rationalizability in games with a continuum of players
play

Rationalizability in Games with a Continuum of Players Pedro - PowerPoint PPT Presentation

Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni Instituto Milenio Sistemas Complejos de Ingenier a


  1. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework We have a game in which: The set of players I is the unit interval of R , I ≡ [0 , 1]. For each player i ∈ I a set of strategies s ( i ) ∈ S ( i ) ≡ S ⊆ R n , ∀ i ∈ I . The payoff functions u ( i )( · ) depend on the other players’ � strategies through the integral of the strategy profile s ( i ) di. There are functions u ( i, · ) : S × co { S } → R such that: � � � u ( i )( s ( i ) , s ) ≡ u i, s ( i ) , s ( i ) di S is compact. A strategy profile is a measurable function s : I → S . s ∈ S I . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  2. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework Definition 1 A Nash Equilibrium is a strategy profile s ∗ ∈ S I such that, � � � � � � s ∗ ( i ) , s ∗ s ∗ for λ -a.e. i ∈ I, u ( i ) ≥ u ( i ) y, ∀ y ∈ S (1) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  3. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework We call A ≡ co { S } . Optimal strategy correspondence B ( i, · ) : A ⇒ S : B ( i, a ) := argmax y ∈ S { u ( i, y, a ) } . (2) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  4. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework We call A ≡ co { S } . Optimal strategy correspondence B ( i, · ) : A ⇒ S : B ( i, a ) := argmax y ∈ S { u ( i, y, a ) } . (2) Best reply to forecasts correspondence B ( i, · ) : P ( A ) ⇒ S : B ( i, µ ) := argmax y ∈ S E µ [ u ( i, y, a )] . (3) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  5. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework We call A ≡ co { S } . Optimal strategy correspondence B ( i, · ) : A ⇒ S : B ( i, a ) := argmax y ∈ S { u ( i, y, a ) } . (2) Best reply to forecasts correspondence B ( i, · ) : P ( A ) ⇒ S : B ( i, µ ) := argmax y ∈ S E µ [ u ( i, y, a )] . (3) � We denote Γ( a ) = I B ( i, a ) di. Equivalently, an equilibrium is a point a ∗ ∈ A such that: � � a ∗ ∈ Γ( a ∗ ) ≡ B ( i, a ∗ ) di ≡ B ( i, δ a ∗ ) di (4) I I Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  6. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Framework U S ×A : the space of real valued continuous functions defined on S × A , endowed with the supremum norm. u : i ∈ I → u ( i ) ∈ U S ×A u ( i ) : S × A → R . HM : The mapping u is measurable. Theorem 2 (Rath, 1992) Every game u has a (pure strategy) Nash Equilibrium. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  7. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Guesnerie (1992) Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  8. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Guesnerie (1992) farmers [0 , 1] ≡ I . cost function c i : R + → R . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  9. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Guesnerie (1992) farmers [0 , 1] ≡ I . cost function c i : R + → R . �� � price p = P q ( i ) di . payoff u ( i, q ( i ) , p ) ≡ pq ( i ) − c i ( q ( i )). Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  10. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Guesnerie (1992) farmers [0 , 1] ≡ I . cost function c i : R + → R . �� � price p = P q ( i ) di . payoff u ( i, q ( i ) , p ) ≡ pq ( i ) − c i ( q ( i )). for a given forecast µ over the price, E µ [ pq ( i ) − c i ( q ( i ))] ≡ E µ [ p ] q ( i ) − c i ( q ( i )) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  11. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Guesnerie (1992) farmers [0 , 1] ≡ I . cost function c i : R + → R . �� � price p = P q ( i ) di . payoff u ( i, q ( i ) , p ) ≡ pq ( i ) − c i ( q ( i )). for a given forecast µ over the price, E µ [ pq ( i ) − c i ( q ( i ))] ≡ E µ [ p ] q ( i ) − c i ( q ( i )) B ( i, p ) ≡ Supply( i )( p ), B ( i, µ ) ≡ Supply( i )( E µ [ p ]) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  12. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  13. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Definition 3 (Bernheim, 1984) s i is a Rationalizable Strategy for player i if there exists some consistent system of beliefs for this player and some subjective product probability measure over the set of strategy profiles of the opponents, that gives zero probability to actions of the opponents of i that are ruled out by this system of beliefs and such that the strategy s i maximizes expected payoff with respect to this probability measure. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  14. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Proposition 4 (Bernheim, 1984) In a game with a finite number of players, compact strategy sets and continuous payoff functions, the set of Rationalizable Strategy Profiles: Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  15. