博弈論完整課件[浙江大學(xué)]game_cha(1)-展示頁(yè)
2025-05-25 18:09本頁(yè)面
【正文】 e thought of as simultaneous. Most economically interesting Bayesian games, however, are dynamic. As we will see in Chapter 5, the existence of private information leads naturally to attempts by informed parties to municate(or mislead) and to attempts by unin formed parties to learn and respond. This are inherently dynamic issues. 一、 Theory: Static Bayesian Games and Bayesian Nash Equilibrium （一） An Example: Cournot Competition under Asymmetric Information … C1(q1)=Cq1 , mon knowledge,but C2(q2)=? , firm 2’s private information Firm 2 knows its cost function and firm 1’s, but firm 1 only knows its cost function(does not know firm 2’s cost function). All of this is mon knowledge. Firm 1 knows: firm 2’s cost function is C2(q2)= CHq2 with probability θ and C2(q2)= CLq2 with probability 1 θ, where CL ＜ CH. mon knowledge q2*(CH) will solve [aq1*q2cH] q2 q2 q2*(CH)= a q1* cH 2 max (1)式： q2*(CL) Will solve [aq1*q2cL] q2 max q2 q2*(CL)= a q1* cL 2 q1* Will solve [aq1q2*(CL)c] q1 [aq1q2*(CH)c] q1 (1θ) θ + { } max q1 （ 2）式： q1* = (1 θ)[aq2*(CL)c]+ θ [aq2*(CH)c] 2 （ 3）式 The solutions to ?1 ,?2 and ?3 are q2*(CH)=(a2CH+C)/3 + (1 θ)(CH – Cl )/6 q2*(CL)= (a2CL+C)/3 – θ(CH – Cl )/6 q1* =[a2C+ θCH +(1 θ)CL ]/3 不完全信息下古諾競(jìng)爭(zhēng)的貝葉斯 NE ?與完全信息下古諾競(jìng)爭(zhēng)的 NE作比較： 令 a=8， C=2， CH=3， CL=2 不完全信息 (θ=50% )： q2*(CH)= , q2*(CL)= , q1*= 完全信息 CH=3 也是共同信息（即 θ=1 ）： q2**(CH)=(a2CH+C)/3 = ＜ q2*(CH)= ＞ q1*= q1** =（ a2C+ CH ） /3 = 完全信息 CL=2也是共同信息（即 θ=0 ）： q2**(CL)=(a2CL+C)/3 =2 ＞ q2*(CL)= q1** =（ a2C+ CL） /3=2 ＜ q1*= ?firm 1的決策要把 CH和 CL都 考慮在內(nèi)， firm 2 知道自己的邊際成本是 CH （或 CL），為什么也 要把 CL（ CH）考慮在內(nèi)？（ 6） （二）、 NormalForm Representation of Static Bayesian Games The normalform representation of an nplayer game of plete information is G={S1, …,S n。μ1, …, μn} We now want to develop the normalform representation of a simultaneousmove game of inplete information, also called a s
黨政相關(guān)相關(guān)推薦
鄂ICP備17016276號(hào)-1