counsel that afternoon.

Day 2. Litigation hold issued across email, chat, calendars, expense systems, phone records, and the personal devices of eleven individuals. IT confirmed suspension of every deletion rule in writing. An affirmative instruction went out that nothing was to be destroyed, and counsel documented delivering it.

Day 3. Counsel applied for a marker. The company knew very little: a pattern of dinners, a suggestive timing correlation, and no admission from anyone. It knew enough to describe conduct that could be a per se offense, which is what a marker requires.

Days 4–30. The internal investigation. Separate counsel retained for four individuals within the first week. The vice president, represented, ultimately described a five-year arrangement among three producers to signal and coordinate quarterly price movements. Two other employees had attended meetings and understood what was happening.

Day 6. Parallel leniency applications filed in the European Union, Canada, and Japan, coordinated by local counsel who had been engaged on day 3. This is the step companies most often perform late, and lateness there costs the foreign leniency entirely.

Day 31. Full proffer to the Antitrust Division. Conditional leniency letter signed.

Then the long part. Three years of cooperation: document productions, twenty-two witness interviews, grand jury testimony by four employees, and — the hardest part internally — assistance in the prosecution of the two competitors and of individuals at those companies.

The civil case. Follow-on class actions filed within a week of the first public charge against a competitor. Ferrocourt cooperated with plaintiffs to secure ACPERA treatment, providing the same account and the same witnesses. Four years later the court found the cooperation satisfactory, and Ferrocourt paid single damages on its own sales — approximately $41 million — rather than treble damages with joint and several liability for the entire conspiracy, which its economists estimated at over $600 million.

One competitor received no leniency and pleaded guilty, paying a fine calculated under 18 U.S.C. § 3571(d) on the volume of affected commerce. Two individuals, including Ferrocourt's former vice president — who did not qualify for individual coverage because he had left before the application — received custodial sentences.

Baptiste-Nwankwo's assessment: "We were first by about eleven weeks. If I had spent a month getting comfortable before calling counsel, we would have been second, and the difference is roughly half a billion dollars. The hardest thing I have ever done professionally was authorize a leniency application on three days of facts."

How an investigation actually begins

Companies imagine a letter. What happens is usually one of these.

The unannounced approach. Federal agents appear at an employee's home, in the evening, and ask to talk. They are not required to give warnings in a non-custodial setting, they may not identify the target, and the employee has no counsel. This is the Division's most effective technique and the reason employees need to know, in advance, that they may decline to speak and may ask for a lawyer.

The search warrant. Agents arrive at the office with a warrant, seize servers and devices, and interview whoever is present. The company's response in the first hour — where counsel is, whether employees know what to do, whether the scope of the warrant is recorded — shapes everything after.

The grand jury subpoena, for documents or testimony, served on the company or on individuals.

The competitor's call. A counterpart's counsel telephones to say their client has applied for leniency. This is sometimes the first notice, and it means the race is already lost.

Or an internal discovery — a compliance audit, a departing employee's disclosure, a document surfaced in unrelated litigation, or an acquisition's diligence.

In every case the first forty-eight hours matter disproportionately, because they determine whether evidence is preserved, whether employees say things that cannot be unsaid, and whether the company is first in line.

The first forty-eight hours

Engage antitrust criminal counsel immediately. Not the company's regular corporate firm unless it has this practice. The judgments required — whether to seek a marker, how to approach employees, whether to cooperate — are made by people who have done it.

Issue a litigation hold at once, covering email, chat platforms, shared drives, calendars, expense and travel records, phone records, and personal devices. Suspend every deletion rule. Confirm implementation in writing, because a hold announced but not implemented is worse than none.

Say nothing to anyone about destroying anything, and say so affirmatively. Obstruction under 18 U.S.C. § 1512 has produced longer sentences than the underlying antitrust offense in more than one case, and the instinct to tidy is strongest in exactly the people who know what is in the files.

Instruct employees on their rights, neutrally. They may speak with agents or decline. They may have counsel present. They must not lie — 18 U.S.C. § 1001 makes a false statement to a federal agent a separate felony, and it is charged. The company will provide counsel. Deliver this as information, not as direction, because instructing employees not to cooperate looks like obstruction.

