Game theory turns oligopoly interdependence into a table of payoffs
At A-level, oligopoly means interdependence: each firm's best move depends on what its rivals do. Game theory makes that exact. It studies situations in which players make decisions and then receive payoffs that depend on what the other players decide. Take two firms choosing whether to advertise. Advertising costs money and mostly wins customers from the rival, so if both advertise the gains cancel out and only the costs remain.
Two firms can advertise or not. Advertising costs money and mainly steals customers from the rival. What do you think they end up doing?
Each firm earns more by advertising whatever the other does, so both advertise, even though each would earn more if neither did. Knowing that the pair would do better does not change what each one does alone.
Advertising is the best reply whatever the rival does, so both advertise
Take Firm A's side. If B advertises, A earns £3 million by advertising and £2 million by not, so A advertises. If B does not advertise, A earns £6 million by advertising and £5 million by not, so A advertises again. Advertising is A's dominant strategy: the one it will pursue regardless of B's decision. B's position is the mirror image, so both advertise and each earns £3 million.
Check yourself
In the advertising game, why does Firm A advertise even when it is sure Firm B will not?
Both firms would do better holding back, yet neither can get there alone
If neither advertised, each would earn £5 million instead of £3 million. Both prefer that outcome, but neither can reach it alone: a firm that holds back while its rival advertises falls to £2 million. This is a prisoner's dilemma, where the gains from cooperation are larger than the rewards from pursuing self-interest. The outcome in which both advertise is Pareto-inferior: another outcome makes both players better off and leaves nobody worse off.
Check yourself
Which statement about the outcome in which both firms advertise is true?
An equilibrium needs no dominant strategy, and a game can have two
Both advertising is a Nash equilibrium: given what the other does, neither firm can gain by changing its own choice. Many games have no dominant strategy. Two carmakers choosing a charging socket each want to match the other. Both picking socket X is an equilibrium, and so is both picking socket Y, because neither gains by switching alone. The model says where play can come to rest, not which resting point is reached: history, convention or a regulator picks.
Check yourself
Two firms each choose to expand or hold. In each cell Firm A's profit comes first, then Firm B's, in £ million. Which cell is the Nash equilibrium?
At P both expand. B would switch to holding and earn £3 million instead of £2 million, so P is not an equilibrium.
Yes. Expanding is A's dominant strategy, and against an expanding A, B's best reply is to hold, earning £3 million rather than £2 million. Neither gains by moving alone.
At R, A holds while B expands. A would switch to expanding and earn £4 million instead of £3 million, so R is not an equilibrium.
At S both hold. A would switch to expanding and earn £7 million instead of £6 million, so S is not an equilibrium.
- Nash equilibrium
- A Nash equilibrium is a set of strategies, one for each player, in which no player can do better by changing strategy while the others keep theirs.
- Dominant strategy
- A dominant strategy is a choice that gives a player a higher payoff than any alternative, whatever the other players choose.
Four ideas for reading a two-player game
Can you name the four ideas you need to read a two-player game?
A payoff matrix is a table giving each player's payoff for every combination of the players' strategies.
Can you think of an example?
The advertising game's matrix has four cells. The cell where both advertise reads 3, 3: each firm earns £3 million, with the row player's payoff written first.
A prisoner's dilemma is a game in which the gains from cooperation are larger than the rewards from pursuing self-interest, yet each player's self-interest wins.
Can you think of an example?
Two suspects are questioned apart. If both stay silent each gets two years; if both confess, five each; one confessing alone gets one year and the silent partner eight. Confessing is dominant, so both serve five.
A Pareto-inferior outcome is one where some other available outcome would make at least one player better off and nobody worse off.
Can you think of an example?
In the advertising game both advertising gives each firm £3 million. Both holding back gives each £5 million, so the equilibrium is Pareto-inferior.
A coordination game is a game in which each player does best by making the same choice as the others, so it has more than one Nash equilibrium.
Can you think of an example?
Two friends separated at a festival both want to meet, at the main stage or at the food tents. Either place works if both go there; the danger is that each waits at a different one.
Doesn't a Nash equilibrium mean the best outcome for everyone?
No. A Nash equilibrium is a resting point: no player can do better by changing alone. The advertising game shows the gap. Both advertising is the equilibrium, yet both holding back would give each firm £5 million instead of £3 million. The equilibrium test moves one player at a time; the efficiency test asks whether some other outcome would make everyone better off together. A game's equilibrium can pass that test or fail it, and a game with two equilibria can have one that is better for both players than the other.
To find the equilibrium in a two-by-two matrix, take each player in turn. For each of the rival's strategies, mark the player's best reply. A cell where both payoffs are marked is a Nash equilibrium. Then ask whether any other cell makes both players better off. In an interview, say the method aloud as you go: the interviewer is listening to how you reason, not only to the answer.
Exam question
Two petrol stations face each other across a road. Both would earn more if both charged a high price, yet they keep undercutting each other. Is that an equilibrium, and is it a good outcome? [4]
It is a prisoner's dilemma. Whatever the rival charges, each station does better charging low: against a high price it takes the rival's customers, and against a low price it avoids losing its own. So both charging low is a Nash equilibrium, and neither gains by raising its price alone. For the two stations it is Pareto-inferior, since both prefer the high-price outcome; for drivers it is good, so whether it is a good outcome depends on whose welfare counts. The model assumes a one-off game. Two stations that face each other every day can punish undercutting tomorrow, and may hold prices up without ever talking.
A strong answer names the game, shows each player's best reply, and states the equilibrium. It then separates the players' welfare from society's, and ends on the assumption that breaks the prediction: this game is repeated daily, which can sustain high prices.
Exam question
Britain drives on the left and most of Europe on the right. Is either side better, and what would it take for Britain to switch? [4]
Choosing a side of the road is a coordination game. Each driver does best by doing what everyone else does, so everyone on the left and everyone on the right are both Nash equilibria, and neither is better in itself. History picked Britain's. No single driver gains by switching, so a change has to happen for everyone at once, set by law, as Sweden did on one day in 1967. The assumption that the two equilibria are equally good breaks once costs count: a British switch would mean new signs, junctions and buses, so the equilibrium we are in is the cheaper one to stay in.
A strong answer names the coordination game and its two equilibria, explains why no individual moves, and says what can move everyone together. It then questions the assumption that the equilibria are equally good once switching costs are counted.
Exam question
Two drivers meet at a bridge only wide enough for one car. Draw me the game. Where does it come to rest, and who wins? [4]
Each driver can go or wait. I would draw a two-by-two payoff matrix: one driver's choices as rows, the other's as columns, both payoffs in each cell. If both go they crash, the worst outcome; if both wait nobody crosses; if one goes and the other waits, the one who goes does best and the one who waits still crosses later. Marking each player's best reply to each of the other's choices gives two Nash equilibria, one goes and the other waits, and neither player has a dominant strategy. The matrix assumes both move at once; if one can commit first, say by driving on to the bridge, that driver wins. Otherwise the model cannot say which equilibrium is reached.
A strong answer sets out players, strategies and payoffs in a matrix, finds the equilibria by marking best replies, and says there is no dominant strategy. It ends on the assumption of simultaneous moves, which commitment breaks.
A game sets out players, strategies and payoffs, often in a payoff matrix. A dominant strategy is best whatever the rival does. A Nash equilibrium is a set of choices where no player gains by changing alone. In a prisoner's dilemma each player's dominant strategy leads to an equilibrium that is Pareto-inferior: both would do better cooperating. A coordination game has more than one equilibrium, and history or rules decide which one is reached.