Representing games Lectures in Game Theory Spring 2011, Part 1 10.12.2010 G.B. Asheim, ECON4240-1 1 Game theory studies multi-person decision problems, and analyzes agents that are rational (have well-defined preferences) reason strategically (take into account their knowledge and beliefs about what others do) Applications Industrial organization (incl. oligopoly theory) Bargaining and auction theory Labor market and financial economics Macro economics International economics 10.12.2010 G.B. Asheim, ECON4240-1 2 Classification Non-cooperative game theory Studies the outcome of individual actions in a situation without external enforcement. Contract and cooperative game theory. Studies the outcome of joint actions in a situation with external enforcement. Seeks to develop solution concepts, prescriptions or predictions about the outcomes of games 10.12.2010 G.B. Asheim, ECON4240-1 3 1 Major tensions of strategic interaction The conflict between individual and group interests. S Strategic i uncertainty. i The specter of inefficient coordination 10.12.2010 4 G.B. Asheim, ECON4240-1 Representing games A game can be analyzed both in and the normal form. the extensive form Stay out Entrant 0, 2 Fight Enter Incumbent Accept 10.12.2010 Incumbent Accept Fight Enter 1, - 1, 1 -1 1 1 - 1, 1 -1 Entrant Stay out 0, 2 0, 2 1, 1 5 G.B. Asheim, ECON4240-1 The extensive form specifies Players: {1, ... , i, ... , n} What actions an acting player can choose among, what an acting player knows. Payoff for each of the players as a function off the h actions that h are realized. l d 2 H 1 Decision node (initial node) 10.12.2010 H L 1, 2 L 1, 1 2 H 2, 1 Decision nodes L Payoffs assigned to players 1 and 2 at terminal nodes 0, 0 G.B. Asheim, ECON4240-1 6 2 Information sets Dynamic 2 H game H 1 L 2 H L L 1, 2 1, 1 2, 1 H 1 L 0, 0 Static game 2 H 1, 2 L 2 H 1, 1 2, 1 L 0, 0 Definition : An information set for player i is a set of decision nodes that satisfies at all decision nodes in the info. set, player i moves, when the info. set is reached, i does not know which of the set' s decision nodes has been reached. 10.12.2010 7 G.B. Asheim, ECON4240-1 Strategy Definition : A strategy for player i is a plan of action that, for each of i' s info. sets, specifies a feasible action. 2 H 1, 2 2 H 1, 2 H H 1 1 L 1, 1 L 1, 1 2 H 2, 1 2 H 2, 2 1 L L 0, 0 L 0, 0 L HH HL LH LL H 1, 2 1, 2 1, 1 1, 1 H H 1, 2 L 2, 1 0, 0 2, 1 0, 0 L 2, 1 0, 0 10.12.2010 L 1, 1 G.B. Asheim, ECON4240-1 8 The normal form specifies Players: {1, ... , i, ... , n} For each player, a strategy set: Si For each player, a payoff function: ui G ( S1 ,, Sn ; u1 ,, un ) Payoff for each player i depends on the strategy profile : ui ( s1 , , sn ) ui ( si , si ) where, for all j , s j S j , and where we write si ( s1 , , si 1 , si 1 , , sn ) 10.12.2010 G.B. Asheim, ECON4240-1 9 3 Classic normal form games Matching Pennies Coordination Prisoners’ Dilemma Pareto Coordination Battle of the Sexes Stag Hunt Hawk-Dove/Chicken Pigs 10.12.2010 G.B. Asheim, ECON4240-1 10 Application of the normal form Stay out Entrant Enter Incumbent Accept Fight Enter 1, 1 - 1, - 1 Fight - 1, - 1 Entrant Stay out 0, 2 0, 2 0, 2 Incumbent Accept 1, 1 Fight Stay out Accept Entrant Enter 10.12.2010 0, 2 Fight - 1, - 1 Incumbent Accept Does a normal form represent p dynay mic interaction in an adequate way? Or should a normal form only be used for the analysis of static interaction? Notice that different extensive forms may have the same normal form. 0, 2 1, 1 G.B. Asheim, ECON4240-1 11 Beliefs, mixed strategies, and exp. utility Strategic uncertainty (uncertainty about opponent choice) leads to beliefs about opponent behavior. If payoffs are von Neumann-Morgenstern utility, then: Expected p payoff p y for p player y i: ui ( si , i ) s i S i i ( si )ui ( si , si ) where i is a prob. distr. over opponent str. profiles. A mixed strategy is a probability distribution over the player’s own strategies. Interpretation: (a) The player randomizes. (b) Her opponents are uncertain. 10.12.2010 G.B. Asheim, ECON4240-1 12 4