Assess the leniency question immediately. Is there conduct that could be a per se offense? Are we plausibly first? A marker can be sought on limited information and preserves the position while the investigation proceeds. This decision cannot wait for certainty.

Coordinate globally on day one. Other jurisdictions' leniency programs have their own queues, and a company that secures United States leniency and applies in Europe three weeks later has lost the European position.

And begin the privileged internal investigation, with documented Upjohn warnings, understanding that individuals will need separate counsel quickly.

Individuals, joint defense, and the conflict at the center of everything

The company and its executives are not aligned. The company can plead, pay, and continue. An individual faces prison. The moment cooperation is contemplated, those interests diverge sharply, and the divergence is the central management problem in cartel defense.

Separate counsel, early. Every employee who may have participated needs their own lawyer, funded by the company where permitted by charter, bylaws, and applicable law. Providing counsel is not obstruction; failing to provide it while interviewing people under a company banner is a real problem.

Upjohn warnings in every interview, documented — the lawyer represents the company, the privilege belongs to the company, and the company may waive it.

Joint defense agreements are common, useful, and dangerous. They permit coordinated defense and privileged information sharing among defendants with common interests. They also create obligations that constrain a later decision to cooperate, and a company that joins a joint defense group and subsequently seeks to cooperate must extract itself carefully. Enter them in writing, with an express exit provision, and understand that the government views them with interest.

Cooperation credit generally requires providing information about individuals. That is a board-level decision that reallocates risk from the entity to its people, and it should be made deliberately, with the conflicts on the table, rather than drifting into it through a series of productions.

And the executives will be approached directly. Prepare them, provide counsel, and accept that some will cooperate against the company. That is what the program is designed to produce.

Resolution

The leniency applicant signs a conditional leniency letter, cooperates for years, and — if the conditions are met — receives no conviction and no fine. Its exposure is the civil case, mitigated by ACPERA if the cooperation is satisfactory.

Everyone else negotiates a plea. The terms that matter: the charged conduct and its time period; the volume of affected commerce, which drives the fine calculation; the fine and any payment schedule; cooperation obligations; the treatment of individuals, including who is carved out for prosecution; and — critically — the factual admissions, which become § 16(a) prima facie evidence in the civil litigation.

Negotiate the volume of commerce hard. It is the base of the sentencing calculation and the anchor of the civil damages claim, and a definition that includes sales outside the conspiracy's scope or period costs twice.

Watch the carve-outs. A plea that carves out named executives for individual prosecution resolves the company's exposure and leaves its people exposed — which is sometimes unavoidable and always a decision to be made consciously.

Then the collateral consequences: debarment and suspension for government contractors, exclusion from federal programs, foreign regulatory consequences, and — in some industries — licensing effects. These are negotiated with different parts of the government than the plea.

Agreement versus parallel conduct

The line that separates lawful competitive behavior from a felony is agreement, and it is worth stating precisely because business people get it wrong in both directions.

What is lawful. Observing competitors' published prices and responding. Reading trade press. Reaching the same conclusion as a rival about market conditions. Following a price leader without any communication. Declining to bid because a job is unattractive. Independent decisions that happen to be parallel are not a conspiracy, however uniform the outcome.

What is not. Any agreement — express or tacit, written or oral, formal or a nod at a dinner — about price, elements of price, output, customers, territories, bidding, or wages. No writing is required. No enforcement mechanism is required. The agreement need not succeed.

Where the evidence comes from. Because the agreement is rarely documented, the government proves it circumstantially: a pattern of communications, opportunities to conspire, actions against independent self-interest, and market behavior inconsistent with independent decision-making. Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007) requires a civil plaintiff to plead enough to suggest an agreement rather than mere parallel conduct — but a criminal case is built on the communications, and the pattern is what makes them meaningful.

The dangerous middle ground is the invitation. A public statement that appears designed to signal willingness to coordinate, an unreciprocated call to a competitor about pricing, or a message that goes unanswered can be prosecuted as an attempt or investigated as evidence of the conspiracy the government believes exists.

