A one-off prisoner's dilemma predicts that every cartel collapses
A-level oligopoly theory says firms are tempted to collude and tempted to cheat. Played once, the cartel game is a prisoner's dilemma: whatever its rival does, each firm earns more by cheating on the agreed price, so both cheat and each earns £3 million instead of £5 million. The way out of a prisoner's dilemma is to find a way to penalise those who do not cooperate. A cartel cannot sign a contract a court will enforce, so the penalty has to come from somewhere else: the future.
A known last round unravels cooperation all the way back to the first
Suppose the two firms know they will play exactly ten rounds, perhaps because a licence runs out. In round ten nothing follows, so it is a one-off game and both cheat. In round nine, both know round ten will be cheating whatever happens now, so good behaviour in round nine buys nothing, and both cheat. The same reasoning runs back to round one. Working from the end of a game to the start like this is backward induction, and with a known finite horizon it predicts cheating from the start.
Check yourself
Two firms know their cartel game will be played for exactly five rounds. What does backward induction predict for round one?
Now nobody knows which round will be the last: after each one there is a good chance of another. Can the cartel hold together?
With no known last round there is no end to reason back from. Cheating wins £1 million once but can cost £2 million in every later round, which deters it when another round is likely enough, but not when the future counts for little.
An uncertain end lets the threat of punishment keep firms honest
If the game might go on after every round, each firm can play a trigger strategy: hold the price while the rival does, and cheat for good once the rival cheats. Cheating now gains £1 million, taking profit from £5 million to £6 million, but every later round then pays £3 million instead of £5 million. With a chance p of another round each time, the expected loss is £2 million times p/(1 − p), where p/(1 − p) is the expected number of further rounds: p + p² + p³ and so on. It outweighs the £1 million gain once p is above one third, so a likely enough future sustains the cartel.
Check yourself
Each round, a cheat gains £1 million and then loses £2 million in every later round. Which chance of another round is too low for the threat to stop cheating?
Cartels break when cheating pays more, is seen later, or can be confessed
Each firm faces a private temptation to produce a little more while counting on the others to hold back. More members make that temptation larger, because a small firm that cheats wins a big share of the market, and make cheating harder to spot. Uncertain demand hides cheating, because a fall in sales could be a rival's price cut or a weak market. Slow detection delays the punishment. Leniency programmes, such as the Competition and Markets Authority's, offer the first member to confess immunity from fines, which turns the cartel itself into a new prisoner's dilemma.
- Repeated game
- A repeated game is a game in which the same players meet in the same situation again and again, so each can respond to what others did before.
- One-shot game
- A one-shot game is a game the players play only once, so no choice made in it can be rewarded or punished later.
Ideas that explain when cartels hold and when they break
Can you name the four ideas that explain when a cartel holds and when it breaks?
Backward induction is solving a game by working out the last move first, then the move before it, back to the start.
Can you think of an example?
Two firms know a shared contract ends after ten rounds. Each reasons that round ten will be cheating, so round nine is too, and so on until neither holds the price at all.
A trigger strategy is a plan to cooperate until a rival breaks the agreement and then to stop cooperating, as punishment.
Can you think of an example?
A ferry operator on a route shared with one rival keeps its fares high, but promises itself that if the rival ever undercuts, it will cut its own fares for good.
Tit-for-tat is a strategy of cooperating in the first round and then copying whatever the rival did in the round before.
Can you think of an example?
In Robert Axelrod's well-known computer tournaments, this simple rule beat far cleverer entries: it was never first to cheat, punished at once, and forgave as soon as the rival returned to cooperating.
A leniency programme is a competition authority's offer of reduced or no fines to a cartel member that reports the cartel and helps prove it.
Can you think of an example?
Under the Competition and Markets Authority's programme, the first cartel member to come forward can escape a fine entirely, so every member fears being second to the door.
If the game is repeated, won't the firms always cooperate?
No. Repetition makes cooperation possible, not certain. It fails if the last round is known, because backward induction unravels it. It fails if the firms care too little about the future, or think another round too unlikely, because then the threat of punishment is too weak. And even when cooperation could last, both firms cheating in every round is still an equilibrium of the repeated game: if each expects the other to cheat, cheating is each one's best reply. Repetition opens a door; it does not push anyone through it.
State the one-shot prediction first, then say what repetition changes: a future in which cheating can be punished. Name the condition, a high enough chance of meeting again, and then the forces that break it: many firms, uncertain demand, slow detection, leniency. An interviewer rewards an answer that says when the argument fails, not only when it works.
Exam question
Why do cartels hold, and what would you do to break one? [6]
Played once, a cartel is a prisoner's dilemma and each member cheats. Cartels hold because the game repeats with no known end: each firm can threaten to cut its price if a rival cheats, as a trigger strategy or tit-for-tat does, and when another round is likely enough, a one-off gain from cheating is outweighed by the profit lost in every round after. To break one, I would attack that threat. A leniency programme rewards the first firm to confess, so each fears being second. Encouraging entry adds members and makes cheating harder to police, and letting large buyers negotiate secret discounts makes cheating harder to spot and slower to punish. The argument assumes firms can see cheating; if prices are secret, the cartel may already be weaker than it looks.
A strong answer states the one-shot model, applies repetition and the condition for cooperation, and then uses that condition to choose the policy. It ends on an assumption, here that cheating is visible, which could change the answer.
Exam question
Two ferry firms share a route under licences that both expire in exactly three years. Should we expect them to keep fares high until then? [4]
Backward induction says no. In the final period there is no future to protect, so each cuts its fare; knowing that, neither gains by holding fares up in the period before, and the argument runs back to today. So collusion should unravel from the start. The prediction assumes both firms are sure the licences end and reason all the way back. If either might win a renewal, the end is uncertain, the future matters again, and high fares could last.
A strong answer applies backward induction step by step, from the last period back, and then questions its assumptions: a known end and flawless reasoning by both firms. A chance of renewal removes the known last round.
Exam question
Why might a price-fixing cartel fall apart in a recession? [4]
A cartel survives while each member believes cheating will be seen and punished. In a recession sales fall for every firm, so a member losing sales cannot tell whether a rival has secretly cut prices or the market has weakened. Punishment becomes a guess, the threat loses force, and firms with spare capacity are tempted to cut prices to fill it. If firms instead expect the slump to be short, the future still matters, and they may hold on through it.
A strong answer names the mechanism, uncertain demand hiding cheating, links it to the weakened threat of punishment, and adds the pull of spare capacity. It ends on the assumption, how long the slump lasts, that decides whether the future still disciplines the members.
Played once, the cartel game is a prisoner's dilemma and each member cheats. With a known last round, backward induction unravels cooperation back to the first round. With no known end, trigger strategies and tit-for-tat can sustain it, if another round is likely enough that future losses outweigh today's gain from cheating. Cartels break when there are many members, when demand is uncertain, when cheating is seen late, and when leniency rewards the first to confess.