行为金融学 讲座7.pdf

14.127 Lecture 7 Xavier Gabaix March 18, 2004 1 Learning in games • Drew Fudenberg and David Levine, The Theory of Learning in Games 1.1 Fictitious play • Let γi denotes frequencies of i’s opponents play t i number of times s−i was played till now γt (s−i) = t • Player i plays the best response BR � i � γt • Big concerns — Asymptotic behavior do we converge or do we cycle? — If we converge, then to what subset of Nash equilibria? • Caveat. Empirical distribution need not converge 1.2 Replicator dynamics i i � � • Call θ s = fraction of players of type i who play s . t i • Postulate dynamics — In discrete time i � i i � i � � −i� i� θ = θ (s ) , ..., θ (sn) = θ + λ BR θ − θ t+1 t+1 1 t+1 t t t — In continuous time � � � � d i −i i θt+1 = λ BR θt − θt dt • Then analyze the dynamics chaos, cycles, fixed points 1.3 Experience weighted attraction model, EWA • Camerer­Ho, Econometrica 1999 • Denote Nt =number of “observation equivalent” past responses such that Nt+1 = ρNt + 1 • Denote — sij − strategy j of player i — si (t) — strategy that i