And the practical instruction for a business audience is simpler than the doctrine: whatever the topic, the moment a conversation with a competitor turns to what either of you will charge, bid, produce, or pay, the conversation is over. Say so, leave, and report it. That instruction is enforceable, teachable, and — in the cases that actually get prosecuted — would have been sufficient.

The government contractor overlay

Bid rigging against a public purchaser carries a second set of consequences that frequently exceeds the antitrust exposure.

Additional criminal charges. Bid rigging on federal procurement is routinely charged alongside conspiracy to defraud the United States under 18 U.S.C. § 371, wire and mail fraud, and false statements under 18 U.S.C. § 1001. Prosecutors frequently prefer the fraud counts, which do not require proving a per se antitrust offense and carry their own penalties.

Civil False Claims Act exposure. A bid submitted pursuant to a rigged process supports a claim that the resulting invoices were false — with treble damages and per-claim penalties, and with the possibility that a competitor or an insider brings it as a qui tam relator before the government acts.

Suspension and debarment. The administrative consequence is often the one that ends the business. Suspension can be immediate and does not require a conviction; debarment can extend for years and reaches affiliates. It is administered by agency suspension and debarment officials — a different part of the government from the prosecutors, on a different timeline, and it is negotiated separately through an administrative agreement.

Contract remedies. Termination for default, withholding of payments, and recoupment of amounts paid.

The practical sequencing point. A company negotiating a criminal resolution must engage the suspension and debarment officials in parallel, not afterward. A plea agreed without an administrative agreement can leave the company criminally resolved and commercially finished.

And for a company that bids to public purchasers, the compliance implication is specific. The conduct that gets prosecuted is complementary bidding, bid rotation, and agreements not to bid — arrangements that participants describe as courtesy among firms who see each other on every list. Train the estimating and bidding teams on exactly that, with examples, and require certification with each bid that no communication with a competitor about the bid occurred.

What to tell the client

One: this is a race with one winner. Leniency goes to the first company to report. The second gets a plea. Everything else follows from that.

Two: the decision cannot wait for certainty. A marker can be obtained on limited facts and preserves the position while you investigate. The instinct to know more before escalating is correct everywhere else and wrong here.

Three: individuals go to prison. The company can plead, pay, and continue; the executive cannot. That divergence is the central management problem, and it means separate counsel, early, for everyone potentially involved.

Four: the fine is not capped by the statute. 18 U.S.C. § 3571(d) permits twice the gross gain or twice the gross loss, which in a substantial cartel produces numbers far beyond any statutory maximum.

Five: the civil exposure usually exceeds the fine. Treble damages under 15 U.S.C. § 15, joint and several liability, no contribution after Texas Industries, a government judgment that is prima facie evidence under 15 U.S.C. § 16(a), and parallel state indirect purchaser actions.

Six: ACPERA is why leniency is worth what it is. A successful applicant that cooperates satisfactorily with civil claimants faces single damages on its own sales rather than treble damages for the whole conspiracy. That line is usually the one that decides the analysis, and it is routinely omitted from the first briefing.

Seven: obstruction is the charge that produces the longest sentences. Section 1512 and § 1001 are charged alongside the antitrust count. Preserve everything, instruct affirmatively, and never suggest that anything be tidied.

Eight: coordinate globally on day one. Foreign leniency programs have their own queues, and a company that secures United States leniency and applies in Europe three weeks later has lost Europe.

And nine: the conduct almost never began with anyone deciding to form a cartel. It began at a trade association dinner, in a benchmarking program, or with a call to a former colleague — which is where a compliance program has to aim.

Where the conduct actually comes from

Cartels are rarely formed by executives who set out to break the law. They emerge from ordinary commercial situations, and knowing the patterns is what makes a compliance program useful.

The trade association. Standards committees, statistical programs, and the dinners around industry meetings. Someone mentions that margins are unsustainable; someone agrees; the conversation continues in the bar. Nobody signs anything. This is the single most common origin, and it is why trade association protocols exist.

The benchmarking program. A third-party consultant collects and redistributes data. If the data is current, disaggregated, and company-identified, the program is a mechanism for coordination whatever its stated purpose. Many long-running industry programs sit closer to that line than participants believe.