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Proposition 4 (Bernheim, 1984) In a game with a finite number of players, compact strategy sets and continuous payoff functions, the set of Rationalizable Strategy Profiles: (i) is the result of the iterative and independent elimination of strategies that are not best-replies to any forecast considering all of the remaining strategy profiles Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  16. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Proposition 4 (Bernheim, 1984) In a game with a finite number of players, compact strategy sets and continuous payoff functions, the set of Rationalizable Strategy Profiles: (i) is the result of the iterative and independent elimination of strategies that are not best-replies to any forecast considering all of the remaining strategy profiles (ii) is the largest set that satisfies being a fixed point of the process of elimination of strategies. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  17. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  18. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States In the setting of Rath (1992), forecasts over the set of states . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  19. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States In the setting of Rath (1992), forecasts over the set of states . If CK is a subset X ⊆ A Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  20. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States In the setting of Rath (1992), forecasts over the set of states . If CK is a subset X ⊆ A � ∀ i ∈ I , s ( i ) ∈ B ( i, X ) ≡ � B ( i, a ) a ∈ X Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  21. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States In the setting of Rath (1992), forecasts over the set of states . If CK is a subset X ⊆ A � ∀ i ∈ I , s ( i ) ∈ B ( i, X ) ≡ � B ( i, a ) a ∈ X � � a = s ( i ) di ∈ B ( i, X ) di. � Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  22. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Define ˜ Pr : P ( A ) → P ( A ) by � ˜ Pr ( X ) ≡ B ( i, X ) di I Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  23. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Define ˜ Pr : P ( A ) → P ( A ) by � ˜ Pr ( X ) ≡ B ( i, X ) di I Define: � � 0 ( A ) ≡ A t ( A ) ≡ ˜ t − 1 ( A ) ˜ ˜ ˜ Pr Pr Pr Pr Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  24. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Define ˜ Pr : P ( A ) → P ( A ) by � ˜ Pr ( X ) ≡ B ( i, X ) di I Define: � � 0 ( A ) ≡ A t ( A ) ≡ ˜ t − 1 ( A ) ˜ ˜ ˜ Pr Pr Pr Pr Point-Rationalizable set, P A , must satisfy: + ∞ � t ( A ) =: P ′ ˜ P A ⊆ (5) Pr A . t =0 P A ≡ ˜ Pr ( P A ) . (6) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  25. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Definition 5 The set of Point-Rationalizable States , P A , is the maximal subset X ⊆ A that satisfies the condition: X ≡ ˜ Pr ( X ) . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  26. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Theorem 6 Let us write ∞ � t ( A ) . ˜ P ′ A := Pr t =0 The set of Point-Rationalizable States of a game u satisfies P A ≡ P ′ A Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  27. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 S ≡ [0 , 1] Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  28. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  29. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . ˜ Pr ( X ) ≡ co { B ( X ) } , Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  30. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  31. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . P ′ A ≡ [0 , ¯ a ]. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  32. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . P ′ A ≡ [0 , ¯ a ]. ˜ Pr ( P ′ A ) ≡ co { B ( P ′ A ) } Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  33. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . P ′ A ≡ [0 , ¯ a ]. ˜ Pr ( P ′ A ) ≡ co { B ( P ′ A ) } ≡ co { B ([0 , ¯ a ]) } Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  34. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . P ′ A ≡ [0 , ¯ a ]. ˜ Pr ( P ′ A ) ≡ co { B ( P ′ a ]) } ≡ co {{ a ∗ }} ≡ { a ∗ } � P ′ A ) } ≡ co { B ([0 , ¯ A . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  35. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Example 1 u ( i ) ≡ u : [0 , 1] 2 → R for all i ∈ I , is such that S ≡ [0 , 1] � a ∗ if a ≤ ¯ a , B ( a ) = { 0 , ¯ a (1 − α ) + aα } if a > ¯ a , a, α ∈ ]0 , 1[ . a ∗ < ¯ where a ∗ , ¯ a . t ( A ) ≡ � 0 , a t � ˜ ˜ Pr ( X ) ≡ co { B ( X ) } , Pr where { a t } + ∞ t =0 satisfies a t = ¯ a (1 − α t ) + α t ց ¯ a . P ′ A ≡ [0 , ¯ a ]. ˜ Pr ( P ′ A ) ≡ co { B ( P ′ a ]) } ≡ co {{ a ∗ }} ≡ { a ∗ } � P ′ A ) } ≡ co { B ([0 , ¯ A . So P ′ A � = P A , which is in fact P A ≡ { a ∗ } . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  36. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ✻ ✻ 1 a 1 ✟✟✟✟ ✛ a 2 . .. .. . . . . . a 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . ❜ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . . . . . . . . . . . . . . . . . . . . . . . . B ( A ) . . . . . . . � . . . . . . . . . . . . . . . . . . . . � ✠ . . . . . . . ✛ . . . . . . . . . . . . . . . . P ′ . . . . . . . . . . r r . . . . . . . . . . A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ✌ . . . . . . . . . . . . . . ✲ . . . . . . . . . . . . . . . . . . . . . r r 0 ¯ 1 a � �� � { a ∗ } ≡ P A Figure: The set of Point-Rationalizable States is not the set P ′ A . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  37. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Lemma 7 In a game u , for a closed set X ⊆ A the correspondence i ⇒ B ( i, X ) is measurable and has compact values. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  38. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Lemma 7 In a game u , for a closed set X ⊆ A the correspondence i ⇒ B ( i, X ) is measurable and has compact values. Existence : i ⇒ B ( i, { a } ) ≡ Γ( a ) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  39. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Lemma 7 In a game u , for a closed set X ⊆ A the correspondence i ⇒ B ( i, X ) is measurable and has compact values. Existence : i ⇒ B ( i, { a } ) ≡ Γ( a ) (Point-)Rationalizability : i ⇒ B ( i, X ) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  40. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  41. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  42. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A A ⊆ ˜ Prove that P ′ Pr ( P ′ A ) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  43. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A A ⊆ ˜ Prove that P ′ Pr ( P ′ A ) Consider the sequence F t : I ⇒ S , t ≥ 0, of correspondences: Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  44. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A A ⊆ ˜ Prove that P ′ Pr ( P ′ A ) Consider the sequence F t : I ⇒ S , t ≥ 0, of correspondences: F 0 ( i ) := S ∀ i ∈ I Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  45. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A A ⊆ ˜ Prove that P ′ Pr ( P ′ A ) Consider the sequence F t : I ⇒ S , t ≥ 0, of correspondences: F 0 ( i ) := S ∀ i ∈ I � � t − 1 ( A ) i, ˜ F t ( i ) := B ∀ i ∈ I Pr t ≥ 1 Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  46. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Proof of Theorem 6 If X ≡ ˜ Pr ( X ) then X ⊆ P ′ A , so P A ⊆ P ′ A Moreover it is always true that ˜ Pr ( P ′ A ) ⊆ P ′ A A ⊆ ˜ Prove that P ′ Pr ( P ′ A ) Consider the sequence F t : I ⇒ S , t ≥ 0, of correspondences: F 0 ( i ) := S ∀ i ∈ I � � t − 1 ( A ) i, ˜ F t ( i ) := B ∀ i ∈ I Pr t ≥ 1 t ( A ) ≡ � we have that ˜ I F t ( i ) di . Pr Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  47. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  48. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  49. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Lemma 7: F 1 is measurable and compact valued. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  50. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Lemma 7: F 1 is measurable and compact valued. t − 1 ( A ) ≡ � I F t − 1 is non By induction over t , for all t ≥ 1, ˜ Pr empty, convex and compact. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  51. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Lemma 7: F 1 is measurable and compact valued. t − 1 ( A ) ≡ � I F t − 1 is non By induction over t , for all t ≥ 1, ˜ Pr empty, convex and compact. F t is measurable and non-empty compact valued. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  52. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Lemma 7: F 1 is measurable and compact valued. t − 1 ( A ) ≡ � I F t − 1 is non By induction over t , for all t ≥ 1, ˜ Pr empty, convex and compact. F t is measurable and non-empty compact valued. Define F : I ⇒ S as the point-wise lim sup of F t : � p-lim sup t F t � F t ( i ) F ( i ) := ( i ) ≡ lim sup t Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  53. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States ∀ i ∈ I the mappings B ( i, · ) : A ⇒ S are u.s.c. and B ( i, X ) is compact for any compact subset X ⊆ A . � I F 0 , is non-empty and compact. From Aumann (1965) A ≡ Lemma 7: F 1 is measurable and compact valued. t − 1 ( A ) ≡ � I F t − 1 is non By induction over t , for all t ≥ 1, ˜ Pr empty, convex and compact. F t is measurable and non-empty compact valued. Define F : I ⇒ S as the point-wise lim sup of F t : � p-lim sup t F t � F t ( i ) F ( i ) := ( i ) ≡ lim sup t From Rockafellar and Wets (1998), F is measurable and compact valued. