L21 - Collusion Flashcards
What are the conditions necessary for oligopolists to collude?
C(1) –> Sellers must be aware of each other’s strategies
- If a firm is not aware, how can it punish a deviation?
- Firms may form cartels to check that each other are conforming to the agreement
C(2) –> Sellers must interact repeatedly
- Incentive “to deviate” must be counteracted by a credible long-term punishment
- Punishments usually mean a price war (period of low prices)
What type of well known game are duopolies effectively competing in?
- They are effectively in a prisoner’s dilemma –> like Bertrand’s model of Oligopolies cause of the nature of the game both players would defect or use the minimum level of prices
- in the Cournots model of oligopoly if they both chose to restrict their output –> they could both enjoy the monopoly profits
- but in the short term –> both firms have a short term incentive to produce more that it said it would
- in Bertrand’s model if they both colluded to raise their prices they both would be better off at the monopolist price –> but if they were to do so they would have the short term incentive to undercut each other to maximise their profit
How do you solve a finitely repeated prisoner’s dilemma?
- Notice that the game is a prisoner’s dilemma
The “one-shot” Nash equilibrium (if game played once) is (deviate, deviate)
Is (cooperate, cooperate) a Nash equilibrium of the repeated game - NAÏVE: could be, if firms threaten to play (deviate, deviate) benefit from deviating in period 1 can be eliminated in the 4 remaining periods…
- CORRECT: use backward induction to find a Nash equilibrium in each subgame
- looking from the final game –> as even in the last game regardless how they played before both firms still have an incentive to deviate
- As this incentive can appear in all previous games their is a sub game Nash equilibria to (deviate, deviate) in all periods
How do you solve an infinitely repeated Prisoner’s Dilemma?
- In an infinitely repeated game, backwards induction cannot be used
(Same is true for an indefinitely repeated game)
Instead we ask whether a certain strategy is a Nash equilibrium - Grim trigger strategy –> Suppose each firm plays cooperate, as long as the other has always done so
But if a player chooses deviates: –> - firms revert to playing the one-shot Nash equilibrium (deviate, deviate) FOREVER
When is a Grim Trigger Strategy defined as a Nash equilibrium?
Does this strategy define a Nash equilibrium?
Yes, if no incentive to deviate from it
To analyse this, we first need to be able to construct the firms’ payoffs:
- i) short-term benefit: playing deviate when other plays cooperate
- ii) Long-term punishment: both playing deviate instead of cooperate
How do you calculate the short-term benefits in the Grim Trigger Strategy?
To calculate the short-term benefits you must work out the difference between each firms (Deviate, Cooperate) and their (Cooperate,Cooperate)
How do you calculate the long-term punishment in the Grim Trigger Strategy?
- First find the difference between (Cooperate, Cooperate) and (Deviate, Deviate) –> for the first period
- however, firms are punished forever
- Therefore we need to get future payoffs in terms of present value
- Therefore to calculate the long-term punishment you divide the original answer by the discount rate r
How do you calculate the present value of future payoffs?
present value = X x 𝛿 𝛿 = 1/(1+r) - where X is the value of the money in the base year - where 𝛿 is the future payoff and - r is the interest rate
How do you calculate Discounted Payoffs?
- Suppose a player expects to receive a payoff of in every future period
- In the present period, the player values the payoff of:
- period 1 = X x 𝛿 =X𝛿
- period 2 = X x 𝛿 x 𝛿 =X𝛿^2
The expected present discounted value of this stream of payoffs is:
- X(𝛿+𝛿^2+𝛿^3+…+𝛿^n…)~= X𝛿/(1-𝛿)
If you substitute in 𝛿=1/(1+r), then X𝛿/(1-𝛿) =X/r
How do you solve an infinitely repeated prisoner’s dilemma?
- By comparing short-term benefits and the long-term punishment
- he strategy (cooperate, cooperate) forever is a Nash equilibrium if:
- short-term benefit is less than long-term punishment
- which is true if interest rate is sufficiently low