The bidding market. Contractors who see each other on every list develop an understanding about who "should" win a given job. Complementary bids, bid rotation, and agreements not to bid arise incrementally, and the participants frequently describe them as courtesy rather than conspiracy.

The customer who tells you. A purchaser says a competitor quoted a particular price. Verifying that with the competitor is the step that converts market intelligence into an agreement.

The former colleague. People move between competitors and keep relationships. A call to a friend at a rival about pricing pressure is how many of these begin.

The labor market agreement. No-poach and wage-fixing arrangements among employers, frequently made by human resources leaders who have never considered that the Sherman Act applies to buying labor as much as to selling products. These have been prosecuted criminally.

The acquisition. A company acquires a target that was participating in a cartel and inherits both the conduct and, if it continues it, the liability. Antitrust diligence in an acquisition should ask about competitor contacts and trade association participation, not only about market shares.

The common thread is that none of these begins with a decision to form a cartel. Compliance training built around "do not conspire" misses all of them; training built around "here is the conversation that starts it, and here is what to do when it starts" does not.

Deciding whether to apply for leniency

The decision is made on incomplete facts, under time pressure, with enormous consequences. Structure it.

Question one: is there conduct that could be a per se offense? Not "did we violate the Sherman Act," which is a conclusion, but: is there evidence of communication with competitors about price, bids, customers, territories, or wages? A pattern of contacts plus a suggestive correlation is enough to ask the question.

Question two: are we plausibly first? Consider what would prompt a competitor to report — a change of ownership or general counsel, a departing executive, an unrelated investigation, a compliance audit. If any competitor has recently experienced one of those, assume the clock is running.

Question three: what is the exposure if we are second? Model it: a fine under 18 U.S.C. § 3571(d) calculated on the volume of affected commerce; treble damages under 15 U.S.C. § 15 with joint and several liability and no contribution after Texas Industries; parallel state indirect purchaser exposure; foreign fines and damages; and individual prosecutions.

Question four: what does leniency actually cost? Termination of the conduct with its commercial consequences; years of cooperation; production of documents and witnesses; assistance in prosecuting competitors and, sometimes, former colleagues; and — under ACPERA — the same cooperation extended to the plaintiffs suing you.

Question five: can we qualify? Were we the leader or originator? Did we coerce others? Have we terminated? Can we make a corporate confession rather than individual ones?

Then the arithmetic. In almost every genuine cartel matter, the modeled difference between leniency with ACPERA single damages and a plea with treble joint-and-several exposure is an order of magnitude. That is why the answer is usually to seek the marker.

And the counsel point that matters most. The instinct to investigate before escalating is professionally correct in every other context and wrong here. A marker can be sought on limited facts, it preserves the position, and it can be withdrawn. Advise the client that the decision is about preserving an option, not about admitting anything — because framed that way, it is a decision a board can make in a day.

The compliance program that would have prevented it

Train the people who actually talk to competitors. Sales, procurement, and executives in trade association roles. Not a general online module — a short, concrete session about what may and may not be said, with real examples from the industry.

Trade associations are the recurring venue. Standards committees, benchmarking programs, statistical reporting, and social events around industry meetings are where cartels form, frequently without anyone deciding to form one. Give attendees rules: an agenda reviewed in advance, counsel present or available, no discussion of price, cost, capacity, customers, or bidding, and a documented obligation to leave and report if such a discussion begins.

Watch the information exchange programs. Aggregated, historical, independently administered benchmarking is generally lawful; current, disaggregated, company-identified exchange is not. Many industry programs sit closer to the second than their participants realize.

Screen the labor-market agreements too. No-poach and wage-fixing agreements among employers have been prosecuted criminally, and they are frequently made by human resources departments that have never heard of the Sherman Act.

Build a reporting channel people will use, and respond to what comes in. The cartel that is discovered internally and reported first costs a fraction of the one discovered by a competitor's leniency application.

Audit where the risk is: concentrated industries, bidding markets, industries with a history of cartel enforcement, and any market where prices move in unison.

And make the escalation path fast. The single most valuable feature of a compliance program in this area is that a general counsel who learns of a possible cartel on Tuesday can reach experienced counsel on Tuesday — because in a race with one winner, days matter.