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  54. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States � I F t for all t ≥ 0. Take a ∈ P ′ A . That is, a ∈ Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  55. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States � I F t for all t ≥ 0. Take a ∈ P ′ A . That is, a ∈ � I s t ∀ t ≥ 0. We get a sequence { s t } t ∈ N , such that a = Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  56. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States � I F t for all t ≥ 0. Take a ∈ P ′ A . That is, a ∈ � I s t ∀ t ≥ 0. We get a sequence { s t } t ∈ N , such that a = � Lemma proved in Aumann (1976) gives that a ∈ I F . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  57. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States � I F t for all t ≥ 0. Take a ∈ P ′ A . That is, a ∈ � I s t ∀ t ≥ 0. We get a sequence { s t } t ∈ N , such that a = � Lemma proved in Aumann (1976) gives that a ∈ I F . Upper semi continuity of B ( i, · ) implies that F ( i ) ⊆ B ( i, P ′ A ) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  58. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States � I F t for all t ≥ 0. Take a ∈ P ′ A . That is, a ∈ � I s t ∀ t ≥ 0. We get a sequence { s t } t ∈ N , such that a = � Lemma proved in Aumann (1976) gives that a ∈ I F . Upper semi continuity of B ( i, · ) implies that F ( i ) ⊆ B ( i, P ′ A ) � � A ) di ≡ ˜ B ( i, P ′ Pr ( P ′ a ∈ F di ⊆ A ) . I I � Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  59. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Characterization of Point-Rationalizable States analogous to Proposition 4. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  60. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Characterization of Point-Rationalizable States analogous to Proposition 4. Keys: (i) identify the adequate convergence concept for the eductive process. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  61. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Characterization of Point-Rationalizable States analogous to Proposition 4. Keys: (i) identify the adequate convergence concept for the eductive process. (ii) measurability requirements. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  62. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Characterization of Point-Rationalizable States analogous to Proposition 4. Keys: (i) identify the adequate convergence concept for the eductive process. (ii) measurability requirements. The set of Point-Rationalizable States is obtained as the integral of the point-wise upper limit of a sequence of set valued mappings. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  63. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Point-Rationalizable States Characterization of Point-Rationalizable States analogous to Proposition 4. Keys: (i) identify the adequate convergence concept for the eductive process. (ii) measurability requirements. The set of Point-Rationalizable States is obtained as the integral of the point-wise upper limit of a sequence of set valued mappings. Corollary 8 The set of Point-Rationalizable States of a game u is well defined, non-empty, compact and convex. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  64. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  65. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States When we consider standard Rationalizability, forecasts are subjective probability distributions over the sets of outcomes. In finite player games, we consider, for each player, product measures over the set of strategies of the opponents. In continuous player games, not trivial. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  66. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States When we consider standard Rationalizability, forecasts are subjective probability distributions over the sets of outcomes. In finite player games, we consider, for each player, product measures over the set of strategies of the opponents. In continuous player games, not trivial. In Rath’s setting, forecasts can be assumed to be (subjective) probability distributions over the set of states. B ( i, · ) : P ( A ) ⇒ S : B ( i, µ ) : = argmax y ∈ S E µ [ u ( i, y, a )] Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  67. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States The process of elimination of non-best-replies to (general) forecasts is described with the mapping ˜ R : B ( A ) → P ( A ): �� � s ∈ S I , s is a measurable selection ˜ R ( X ) := s ( i ) di : . of i ⇒ B ( i, P ( X )) I � ≡ B ( i, P ( X )) di I Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  68. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States The process of elimination of non-best-replies to (general) forecasts is described with the mapping ˜ R : B ( A ) → P ( A ): �� � s ∈ S I , s is a measurable selection ˜ R ( X ) := s ( i ) di : . of i ⇒ B ( i, P ( X )) I � ≡ B ( i, P ( X )) di I Proposition 9 In a game u , if X ⊆ A is nonempty and closed, then ˜ R ( X ) is nonempty, convex and closed. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  69. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States The Eductive Procedure: on each iteration, the states that are not reached by the process ˜ R are eliminated: � � ˜ R t +1 ( A ) := ˜ ˜ ˜ R 0 ( A ) := A , R t ( A ) R . ∞ � ˜ R ′ R t ( A ) . A := t =0 Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  70. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States The Eductive Procedure: on each iteration, the states that are not reached by the process ˜ R are eliminated: � � ˜ R t +1 ( A ) := ˜ ˜ ˜ R 0 ( A ) := A , R t ( A ) R . ∞ � ˜ R ′ R t ( A ) . A := t =0 Theorem 10 In a game u , the set R ′ A is non empty, convex and closed. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  71. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States Definition 11 The set of Rationalizable States is the maximal subset X ⊆ A that satisfies: ˜ R ( X ) ≡ X and we note it R A . Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  72. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States Definition 11 The set of Rationalizable States is the maximal subset X ⊆ A that satisfies: ˜ R ( X ) ≡ X and we note it R A . Theorem 12 The set of Rationalizable States of a game u satisfies R A ≡ R ′ A Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  73. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States Definition 11 The set of Rationalizable States is the maximal subset X ⊆ A that satisfies: ˜ R ( X ) ≡ X and we note it R A . Theorem 12 The set of Rationalizable States of a game u satisfies R A ≡ R ′ A The proof mimics that of Theorem 6, taking into account that if X is compact, then when P ( X ) is endowed with the weak* topology, we preserve continuity properties of payoffs and P ( X ) is compact and metrizable, (since we use the norm in R n ). Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  74. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizable States Proposition 13 If in a game u , we have ∀ µ ∈ P ( A ) : E µ [ u ( i, y, a )] ≡ u ( i, y, E µ [ a ]) then P A ≡ R A Proposition 13 says that if the utility functions are affine in the state variable, then we have that the Point-Rationalizable States set is equal to the set of Rationalizable States. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  75. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  76. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) Iterative elimination of unreasonable prices q ( i )( µ ) ≡ Supply( i )( E µ [ p ]) Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  77. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) Iterative elimination of unreasonable prices q ( i )( µ ) ≡ Supply( i )( E µ [ p ]) � Supply( i )( p ′ ) q ( i ) ∈ p ′ ∈ [0 ,p max ] Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  78. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) Iterative elimination of unreasonable prices q ( i )( µ ) ≡ Supply( i )( E µ [ p ]) � Supply( i )( p ′ ) q ( i ) ∈ ≡ Supply( i )([0 , p max ]) p ′ ∈ [0 ,p max ] Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  79. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) Iterative elimination of unreasonable prices q ( i )( µ ) ≡ Supply( i )( E µ [ p ]) � Supply( i )( p ′ ) q ( i ) ∈ ≡ Supply( i )([0 , p max ]) p ′ ∈ [0 ,p max ] �� � p ∈ P Supply( i )([0 , p max ]) di I Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  80. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Rationalizability in Guesnerie (1992) q ✻ ❅ P − 1 ( p ) ✠ ❅ S ( p ) ❅ ❄ ✦ ❅ ✦✦✦✦✦✦✦✦✦✦✦✦ ❅ ❅ q ∗ ❅ ❅ ❅ ❅ ✛ ❅ ✲ p 1 p ∗ p max p ✻ ❄ min p 2 max Figure: The eductive process Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  81. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary Outline 1 Motivation 2 Games with a continuum of players (Rath, 1992) Framework Guesnerie (1992) 3 Rationalizable Strategies in games with a finite number of players 4 State Rationalizability Point-Rationalizable States Rationalizable States Rationalizability in Guesnerie (1992) 5 Other Results 6 Summary Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

  82. Motivation The setting in Rath (1992) Rationalizable Strategies State Rationalizability Other Results Summary For more results see: Pedro Jara-Moroni. Rationalizability in games with a continuum of players. Paris School of Economics WP, 2007. Roger Guesnerie and Pedro Jara-Moroni. Expectational coordination in a class of economic models : strategic substitutabilities versus strategic complementarities. Paris School of Economics WP, 2007. Rationalizability in Games with a Continuum of Players Pedro Jara-Moroni, Postdoc at SCI-DIM-U. de Chile

Recommend


More recommend